ProASIC3 nano FPGA Fabric
User’s Guide
ProASIC3 nano FPGA Fabric User’s Guide
Table of Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Revision History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Related Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1 FPGA Array Architecture in Low Power Flash Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Device Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
FPGA Array Architecture Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Device Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2 Low Power Modes in ProASIC3/E and ProASIC3 nano FPGAs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Power Consumption Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Static (Idle) Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
User Low Static (Idle) Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Sleep Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Shutdown Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21
21
22
23
25
27
28
28
29
3 Global Resources in Low Power Flash Devices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Global Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Global Resource Support in Flash-Based Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
VersaNet Global Network Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chip and Quadrant Global I/Os . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Spine Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using Clock Aggregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Design Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31
31
32
33
35
41
44
46
58
58
59
4 Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs . . . . . . . . . . . . . . 61
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Overview of Clock Conditioning Circuitry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CCC Support in Microsemi’s Flash Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Global Buffers with No Programmable Delays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Global Buffer with Programmable Delay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Global Buffers with PLL Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Global Input Selections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Device-Specific Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Revision 5
61
61
63
64
64
67
71
78
2
ProASIC3 nano FPGA Fabric User’s Guide
PLL Core Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Software Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
Detailed Usage Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Recommended Board-Level Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5 FlashROM in Microsemi’s Low Power Flash Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Architecture of User Nonvolatile FlashROM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
FlashROM Support in Flash-Based Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
FlashROM Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
FlashROM Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Programming and Accessing FlashROM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
FlashROM Design Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Custom Serialization Using FlashROM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
6 SRAM and FIFO Memories in Microsemi's Low Power Flash Devices . . . . . . . . . . . . . . . . . . . . . . . 131
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Device Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
SRAM/FIFO Support in Flash-Based Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
SRAM and FIFO Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Memory Blocks and Macros . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Initializing the RAM/FIFO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
Software Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
7 I/O Structures in nano Devices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Low Power Flash Device I/O Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
nano Standard I/Os . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
I/O Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
I/O Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Wide Range I/O Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
I/O Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
Simultaneously Switching Outputs (SSOs) and Printed Circuit Board Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
I/O Software Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
User I/O Naming Convention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
I/O Bank Architecture and CCC Naming Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Board-Level Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Revision 5
3
Table of Contents
8 I/O Software Control in Low Power Flash Devices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Flash FPGAs I/O Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Software-Controlled I/O Attributes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Implementing I/Os in Microsemi Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Assigning Technologies and VREF to I/O Banks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
186
187
188
198
203
203
204
9 DDR for Microsemi’s Low Power Flash Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Double Data Rate (DDR) Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
DDR Support in Flash-Based Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
I/O Cell Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Input Support for DDR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Output Support for DDR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Instantiating DDR Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Design Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
10 Programming Flash Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Summary of Programming Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Programming Support in Flash Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
General Flash Programming Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Important Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
11 Security in Low Power Flash Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Security in Programmable Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Security Support in Flash-Based Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Security Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Security Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Security in Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
FlashROM Security Use Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Generating Programming Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
235
236
237
238
242
245
247
258
258
258
259
259
12 In-System Programming (ISP) of Microsemi’s Low Power Flash Devices Using FlashPro4/3/3X . . . 261
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
ISP Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
ISP Support in Flash-Based Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
Programming Voltage (VPUMP) and VJTAG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
Nonvolatile Memory (NVM) Programming Voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
4
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
IEEE 1532 (JTAG) Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Security in ARM-Enabled Low Power Flash Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
FlashROM and Programming Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Programming Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ISP Programming Header Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Board-Level Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
264
264
265
267
268
269
271
272
272
273
13 Core Voltage Switching Circuit for IGLOO and ProASIC3L In-System Programming . . . . . . . . . . . . 275
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
Microsemi’s Flash Families Support Voltage Switching Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
Circuit Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Circuit Verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
DirectC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
14 Microprocessor Programming of Microsemi’s Low Power Flash Devices . . . . . . . . . . . . . . . . . . . . . 283
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Microprocessor Programming Support in Flash Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
Programming Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
Implementation Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
Hardware Requirement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
15 Boundary Scan in Low Power Flash Devices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
Boundary Scan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
TAP Controller State Machine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Microsemi’s Flash Devices Support the JTAG Feature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Boundary Scan Support in Low Power Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Boundary Scan Opcodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Boundary Scan Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Board-Level Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Advanced Boundary Scan Register Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
291
291
292
293
293
293
294
295
296
16 UJTAG Applications in Microsemi’s Low Power Flash Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
UJTAG Support in Flash-Based Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
UJTAG Macro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
UJTAG Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
Typical UJTAG Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
Revision 5
5
Table of Contents
17 Power-Up/-Down Behavior of Low Power Flash Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
Flash Devices Support Power-Up Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
Power-Up/-Down Sequence and Transient Current . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
I/O Behavior at Power-Up/-Down . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
Cold-Sparing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
Hot-Swapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
Related Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
List of Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
A Summary of Changes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
History of Revision to Chapters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
B Product Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
Customer Service . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Customer Technical Support Center . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Technical Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Website . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Contacting the Customer Technical Support Center . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ITAR Technical Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
321
321
321
321
321
322
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
6
R e vi s i o n 5
Introduction
Contents
This user’s guide contains information to help designers understand and use Microsemi's ProASIC®3
nano devices. Each chapter addresses a specific topic. Most of these chapters apply to other Microsemi
device families as well. When a feature or description applies only to a specific device family, this is made
clear in the text.
Revision History
The revision history for each chapter is listed at the end of the chapter. Most of these chapters were
formerly included in device handbooks. Some were originally application notes or information included in
device datasheets.
A "Summary of Changes" table at the end of this user’s guide lists the chapters that were changed in
each revision of the document, with links to the "List of Changes" sections for those chapters.
Related Information
Refer to the ProASIC3 nano Low Power Flash FPGAs datasheet for detailed specifications, timing, and
package and pin information.
The website for ProASIC3 nano devices is /www.microsemi.com/soc/products/pa3nano/default.aspx.
Revision 5
7
1 – FPGA Array Architecture in Low Power Flash
Devices
Device Architecture
Advanced Flash Switch
Unlike SRAM FPGAs, the low power flash devices use a live-at-power-up ISP flash switch as their
programming element. Flash cells are distributed throughout the device to provide nonvolatile,
reconfigurable programming to connect signal lines to the appropriate VersaTile inputs and outputs. In
the flash switch, two transistors share the floating gate, which stores the programming information
(Figure 1-1). One is the sensing transistor, which is only used for writing and verification of the floating
gate voltage. The other is the switching transistor. The latter is used to connect or separate routing nets,
or to configure VersaTile logic. It is also used to erase the floating gate. Dedicated high-performance
lines are connected as required using the flash switch for fast, low-skew, global signal distribution
throughout the device core. Maximum core utilization is possible for virtually any design. The use of the
flash switch technology also removes the possibility of firm errors, which are increasingly common in
SRAM-based FPGAs.
Floating Gate
Sensing
Switch In
Switching
Word
Switch Out
Figure 1-1 • Flash-Based Switch
Revision 5
9
FPGA Array Architecture in Low Power Flash Devices
FPGA Array Architecture Support
The flash FPGAs listed in Table 1-1 support the architecture features described in this document.
Table 1-1 • Flash-Based FPGAs
Family*
Series
IGLOO®
®
ProASIC 3
Fusion
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO nano
The industry’s lowest-power, smallest-size solution
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
ProASIC3
Low power, high-performance 1.5 V FPGAs
ProASIC3E
Higher density ProASIC3 FPGAs with six PLLs and additional I/O standards
ProASIC3 nano
Lowest-cost solution with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V with Flash*Freeze technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Automotive ProASIC3
ProASIC3 FPGAs qualified for automotive applications
Fusion
Mixed signal FPGA integrating ProASIC3 FPGA fabric, programmable
analog block, support for ARM® Cortex™-M1 soft processors, and flash
memory into a monolithic device
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO devices as listed
in Table 1-1. Where the information applies to only one product line or limited devices, these exclusions
will be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 1-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
10
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Device Overview
Low power flash devices consist of multiple distinct programmable architectural features (Figure 1-5 on
page 13 through Figure 1-7 on page 14):
•
FPGA fabric/core (VersaTiles)
•
Routing and clock resources (VersaNets)
•
FlashROM
•
Dedicated SRAM and/or FIFO
–
30 k gate and smaller device densities do not support SRAM or FIFO.
–
Automotive devices do not support FIFO operation.
•
I/O structures
•
Flash*Freeze technology and low power modes
Bank 1*
I/Os
Bank 1
Bank 0
VersaTile
User Nonvolatile
FlashROM
Flash*Freeze†
Technology
Charge
Pumps
CCC-GL
Bank 1
Notes: * Bank 0 for the 30 k devices
† Flash*Freeze mode is supported on IGLOO devices.
Figure 1-2 • IGLOO and ProASIC3 nano Device Architecture Overview with Two I/O Banks (applies to 10 k and
30 k device densities, excluding IGLOO PLUS devices)
Revision 5
11
FPGA Array Architecture in Low Power Flash Devices
Bank 0
Bank 0
Bank 1
CCC
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
I/Os
ISP AES
Decryption
User Nonvolatile
FlashRom
Flash*Freeze†
Technology
Charge
Pumps
Bank 0
Bank 1
VersaTile
Bank 1
Note: † Flash*Freeze mode is supported on IGLOO devices.
Figure 1-3 • IGLOO Device Architecture Overview with Two I/O Banks with RAM and PLL
(60 k and 125 k gate densities)
Bank 1
I/Os
Bank 1
Bank 0
VersaTile
User Nonvolatile
FlashROM
Flash*Freeze†
Technology
Charge Pumps
CCC-GL
Bank 1
Note: † Flash*Freeze mode is supported on IGLOO devices.
Figure 1-4 • IGLOO Device Architecture Overview with Three I/O Banks
(AGLN015, AGLN020, A3PN015, and A3PN020)
12
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Bank 0
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
Bank 1
Bank 3
CCC
I/Os
Bank 1
Bank 3
VersaTile
ISP AES
Decryption*
User Nonvolatile
FlashRom
Flash*Freeze
Technology
†
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
Charge
Pumps
Bank 2
Note: Flash*Freeze technology only applies to IGLOO and ProASIC3L families.
Figure 1-5 • IGLOO, IGLOO nano, ProASIC3 nano, and ProASIC3/L Device Architecture Overview with Four
I/O Banks (AGL600 device is shown)
Bank 0
Bank 1
Bank 3
CCC*
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
I/Os
ISP AES
Decryption*
User Nonvolatile
FlashRom
Flash*Freeze
Technology
Charge
Pumps
Bank 1
Bank 3
VersaTile
Bank 2
Note: * AGLP030 does not contain a PLL or support AES security.
Figure 1-6 • IGLOO PLUS Device Architecture Overview with Four I/O Banks
Revision 5
13
FPGA Array Architecture in Low Power Flash Devices
Bank 0
Bank 1
CCC
Bank 2
Bank 7
RAM Block
4,608-Bit Dual-Port SRAM
or FIFO Block
Pro I/Os
Bank 3
Bank 6
VersaTile
ISP AES
Decryption
User Nonvolatile
FlashRom
Bank 5
Flash*Freeze†
Technology
Charge
Pumps
RAM Block
4,608-Bit Dual-Port SRAM
or FIFO Block
Bank 4
Note: Flash*Freeze technology only applies to IGLOOe devices.
Figure 1-7 • IGLOOe and ProASIC3E Device Architecture Overview (AGLE600 device is shown)
I/O State of Newly Shipped Devices
Devices are shipped from the factory with a test design in the device. The power-on switch for VCC is
OFF by default in this test design, so I/Os are tristated by default. Tristated means the I/O is not actively
driven and floats. The exact value cannot be guaranteed when it is floating. Even in simulation software,
a tristate value is marked as unknown. Due to process variations and shifts, tristated I/Os may float
toward High or Low, depending on the particular device and leakage level.
If there is concern regarding the exact state of unused I/Os, weak pull-up/pull-down should be added to
the floating I/Os so their state is controlled and stabilized.
14
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Core Architecture
VersaTile
The proprietary IGLOO and ProASIC3 device architectures provide granularity comparable to gate
arrays. The device core consists of a sea-of-VersaTiles architecture.
As illustrated in Figure 1-8, there are four inputs in a logic VersaTile cell, and each VersaTile can be
configured using the appropriate flash switch connections:
•
Any 3-input logic function
•
Latch with clear or set
•
D-flip-flop with clear or set
•
Enable D-flip-flop with clear or set (on a 4th input)
VersaTiles can flexibly map the logic and sequential gates of a design. The inputs of the VersaTile can be
inverted (allowing bubble pushing), and the output of the tile can connect to high-speed, very-long-line
routing resources. VersaTiles and larger functions can be connected with any of the four levels of routing
hierarchy.
When the VersaTile is used as an enable D-flip-flop, SET/CLR is supported by a fourth input. The
SET/CLR signal can only be routed to this fourth input over the VersaNet (global) network. However, if, in
the user’s design, the SET/CLR signal is not routed over the VersaNet network, a compile warning
message will be given, and the intended logic function will be implemented by two VersaTiles instead of
one.
The output of the VersaTile is F2 when the connection is to the ultra-fast local lines, or YL when the
connection is to the efficient long-line or very-long-line resources.
0
1
Data
X3
Y
Pin 1
0
1
0
1
F2
YL
CLK
X2
0
1
CLR/
Enable
X1
CLR
XC*
Legend:
Via (hard connection)
Switch (flash connection)
Ground
* This input can only be connected to the global clock distribution network.
Figure 1-8 • Low Power Flash Device Core VersaTile
Revision 5
15
FPGA Array Architecture in Low Power Flash Devices
Array Coordinates
During many place-and-route operations in the Microsemi Designer software tool, it is possible to set
constraints that require array coordinates. Table 1-2 provides array coordinates of core cells and memory
blocks for IGLOO and ProASIC3 devices. Table 1-3 provides the information for IGLOO PLUS devices.
Table 1-4 on page 17 provides the information for IGLOO nano and ProASIC3 nano devices. The array
coordinates are measured from the lower left (0, 0). They can be used in region constraints for specific
logic groups/blocks, designated by a wildcard, and can contain core cells, memories, and I/Os.
I/O and cell coordinates are used for placement constraints. Two coordinate systems are needed
because there is not a one-to-one correspondence between I/O cells and core cells. In addition, the I/O
coordinate system changes depending on the die/package combination. It is not listed in Table 1-2. The
Designer ChipPlanner tool provides the array coordinates of all I/O locations. I/O and cell coordinates are
used for placement constraints. However, I/O placement is easier by package pin assignment.
Figure 1-9 on page 17 illustrates the array coordinates of a 600 k gate device. For more information on
how to use array coordinates for region/placement constraints, see the Designer User's Guide or online
help (available in the software) for software tools.
Table 1-2 • IGLOO and ProASIC3 Array Coordinates
VersaTiles
Device
Min.
Memory Rows
Max.
Entire Die
Bottom
Top
Min.
Max.
IGLOO
ProASIC3/
ProASIC3L
x
y
x
y
(x, y)
(x, y)
(x, y)
(x, y)
AGL015
A3P015
3
2
34
13
None
None
(0, 0)
(37, 15)
AGL030
A3P030
3
3
66
13
None
None
(0, 0)
(69, 15)
AGL060
A3P060
3
2
66
25
None
(3, 26)
(0, 0)
(69, 29)
AGL125
A3P125
3
2
130
25
None
(3, 26)
(0, 0)
(133, 29)
AGL250
A3P250/L
3
2
130
49
None
(3, 50)
(0, 0)
(133, 53)
AGL400
A3P400
3
2
194
49
None
(3, 50)
(0, 0)
(197, 53)
AGL600
A3P600/L
3
4
194
75
(3, 2)
(3, 76)
(0, 0)
(197, 79)
AGL1000
A3P1000/L
3
4
258
99
(3, 2)
(3, 100)
(0, 0)
(261, 103)
AGLE600
A3PE600/L,
RT3PE600L
3
4
194
75
(3, 2)
(3, 76)
(0, 0)
(197, 79)
A3PE1500
3
4
322
123
(3, 2)
(3, 124)
(0, 0)
(325, 127)
A3PE3000/L,
RT3PE3000L
3
6
450
173
(3, 2)
or
(3, 4)
(3, 174)
or
(3, 176)
(0, 0)
(453, 179)
AGLE3000
Table 1-3 • IGLOO PLUS Array Coordinates
VersaTiles
Device
16
Min.
Memory Rows
Max.
Entire Die
Bottom
Top
Min.
Max.
IGLOO PLUS
x
y
x
y
(x, y)
(x, y)
(x, y)
(x, y)
AGLP030
2
3
67
13
None
None
(0, 0)
(69, 15)
AGLP060
2
2
67
25
None
(3, 26)
(0, 0)
(69, 29)
AGLP125
2
2
131
25
None
(3, 26)
(0, 0)
(133, 29)
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Table 1-4 • IGLOO nano and ProASIC3 nano Array Coordinates
VersaTiles
Device
Memory Rows
Entire Die
Min.
Max.
Bottom
Top
Min.
Max.
IGLOO nano
ProASIC3 nano
(x, y)
(x, y)
(x, y)
(x, y)
(x, y)
(x, y)
AGLN010
A3P010
(0, 2)
(32, 5)
None
None
(0, 0)
(34, 5)
AGLN015
A3PN015
(0, 2)
(32, 9)
None
None
(0, 0)
(34, 9)
AGLN020
A3PN020
(0, 2)
32, 13)
None
None
(0, 0)
(34, 13)
AGLN060
A3PN060
(3, 2)
(66, 25)
None
(3, 26)
(0, 0)
(69, 29)
AGLN125
A3PN125
(3, 2)
(130, 25)
None
(3, 26)
(0, 0)
(133, 29)
AGLN250
A3PN250
(3, 2)
(130, 49)
None
(3, 50)
(0, 0)
(133, 49)
Top Row (7, 79) to (189, 79)
Bottom Row (5, 78) to (192, 78)
(0, 79)
I/O Tile
(197, 79)
Memory (3, 77)
Blocks (3, 76)
(194, 77) Memory
(194, 76) Blocks
VersaTile (Core)
(3, 75)
(194, 75)
VersaTile (Core)
(194, 4)
VersaTile (Core)
VersaTile (Core)
(3, 4)
(194, 3) Memory
(194, 2) Blocks
Memory (3, 3)
Blocks (3, 2)
(197, 1)
(0, 0)
I/O Tile
UJTAG FlashROM
Top Row (5, 1) to (168, 1)
Bottom Row (7, 0) to (165, 0)
(197, 0)
Top Row (169, 1) to (192, 1)
Note: The vertical I/O tile coordinates are not shown. West-side coordinates are {(0, 2) to (2, 2)} to {(0, 77) to (2, 77)};
east-side coordinates are {(195, 2) to (197, 2)} to {(195, 77) to (197, 77)}.
Figure 1-9 • Array Coordinates for AGL600, AGLE600, A3P600, and A3PE600
Revision 5
17
FPGA Array Architecture in Low Power Flash Devices
Routing Architecture
The routing structure of low power flash devices is designed to provide high performance through a
flexible four-level hierarchy of routing resources: ultra-fast local resources; efficient long-line resources;
high-speed, very-long-line resources; and the high-performance VersaNet networks.
The ultra-fast local resources are dedicated lines that allow the output of each VersaTile to connect
directly to every input of the eight surrounding VersaTiles (Figure 1-10). The exception to this is that the
SET/CLR input of a VersaTile configured as a D-flip-flop is driven only by the VersaTile global network.
The efficient long-line resources provide routing for longer distances and higher-fanout connections.
These resources vary in length (spanning one, two, or four VersaTiles), run both vertically and
horizontally, and cover the entire device (Figure 1-11 on page 19). Each VersaTile can drive signals onto
the efficient long-line resources, which can access every input of every VersaTile. Routing software
automatically inserts active buffers to limit loading effects.
The high-speed, very-long-line resources, which span the entire device with minimal delay, are used to
route very long or high-fanout nets: length ±12 VersaTiles in the vertical direction and length ±16 in the
horizontal direction from a given core VersaTile (Figure 1-12 on page 19). Very long lines in low power
flash devices have been enhanced over those in previous ProASIC families. This provides a significant
performance boost for long-reach signals.
The high-performance VersaNet global networks are low-skew, high-fanout nets that are accessible from
external pins or internal logic. These nets are typically used to distribute clocks, resets, and other highfanout nets requiring minimum skew. The VersaNet networks are implemented as clock trees, and
signals can be introduced at any junction. These can be employed hierarchically, with signals accessing
every input of every VersaTile. For more details on VersaNets, refer to the "Global Resources in Low
Power Flash Devices" section on page 31.
Long Lines
L
Inputs
L
L
L
Output
L
L
L
L
Ultra-Fast Local Lines
(connects a VersaTile to the
adjacent VersaTile, I/O buffer,
or memory block)
L
Note: Input to the core cell for the D-flip-flop set and reset is only available via the VersaNet global
network connection.
Figure 1-10 • Ultra-Fast Local Lines Connected to the Eight Nearest Neighbors
18
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Spans 4 VersaTiles
Spans 2 VersaTiles
Spans 1 VersaTile
VersaTile
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
Spans 1 VersaTile
Spans 2 VersaTiles
Spans 4 VersaTiles
Figure 1-11 • Efficient Long-Line Resources
High-Speed, Very-Long-Line Resources
Pad Ring
SRAM
I/O Ring
Pad Ring
I/O Ring
16×12 Block of VersaTiles
Pad Ring
Figure 1-12 • Very-Long-Line Resources
Revision 5
19
FPGA Array Architecture in Low Power Flash Devices
Related Documents
User’s Guides
Designer User's Guide
http://www.microsemi.com/soc/documents/designer_ug.pdf
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
Changes
Page
August 2012
The "I/O State of Newly Shipped Devices" section is new (SAR 39542).
14
July 2010
This chapter is no longer published separately with its own part number and version
but is now part of several FPGA fabric user’s guides.
N/A
v1.4
(December 2008)
IGLOO nano and ProASIC3 nano devices were added to Table 1-1 • Flash-Based
FPGAs.
10
Figure 1-2 • IGLOO and ProASIC3 nano Device Architecture Overview with Two I/O
Banks (applies to 10 k and 30 k device densities, excluding IGLOO PLUS devices)
through Figure 1-5 • IGLOO, IGLOO nano, ProASIC3 nano, and ProASIC3/L Device
Architecture Overview with Four I/O Banks (AGL600 device is shown) are new.
11, 12
v1.3
(October 2008)
v1.2
(June 2008)
v1.1
(March 2008)
20
Table 1-4 • IGLOO nano and ProASIC3 nano Array Coordinates is new.
17
The title of this document was changed from "Core Architecture of IGLOO and
ProASIC3 Devices" to "FPGA Array Architecture in Low Power Flash Devices."
9
The "FPGA Array Architecture Support" section was revised to include new families
and make the information more concise.
10
Table 1-2 • IGLOO and ProASIC3 Array Coordinates was updated to include Military
ProASIC3/EL and RT ProASIC3 devices.
16
The following changes were made to the family descriptions in Table 1-1 • FlashBased FPGAs:
10
•
ProASIC3L was updated to include 1.5 V.
•
The number of PLLs for ProASIC3E was changed from five to six.
Table 1-1 • Flash-Based FPGAs and the accompanying text was updated to include
the IGLOO PLUS family. The "IGLOO Terminology" section and "Device Overview"
section are new.
10
The "Device Overview" section was updated to note that 15 k devices do not
support SRAM or FIFO.
11
Figure 1-6 • IGLOO PLUS Device Architecture Overview with Four I/O Banks is
new.
13
Table 1-2 • IGLOO and ProASIC3 Array Coordinates was updated to add A3P015
and AGL015.
16
Table 1-3 • IGLOO PLUS Array Coordinates is new.
16
R e visio n 5
2 – Low Power Modes in ProASIC3/E and
ProASIC3 nano FPGAs
Introduction
The demand for low power systems and semiconductors, combined with the strong growth observed for
value-based FPGAs, is driving growing demand for low power FPGAs. For portable and battery-operated
applications, power consumption has always been the greatest challenge. The battery life of a system
and on-board devices has a direct impact on the success of the product. As a result, FPGAs used in
these applications should meet low power consumption requirements.
ProASIC®3/E and ProASIC3 nano FPGAs offer low power consumption capability inherited from their
nonvolatile and live-at-power-up (LAPU) flash technology. This application note describes the power
consumption and how to use different power saving modes to further reduce power consumption for
power-conscious electronics design.
Power Consumption Overview
In evaluating the power consumption of FPGA technologies, it is important to consider it from a system
point of view. Generally, the overall power consumption should be based on static, dynamic, inrush, and
configuration power. Few FPGAs implement ways to reduce static power consumption utilizing sleep
modes.
SRAM-based FPGAs use volatile memory for their configuration, so the device must be reconfigured
after each power-up cycle. Moreover, during this initialization state, the logic could be in an indeterminate
state, which might cause inrush current and power spikes. More complex power supplies are required to
eliminate potential system power-up failures, resulting in higher costs. For portable electronics requiring
frequent power-up and -down cycles, this directly affects battery life, requiring more frequent recharging
or replacement.
SRAM-Based FPGA Total Power Consumption = Pstatic + Pdynamic + Pinrush + Pconfig
EQ 1
ProASIC3/E Total Power Consumption = Pstatic + Pdynamic
EQ 2
Unlike SRAM-based FPGAs, Microsemi flash-based FPGAs are nonvolatile and do not require power-up
configuration. Additionally, Microsemi nonvolatile flash FPGAs are live at power-up and do not require
additional support components. Total power consumption is reduced as the inrush current and
configuration power components are eliminated.
Note that the static power component can be reduced in flash FPGAs (such as the ProASIC3/E devices)
by entering User Low Static mode or Sleep mode. This leads to an extremely low static power
component contribution to the total system power consumption.
The following sections describe the usage of Static (Idle) mode to reduce the power component, User
Low Static mode to reduce the static power component, and Sleep mode and Shutdown mode to achieve
a range of power consumption when the FPGA or system is idle. Table 2-1 on page 22 summarizes the
different low power modes offered by ProASIC3/E devices.
Revision 5
21
Low Power Modes in ProASIC3/E and ProASIC3 nano FPGAs
Table 2-1 • ProASIC3/E/nano Low Power Modes Summary
Mode
Active
Power Supplies / Clock Status
On – All, clock
Needed to Start Up
N/A (already active)
Off – None
Static (Idle) On – All
Initiate clock source.
Off – No active clock in FPGA
No need to initialize volatile
contents.
Optional: Enter User Low Static (Idle) Mode by enabling
ULSICC macro to further reduce power consumption by
powering down FlashROM.
Sleep
On – VCCI
Need to turn on core.
Off – VCC (core voltage), VJTAG (JTAG DC voltage), Load states
and VPUMP (programming voltage)
memory.
from
external
LAPU enables immediate operation when power As needed, restore volatile
returns.
contents from external memory.
Optional: Save state of volatile contents in external
memory.
Shutdown
On – None
Need to turn on VCC, VCCI.
Off – All power supplies
Applicable to all ProASIC3 nano devices, cold-sparing
and hot-insertion allow the device to be powered down
without bringing down the system. LAPU enables
immediate operation when power returns.
Static (Idle) Mode
In Static (Idle) mode, the clock inputs are not switching and the static power consumption is the minimum
power required to keep the device powered up. In this mode, I/Os are only drawing the minimum leakage
current specified in the datasheet. Also, in Static (Idle) mode, embedded SRAM, I/Os, and registers
retain their values, so the device can enter and exit this mode without any penalty.
If the embedded PLLs are used as the clock source, Static (Idle) mode can be entered easily by pulling
LOW the PLL POWERDOWN pin (active-low). By pulling the PLL POWERDOWN pin to LOW, the PLL is
turned off. Refer to Figure 2-1 on page 23 for more information.
22
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
CLKA
POWERDOWN
GLA
GLB
YB
GLC
YC
LOCK
OADIV[4:0]*
OAMUX[2:0]*
DLYGLA[4:0]*
OBDIV[4:0]*
OBMUX[2:0]*
DLYYB[4:0]*
DLYGLB[4:0]*
OCDIV[4:0]*
OCMUX[2:0]*
DLYYC[4:0]*
DLYGLC[4:0]*
FINDIV[6:0]*
FBDIV[6:0]*
FBDLY[4:0]*
FBSEL[1:0]*
XDLYSEL*
VCOSEL[2:0]*
Figure 2-1 • CCC/PLL Macro
User Low Static (Idle) Mode
User Low Static (Idle) mode is an advanced feature supported by ProASIC3/E devices to reduce static
(idle) power consumption. Entering and exiting this mode is made possible using the ULSICC macro by
setting its value to disable/enable the User Low Static (Idle) mode. Under typical operating conditions,
characterization results show up to 25% reduction of the static (idle) power consumption. The greatest
power savings in terms of percentage are seen in the smaller members of the ProASIC3 family. The
active-high control signal for User Low Static (Idle) mode can be generated by internal or external logic.
When the device is operating in User Low Static (Idle) mode, FlashROM functionality is temporarily
disabled to save power. If FlashROM functionality is needed, the device can exit User Low Static mode
temporarily and re-enter the mode once the functionality is no longer needed.
To utilize User Low Static (Idle) mode, simply instantiate the ULSICC macro (Table 2-2 on page 24) in
your design, and connect the input port to either an internal logic signal or a device package pin, as
illustrated in Figure 2-2 on page 24 or Figure 2-3 on page 25, respectively. The attribute is used so the
Synplify® synthesis tool will not optimize the instance with no output port.
This mode can be used to lower standard static (idle) power consumption when the FlashROM feature is
not needed. Configuring the device to enter User Low Static (Idle) mode is beneficial when the FPGA
enters and exits static mode frequently and lowering power consumption as much as possible is desired.
The device is still functional, and data is retained in this state so the device can enter and exit this mode
quickly, resulting in reduced total power consumption. The device can also stay in User Low Static mode
when the FlashROM feature is not used in the device.
Revision 5
23
Low Power Modes in ProASIC3/E and ProASIC3 nano FPGAs
Table 2-2 • Using ULSICC Macro*
VHDL
COMPONENT ULSICC
port (
LSICC
END COMPONENT;
Verilog
: in
module ULSICC(LSICC);
input LSICC;
STD_ULOGIC); endmodule
Example:
Example:
COMPONENT ULSICC
port (
LSICC
END COMPONENT;
ULSICC U1(.LSICC(myInputSignal))
/* synthesis syn_noprune=1 */;
: in
STD_ULOGIC);
attribute syn_noprune : boolean;
attribute syn_noprune of u1 : label is true;
u1: ULSICC port map(myInputSignal);
Note: *Supported in Libero® software v7.2 and newer versions.
ProASIC3/E Device
Internal
Signal
Programming
Circuitry
ULSICC
Macro
FlashROM
Figure 2-2 • User Low Static (Idle) Mode Application—Internal Control Signal
24
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
ProASIC3/E/nano Device
External
Signal
Programming
Circuitry
ULSICC
Macro
FlashROM
Any User's I/O
Figure 2-3 • User Low Static (Idle) Mode Application—External Control Signal
Normal Operation
User Low Static Mode
Normal Operation
ULSICC Signal
1 μs
1 μs
Figure 2-4 • User Low Static (Idle) Mode Timing Diagram
Sleep Mode
ProASIC3/E and ProASIC3 nano FPGAs support Sleep mode when device functionality is not required.
In Sleep mode, the VCC (core voltage), VJTAG (JTAG DC voltage), and VPUMP (programming voltage)
are grounded, resulting in the FPGA core being turned off to reduce power consumption. While the
ProASIC3/E device is in Sleep mode, the rest of the system is still operating and driving the input buffers
of the ProASIC3/E device. The driven inputs do not pull up power planes, and the current draw is limited
to a minimal leakage current.
Table 2-3 shows the status of the power supplies in Sleep mode. When a power supply is powered off,
the corresponding power pin can be left floating or grounded.
Table 2-3 • Sleep Mode—Power Supply Requirements for ProASIC3/E/nano Devices
Power Supplies
ProASIC3/E/nano Device
VCC
Powered off
VCCI = VMV
Powered on
VJTAG
Powered off
VPUMP
Powered off
Revision 5
25
Low Power Modes in ProASIC3/E and ProASIC3 nano FPGAs
Table 2-4 shows the current draw in Sleep mode for an A3P250 device with the following test conditions:
VCCI = VMV; VCC = VJTAG = VPUMP = GND.
Table 2-4 • A3P250 Current Draw in Sleep Mode
A3P250
ICCI (µA)
ICCI (µA) per Bank
VCCI = 3.3 V
31.57
7.89
VCCI = 2.5 V
23.96
5.99
VCCI = 1.8 V
17.32
4.33
VCCI = 1.5 V
14.46
3.62
ICC FPGA Core
0.0
0.0
Leakage Current per I/O
0.1
0.1
VPUMP
0.0
0.0
Typical Conditions
Note: The data in this table were taken under typical conditions and are based on characterization. The
data is not guaranteed.
Table 2-5 shows the current draw in Sleep mode for an A3PE600 device with the following test
conditions: VCCI = VMV; VCC = VJTAG = VPUMP = GND.
Table 2-5 • A3PE600 Current Draw in Sleep Mode
A3PE600
Typical Conditions
ICCI (µA)
ICCI (µA) per Bank
VCCI = 3.3 V
59.85
7.48
VCCI = 2.5 V
45.50
5.69
VCCI = 1.8 V
32.98
4.12
VCCI = 1.5 V
27.66
3.46
VCCI = 0 V or Floating
0.0
0.0
ICC FPGA Core
0.0
0.0
Leakage Current per I/O
0.1
0.1
IPUMP
0.0
0.0
Note: The data in this table were taken under typical conditions and are based on characterization. The
data is not guaranteed.
ProASIC3/E and ProASIC3 nano devices were designed such that before device power-up, all I/Os are
in tristate mode. The I/Os will remain tristated during power-up until the last voltage supply (VCC or
VCCI) is powered to its functional level. After the last supply reaches the functional level, the outputs will
exit the tristate mode and drive the logic at the input of the output buffer. The behavior of user I/Os is
independent of the VCC and VCCI sequence or the state of other FPGA voltage supplies (VPUMP and
VJTAG). During power-down, device I/Os become tristated once the first power supply (VCC or VCCI)
drops below its brownout voltage level. The I/O behavior during power-down is also independent of
voltage supply sequencing.
Figure 2-5 on page 27 shows a timing diagram for the FPGA core entering the activation and
deactivation trip points for a typical application when the VCC power supply ramp rate is 100 µs (ramping
from 0 V to 1.5 V). This is, in fact, the timing diagram for the FPGA entering and exiting Sleep mode, as it
is dependent on powering down or powering up VCC. Depending on the ramp rate of the power supply
and board-level configurations, the user can easily calculate how long it takes for the core to become
active or inactive. For more information, refer to the "Power-Up/-Down Behavior of Low Power Flash
Devices" section on page 307.
26
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
VCC
VCC = 1.5 V
Deactivation Trip Point
Vd = 0.75 ± 0.25 V
Activation Trip Point
Va = 0.85 ± 0.25 V
t
Sleep Mode
Figure 2-5 • Entering and Exiting Sleep Mode—Typical Timing Diagram
Shutdown Mode
For all ProASIC3/E and ProASIC3 nano devices, shutdown mode can be entered by turning off all power
supplies when device functionality is not needed. Cold-sparing and hot-insertion features in ProASIC3
nano devices enable the device to be powered down without turning off the entire system. When power
returns, the live at power-up feature enables immediate operation of the device.
Using Sleep Mode or Shutdown Mode in the System
Depending on the power supply and components used in an application, there are many ways to turn the
power supplies connected to the device on or off. For example, Figure 2-6 shows how a microprocessor
is used to control a power FET. It is recommended that power FETs with low on resistance be used to
perform the switching action.
1.5 V Power
Supply
P-Channel
Power FET
Microprocessor
Power On/Off
Control Signal
VCC, VJTAG, and VPUMP Pins
ProASIC3/E/nano
Figure 2-6 • Controlling Power On/Off State Using Microprocessor and Power FET
Revision 5
27
Low Power Modes in ProASIC3/E and ProASIC3 nano FPGAs
Alternatively, Figure 2-7 shows how a microprocessor can be used with a voltage regulator's shutdown
pin to turn the power supplies connected to the device on or off.
Microprocessor
Shutdown
Control Signal
for VCCI
Shutdown
Control Signal
for VCC, VJTAG, and VPUMP
VCCI Power Pin
Power
Supply
Voltage
Regulator
ProASIC3/E/nano
VCC, VJTAG, and VPUMP
Power Pins
Figure 2-7 • Controlling Power On/Off State Using Microprocessor and Voltage Regulator
Though Sleep mode or Shutdown mode can be used to save power, the content of the SRAM and the
state of the registers is lost when power is turned off if no other measure is taken. To keep the original
contents of the device, a low-cost external serial EEPROM can be used to save and restore the device
contents when entering and exiting Sleep mode. In the Embedded SRAM Initialization Using External
Serial EEPROM application note, detailed information and a reference design are provided to initialize
the embedded SRAM using an external serial EEPROM. The user can easily customize the reference
design to save and restore the FPGA state when entering and exiting Sleep mode. The microcontroller
will need to manage this activity, so before powering down VCC, the data must be read from the FPGA
and stored externally. Similarly, after the FPGA is powered up, the microcontroller must allow the FPGA
to load the data from external memory and restore its original state.
Conclusion
Microsemi ProASIC3/E and ProASIC3 nano FPGAs inherit low power consumption capability from their
nonvolatile and live-at-power-up flash-based technology. Power consumption can be reduced further
using the Static (Idle), User Low Static (Idle), Sleep, or Shutdown power modes. All these features result
in a low-power, cost-effective, single-chip solution designed specifically for power-sensitive electronics
applications.
Related Documents
Application Notes
Embedded SRAM Initialization Using External Serial EEPROM
http://www.microsemi.com/soc/documents/EmbeddedSRAMInit_AN.pdf
28
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
Changes
Page
June 2011
Table 2-1 • ProASIC3/E/nano Low Power Modes Summary and the "Shutdown
Mode" section were revised to remove reference to ProASIC3/E devices (SAR
24526).
22, 27
July 2010
This chapter is no longer published separately with its own part number and version
but is now part of several FPGA fabric user’s guides.
N/A
v1.2
(August 2008)
References to ProASIC3 nano devices were added to the document where
appropriate.
N/A
VJTAG and VPUMP were noted as "Off" in the Sleep Mode section of Table 2-1 •
ProASIC3/E/nano Low Power Modes Summary.
22
The "Sleep Mode" section, including Table 2-3 • Sleep Mode—Power Supply
Requirements for ProASIC3/E/nano Devices, was revised to state that VJTAG and
VPUMP are powered off during Sleep mode.
25
The text above Table 2-4 • A3P250 Current Draw in Sleep Mode and Table 2-5 •
A3PE600 Current Draw in Sleep Mode was revised to state "VCC = VJTAG =
VPUMP = GND."
26
Figure 2-6 • Controlling Power On/Off State Using Microprocessor and Power FET
and Figure 2-7 • Controlling Power On/Off State Using Microprocessor and Voltage
Regulator were revised to show shutdown of VJTAG and VPUMP during Sleep
mode.
27, 28
v1.1
(March 2008)
The part number for this document was changed from 51700094-002-0 to
51700094-003-1.
N/A
v1.0
(January 2008)
The Power Supplies / Clock Status description was updated for Static (Idle) in
Table 2-1 • ProASIC3/E/nano Low Power Modes Summary.
22
Programming information was updated in the "User Low Static (Idle) Mode"
section.
23
The "User Low Static (Idle) Mode" section was updated to include information
about allowing programming in the ULSICC mode.
23
Figure 2-2 • User Low Static (Idle) Mode Application—Internal Control Signal was
updated.
24
Figure 2-3 • User Low Static (Idle) Mode Application—External Control Signal was
updated.
25
In Table 2-4 • A3P250 Current Draw in Sleep Mode, "VCCI = 1.5 V" was changed
from 3.6158 to 3.62.
26
In Table 2-5 • A3PE600 Current Draw in Sleep Mode, "VCCI = 2.5 V" was changed
from 5.6875 to 3.69.
26
51900138-2/10.06
51900138-1/6.06
Revision 5
29
3 – Global Resources in Low Power Flash Devices
Introduction
IGLOO, Fusion, and ProASIC3 FPGA devices offer a powerful, low-delay VersaNet global network
scheme and have extensive support for multiple clock domains. In addition to the Clock Conditioning
Circuits (CCCs) and phase-locked loops (PLLs), there is a comprehensive global clock distribution
network called a VersaNet global network. Each logical element (VersaTile) input and output port has
access to these global networks. The VersaNet global networks can be used to distribute low-skew clock
signals or high-fanout nets. In addition, these highly segmented VersaNet global networks contain spines
(the vertical branches of the global network tree) and ribs that can reach all the VersaTiles inside their
region. This allows users the flexibility to create low-skew local clock networks using spines. This
document describes VersaNet global networks and discusses how to assign signals to these global
networks and spines in a design flow. Details concerning low power flash device PLLs are described in
the "Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs" section on
page 61. This chapter describes the low power flash devices’ global architecture and uses of these global
networks in designs.
Global Architecture
Low power flash devices offer powerful and flexible control of circuit timing through the use of global
circuitry. Each chip has up to six CCCs, some with PLLs.
•
In IGLOOe, ProASIC3EL, and ProASIC3E devices, all CCCs have PLLs—hence, 6 PLLs per
device (except the PQ208 package, which has only 2 PLLs).
•
In IGLOO, IGLOO nano, IGLOO PLUS, ProASIC3, and ProASIC3L devices, the west CCC
contains a PLL core (except in 10 k through 30 k devices).
•
In Fusion devices, the west CCC also contains a PLL core. In the two larger devices (AFS600 and
AFS1500), the west and east CCCs each contain a PLL.
Refer to Table 4-6 on page 84 for details. Each PLL includes delay lines, a phase shifter (0°, 90°, 180°,
270°), and clock multipliers/dividers. Each CCC has all the circuitry needed for the selection and
interconnection of inputs to the VersaNet global network. The east and west CCCs each have access to
three chip global lines on each side of the chip (six chip global lines total). The CCCs at the four corners
each have access to three quadrant global lines in each quadrant of the chip (except in 10 k through 30 k
gate devices).
The nano 10 k, 15 k, and 20 k devices support four VersaNet global resources, and 30 k devices support
six global resources. The 10 k through 30 k devices have simplified CCCs called CCC-GLs.
The flexible use of the VersaNet global network allows the designer to address several design
requirements. User applications that are clock-resource-intensive can easily route external or gated
internal clocks using VersaNet global routing networks. Designers can also drastically reduce delay
penalties and minimize resource usage by mapping critical, high-fanout nets to the VersaNet global
network.
Note: Microsemi recommends that you choose the appropriate global pin and use the appropriate global
resource so you can realize these benefits.
The following sections give an overview of the VersaNet global network, the structure of the global
network, access point for the global networks, and the clock aggregation feature that enables a design to
have very low clock skew using spines.
Revision 5
31
Global Resources in Low Power Flash Devices
Global Resource Support in Flash-Based Devices
The flash FPGAs listed in Table 3-1 support the global resources and the functions described in this
document.
Table 3-1 • Flash-Based FPGAs
Family*
Series
IGLOO
ProASIC3
Fusion
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
IGLOO nano
The industry’s lowest-power, smallest-size solution
ProASIC3
Low power, high-performance 1.5 V FPGAs
ProASIC3E
Higher density ProASIC3 FPGAs with six PLLs and additional I/O standards
ProASIC3 nano
Lowest-cost solution with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V with Flash*Freeze technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Automotive ProASIC3
ProASIC3 FPGAs qualified for automotive applications
Fusion
Mixed signal FPGA integrating ProASIC3 FPGA fabric, programmable
analog block, support for ARM® Cortex™-M1 soft processors, and flash
memory into a monolithic device
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO products as
listed in Table 3-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 3-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
32
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
VersaNet Global Network Distribution
One of the architectural benefits of low power flash architecture is the set of powerful, low-delay
VersaNet global networks that can access the VersaTiles, SRAM, and I/O tiles of the device. Each device
offers a chip global network with six global lines (except for nano 10 k, 15 k, and 20 k gate devices) that
are distributed from the center of the FPGA array. In addition, each device (except the 10 k through 30 k
gate device) has four quadrant global networks, each consisting of three quadrant global net resources.
These quadrant global networks can only drive a signal inside their own quadrant. Each VersaTile has
access to nine global line resources—three quadrant and six chip-wide (main) global networks—and a
total of 18 globals are available on the device (3 × 4 regional from each quadrant and 6 global).
Figure 3-1 shows an overview of the VersaNet global network and device architecture for devices 60 k
and above. Figure 3-2 and Figure 3-3 on page 34 show simplified VersaNet global networks.
The VersaNet global networks are segmented and consist of spines, global ribs, and global multiplexers
(MUXes), as shown in Figure 3-1. The global networks are driven from the global rib at the center of the
die or quadrant global networks at the north or south side of the die. The global network uses the MUX
trees to access the spine, and the spine uses the clock ribs to access the VersaTile. Access is available
to the chip or quadrant global networks and the spines through the global MUXes. Access to the spine
using the global MUXes is explained in the "Spine Architecture" section on page 41.
These VersaNet global networks offer fast, low-skew routing resources for high-fanout nets, including
clock signals. In addition, these highly segmented global networks offer users the flexibility to create lowskew local clock networks using spines for up to 252 internal/external clocks or other high-fanout nets in
low power flash devices. Optimal usage of these low-skew networks can result in significant
improvement in design performance.
Quadrant Global Pads
T1
T2
High-Performance
Global Network
T3
I/ORing
Pad Ring
Pad Ring
Top Spine
Chip (main)
Global Pads
Chip (main)
Global Pads
Spine
Ribs
I/O Ring
Bottom Spine
Scope of Spine
(shaded area
plus local RAMs
and I/Os)
Spine-Selection
MUX
Embedded
RAM Blocks
Pad Ring
B1
B2
B3
Logic Tiles
Note: Not applicable to 10 k through 30 k gate devices
Figure 3-1 • Overview of VersaNet Global Network and Device Architecture
Revision 5
33
Global Resources in Low Power Flash Devices
2
2
2
2
Chip (main) Global
Network
2
2
Global Drivers
Global Drivers
Figure 3-2 • Simplified VersaNet Global Network (30 k gates and below)
North Quadrant Global Network
Quadrant Global Spine
CCC
3
3
Chip Global Spine
3
Chip (main)
Global
6 Network
6
3
6
6
3
CCC
6
6
3
CCC
3
CCC
3
6
3
6
3
South Quadrant Global Network
Figure 3-3 • Simplified VersaNet Global Network (60 k gates and above)
34
R e visio n 5
CCC
CCC
ProASIC3 nano FPGA Fabric User’s Guide
Chip and Quadrant Global I/Os
The following sections give an overview of naming conventions and other related I/O information.
Naming of Global I/Os
In low power flash devices, the global I/Os have access to certain clock conditioning circuitry and have
direct access to the global network. Additionally, the global I/Os can be used as regular I/Os, since they
have identical capabilities to those of regular I/Os. Due to the comprehensive and flexible nature of the
I/Os in low power flash devices, a naming scheme is used to show the details of the I/O. The global I/O
uses the generic name Gmn/IOuxwByVz. Note that Gmn refers to a global input pin and IOuxwByVz
refers to a regular I/O Pin, as these I/Os can be used as either global or regular I/Os. Refer to the I/O
Structures chapter of the user’s guide for the device that you are using for more information on this
naming convention.
Figure 3-4 represents the global input pins connection. It shows all 54 global pins available to
access the 18 global networks in ProASIC3E families.
Quadrant Global
Location A
+
+
+
GAAO/IOuxwByVz
GAA1/IOuxwByVz
GAA2/IOuxwByVz
GABO/IOuxwByVz
GAB1/IOuxwByVz
GAB2/IOuxwByVz
GACO/IOuxwByVz
GAC1/IOuxwByVz
GAC2/IOuxwByVz
+
+
+
+
+
Bankx
Bankx
+
+
3
Quadrant Global Spine
Bankx
3
Chip Global
Location C
6
6
6
6
GCAO/IOuxwByVz
GCA1/IOuxwByVz
GCA2/IOuxwByVz
GCBO/IOuxwByVz
GCB1/IOuxwByVz
GCB2/IOuxwByVz
GCCO/IOuxwByVz
GCC1/IOuxwByVz
GCC2/IOuxwByVz
+
6
3
+
3
+
+
3
Bankx
+
6
+
+
Chip Global Spine
+
3
3
3
3
Bankx
Bankx
+
6
+
6
+
Chip Global
Location F
GFAO/IOuxwByVz
GFA1/IOuxwByVz
GFA2/IOuxwByVz
GFBO/IOuxwByVz
GFB1/IOuxwByVz
GFB2/IOuxwByVz
GFCO/IOuxwByVz
GFC1/IOuxwByVz
GFC2/IOuxwByVz
3
+
3
3
3
3
+
Bankx
+
Bankx
+
+
+
+
+
+
CCC w it h PLL
+
CCC w it h or w it hout PLL
+
+
Quadrant Global
Location E
GBAO/IOuxwByVz
GBA1/IOuxwByVz
GBA2/IOuxwByVz
GBBO/IOuxwByVz
GBB1/IOuxwByVz
GBB2/IOuxwByVz
GBCO/IOuxwByVz
GBC1/IOuxwByVz
GBC2/IOuxwByVz
+
+
GEAO/IOuxwByVz
GEAC/IOuxwByVz
GEA2/IOuxwByVz
GEBO/IOuxwByVz
GEB1/IOuxwByVz
GEB2/IOuxwByVz
GECO/IOuxwByVz
GEC1/IOuxwByVz
GEC2/IOuxwByVz
Quadrant Global
Location B
+
CCC w it hout PLL
Quadrant Global
Location D
GDAO/IOuxwByVz
GDA1/IOuxwByVz
GDA2/IOuxwByVz
GDBO/IOuxwByVz
GDB1/IOuxwByVz
GDB2/IOuxwByVz
GDCO/IOuxwByVz
GDC1/IOuxwByVz
GDC2/IOuxwByVz
Figure 3-4 • Global Connections Details
Revision 5
35
Global Resources in Low Power Flash Devices
Figure 3-5 shows more detailed global input connections. It shows the global input pins connection to the
northwest quadrant global networks. Each global buffer, as well as the PLL reference clock, can be
driven from one of the following:
•
3 dedicated single-ended I/Os using a hardwired connection
•
2 dedicated differential I/Os using a hardwired connection (not supported for IGLOO nano or
ProASIC3 nano devices)
•
The FPGA core
Each shaded box represents an
INBUF or INBUF_LVDS/LVPECL
macro, as appropriate.
To Core
Sample Pin Names
GAA0/IO0NDB0V0
1
GAA1/IO00PDB0V0
1
+
Source for CCC
(CLKA or CLKB or CLKC)
GAA2/IO13PDB7V1
1
Routed Clock
2
(from FPGA core)
+
GAA[0:2]: GA represents global in the northwest corner
of the device. A[0:2]: designates specific A clock source.
Note: Differential inputs are not supported for IGLOO nano or ProASIC3 nano devices.
Figure 3-5 • Global I/O Overview
36
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Figure 3-6 shows all nine global inputs for the location A connected to the top left quadrant global
network via CCC.
GAAO/
IOuxwByVz
GAA1/
IOuxwByVz
MUX
CLKA
MUX
CLKB
Quadrant Global for CLKA
GAA2/
IOuxwByVz
GABO/
IOuxwByVz
GAB1/
IOuxwByVz
CCC
Quadrant Global for CLKB
GAB2/
IOuxwByVz
GACO/
IOuxwByVz
GAC1/
IOuxwByVz
MUX
CLKC
Quadrant Global for CLKC
GAC2/
IOuxwByVz
Figure 3-6 • Global Inputs
Since each bank can have a different I/O standard, the user should be careful to choose the correct
global I/O for the design. There are 54 global pins available to access 18 global networks. For the singleended and voltage-referenced I/O standards, you can use any of these three available I/Os to access the
global network. For differential I/O standards such as LVDS and LVPECL, the I/O macro needs to be
placed on (A0, A1), (B0, B1), (C0, C1), or a similar location. The unassigned global I/Os can be used
as regular I/Os. Note that pin names starting with GF and GC are associated with the chip global
networks, and GA, GB, GD, and GE are used for quadrant global networks. Table 3-2 on page 38 and
Table 3-3 on page 39 show the general chip and quadrant global pin names.
Revision 5
37
Global Resources in Low Power Flash Devices
Table 3-2 • Chip Global Pin Name
I/O Type
Single-Ended
Beginning of I/O Name
Notes
GFAO/IOuxwByVz
Only one of the I/Os can be directly connected to a chip
global at a time.
GFA1/IOuxwByVz
GFA2/IOuxwByVz
GFBO/IOuxwByVz
GFB1/IOuxwByVz
Only one of the I/Os can be directly connected to a chip
global at a time.
GFB2/IOuxwByVz
GFC0/IOuxwByVz
GFC1/IOuxwByVz
Only one of the I/Os can be directly connected to a chip
global at a time.
GFC2/IOuxwByVz
GCAO/IOuxwByVz
GCA1/IOuxwByVz
Only one of the I/Os can be directly connected to a chip
global at a time.
GCA2/IOuxwByVz
GCBO/IOuxwByVz
GCB1/IOuxwByVz
Only one of the I/Os can be directly connected to a chip
global at a time.
GCB2/IOuxwByVz
GCC0/IOuxwByVz
GCC1/IOuxwByVz
Only one of the I/Os can be directly connected to a chip
global at a time.
GCC2/IOuxwByVz
Differential I/O Pairs
GFAO/IOuxwByVz
The output of the different pair will drive the chip global.
GFA1/IOuxwByVz
GFBO/IOuxwByVz
The output of the different pair will drive the chip global.
GFB1/IOuxwByVz
GFCO/IOuxwByVz
The output of the different pair will drive the chip global.
GFC1/IOuxwByVz
GCAO/IOuxwByVz
The output of the different pair will drive the chip global.
GCA1/IOuxwByVz
GCBO/IOuxwByVz
The output of the different pair will drive the chip global.
GCB1/IOuxwByVz
GCCO/IOuxwByVz
The output of the different pair will drive the chip global.
GCC1/IOuxwByVz
Note: Only one of the I/Os can be directly connected to a quadrant at a time.
38
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Table 3-3 • Quadrant Global Pin Name
I/O Type
Single-Ended
Beginning of I/O Name
Notes
GAAO/IOuxwByVz
Only one of the I/Os can be directly connected to a
quadrant global at a time
GAA1/IOuxwByVz
GAA2/IOuxwByVz
GABO/IOuxwByVz
GAB1/IOuxwByVz
Only one of the I/Os can be directly connected to a
quadrant global at a time.
GAB2/IOuxwByVz
GAC0/IOuxwByVz
GAC1/IOuxwByVz
Only one of the I/Os can be directly connected to a
quadrant global at a time.
GAC2/IOuxwByVz
GBAO/IOuxwByVz
GBA1/IOuxwByVz
Only one of the I/Os can be directly connected to a global
at a time.
GBA2/IOuxwByVz
GBBO/IOuxwByVz
GBB1/IOuxwByVz
Only one of the I/Os can be directly connected to a global
at a time.
GBB2/IOuxwByVz
GBC0/IOuxwByVz
GBC1/IOuxwByVz
Only one of the I/Os can be directly connected to a global
at a time.
GBC2/IOuxwByVz
GDAO/IOuxwByVz
GDA1/IOuxwByVz
Only one of the I/Os can be directly connected to a global
at a time.
GDA2/IOuxwByVz
GDBO/IOuxwByVz
GDB1/IOuxwByVz
Only one of the I/Os can be directly connected to a global
at a time.
GDB2/IOuxwByVz
GDC0/IOuxwByVz
GDC1/IOuxwByVz
Only one of the I/Os can be directly connected to a global
at a time.
GDC2/IOuxwByVz
GEAO/IOuxwByVz
GEA1/IOuxwByVz
Only one of the I/Os can be directly connected to a global
at a time.
GEA2/IOuxwByVz
GEBO/IOuxwByVz
GEB1/IOuxwByVz
Only one of the I/Os can be directly connected to a global
at a time.
GEB2/IOuxwByVz
GEC0/IOuxwByVz
GEC1/IOuxwByVz
Only one of the I/Os can be directly connected to a global
at a time.
GEC2/IOuxwByVz
Note: Only one of the I/Os can be directly connected to a quadrant at a time.
Revision 5
39
Global Resources in Low Power Flash Devices
Table 3-3 • Quadrant Global Pin Name (continued)
Differential I/O Pairs
GAAO/IOuxwByVz
The output of the different pair will drive the global.
GAA1/IOuxwByVz
GABO/IOuxwByVz
The output of the different pair will drive the global.
GAB1/IOuxwByVz
GACO/IOuxwByVz
The output of the different pair will drive the global.
GAC1/IOuxwByVz
GBAO/IOuxwByVz
The output of the different pair will drive the global.
GBA1/IOuxwByVz
GBBO/IOuxwByVz
The output of the different pair will drive the global.
GBB1/IOuxwByVz
GBCO/IOuxwByVz
The output of the different pair will drive the global.
GBC1/IOuxwByVz
GDAO/IOuxwByVz
The output of the different pair will drive the global.
GDA1/IOuxwByVz
GDBO/IOuxwByVz
The output of the different pair will drive the global.
GDB1/IOuxwByVz
GDCO/IOuxwByVz
The output of the different pair will drive the global.
GDC1/IOuxwByVz
GEAO/IOuxwByVz
The output of the different pair will drive the global.
GEA1/IOuxwByVz
GEBO/IOuxwByVz
The output of the different pair will drive the global.
GEB1/IOuxwByVz
GECO/IOuxwByVz
The output of the different pair will drive the global.
GEC1/IOuxwByVz
Note: Only one of the I/Os can be directly connected to a quadrant at a time.
Unused Global I/O Configuration
The unused clock inputs behave similarly to the unused Pro I/Os. The Microsemi Designer software
automatically configures the unused global pins as inputs with pull-up resistors if they are not used as
regular I/O.
I/O Banks and Global I/O Standards
In low power flash devices, any I/O or internal logic can be used to drive the global network. However,
only the global macro placed at the global pins will use the hardwired connection between the I/O and
global network. Global signal (signal driving a global macro) assignment to I/O banks is no different from
regular I/O assignment to I/O banks with the exception that you are limited to the pin placement location
available. Only global signals compatible with both the VCCI and VREF standards can be assigned to the
same bank.
40
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Spine Architecture
The low power flash device architecture allows the VersaNet global networks to be segmented. Each of
these networks contains spines (the vertical branches of the global network tree) and ribs that can reach
all the VersaTiles inside its region. The nine spines available in a vertical column reside in global
networks with two separate regions of scope: the quadrant global network, which has three spines, and
the chip (main) global network, which has six spines. Note that the number of quadrant globals and
globals/spines per tree varies depending on the specific device. Refer to Table 3-4 for the clocking
resources available for each device. The spines are the vertical branches of the global network tree,
shown in Figure 3-3 on page 34. Each spine in a vertical column of a chip (main) global network is further
divided into two spine segments of equal lengths: one in the top and one in the bottom half of the die
(except in 10 k through 30 k gate devices).
Top and bottom spine segments radiating from the center of a device have the same height. However,
just as in the ProASICPLUS® family, signals assigned only to the top and bottom spine cannot access the
middle two rows of the die. The spines for quadrant clock networks do not cross the middle of the die and
cannot access the middle two rows of the architecture.
Each spine and its associated ribs cover a certain area of the device (the "scope" of the spine; see
Figure 3-3 on page 34). Each spine is accessed by the dedicated global network MUX tree architecture,
which defines how a particular spine is driven—either by the signal on the global network from a CCC, for
example, or by another net defined by the user. Details of the chip (main) global network spine-selection
MUX are presented in Figure 3-8 on page 44. The spine drivers for each spine are located in the middle
of the die.
Quadrant spines can be driven from user I/Os or an internal signal from the north and south sides of the
die. The ability to drive spines in the quadrant global networks can have a significant effect on system
performance for high-fanout inputs to a design. Access to the top quadrant spine regions is from the top
of the die, and access to the bottom quadrant spine regions is from the bottom of the die. The A3PE3000
device has 28 clock trees and each tree has nine spines; this flexible global network architecture enables
users to map up to 252 different internal/external clocks in an A3PE3000 device.
Table 3-4 • Globals/Spines/Rows for IGLOO and ProASIC3 Devices
ProASIC3/
ProASIC3L
Devices
IGLOO
Devices
Rows
Globals/ Total
in
Quadrant
Spines Spines VersaTiles
in Each
Total
Each
Globals Clock
per
per
Chip
Trees
Tree Device
Tree
VersaTiles Spine
Globals (4×3)
A3PN010
AGLN010
4
0
1
0
0
260
260
4
A3PN015
AGLN015
4
0
1
0
0
384
384
6
A3PN020
AGLN020
4
0
1
0
0
520
520
6
A3PN060
AGLN060
6
12
4
9
36
384
1,536
12
A3PN125
AGLN125
6
12
8
9
72
384
3,072
12
A3PN250
AGLN250
6
12
8
9
72
768
6,144
24
A3P015
AGL015
6
0
1
9
9
384
384
12
A3P030
AGL030
6
0
2
9
18
384
768
12
A3P060
AGL060
6
12
4
9
36
384
1,536
12
A3P125
AGL125
6
12
8
9
72
384
3,072
12
A3P250/L
AGL250
6
12
8
9
72
768
6,144
24
A3P400
AGL400
6
12
12
9
108
768
9,216
24
A3P600/L
AGL600
6
12
12
9
108
1,152
13,824
36
A3P1000/L
AGL1000
6
12
16
9
144
1,536
24,576
48
A3PE600/L
AGLE600
6
12
12
9
108
1,120
13,440
35
A3PE1500
6
12
20
9
180
1,888
37,760
59
A3PE3000/L AGLE3000
6
12
28
9
252
2,656
74,368
83
Revision 5
41
Global Resources in Low Power Flash Devices
Table 3-5 • Globals/Spines/Rows for IGLOO PLUS Devices
IGLOO
PLUS
Devices
Quadrant
Chip
Globals Clock
Globals
(4×3)
Trees
Rows
Globals/
Total
in
Spines
Spines
VersaTiles
Total
Each
per Tree per Device in Each Tree VersaTiles Spine
AGLP030
6
0
2
9
18
384*
792
12
AGLP060
6
12
4
9
36
384*
1,584
12
AGLP125
6
12
8
9
72
384*
3,120
12
Note: *Clock trees that are located at far left and far right will support more VersaTiles.
Table 3-6 • Globals/Spines/Rows for Fusion Devices
42
Clock
Trees
Globals/
Spines
per
Tree
Total
Spines
per
Device
VersaTiles
in
Each
Tree
Total
VersaTiles
Rows
in
Each
Spine
Fusion
Device
Chip
Globals
Quadrant
Globals
(4×3)
AFS090
6
12
6
9
54
384
2,304
12
AFS250
6
12
8
9
72
768
6,144
24
AFS600
6
12
12
9
108
1,152
13,824
36
AFS1500
6
12
20
9
180
1,920
38,400
60
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Spine Access
The physical location of each spine is identified by the letter T (top) or B (bottom) and an accompanying
number (Tn or Bn). The number n indicates the horizontal location of the spine; 1 refers to the first spine
on the left side of the die. Since there are six chip spines in each spine tree, there are up to six spines
available for each combination of T (or B) and n (for example, six T1 spines). Similarly, there are three
quadrant spines available for each combination of T (or B) and n (for example, four T1 spines), as shown
in Figure 3-7.
Tn
Tn+1
Tn+2
Tn+3
Tn+4
Tn
Tn+1
Tn+2
Tn+3
Tn+4
C
Global
Network
B
A
A
B
Global
Network
C
Figure 3-7 • Chip Global Aggregation
A spine is also called a local clock network, and is accessed by the dedicated global MUX architecture.
These MUXes define how a particular spine is driven. Refer to Figure 3-8 on page 44 for the global MUX
architecture. The MUXes for each chip global spine are located in the middle of the die. Access to the top
and bottom chip global spine is available from the middle of the die. There is no control dependency
between the top and bottom spines. If a top spine, T1, of a chip global network is assigned to a net, B1 is
not wasted and can be used by the global clock network. The signal assigned only to the top or bottom
spine cannot access the middle two rows of the architecture. However, if a spine is using the top and
bottom at the same time (T1 and B1, for instance), the previous restriction is lifted.
The MUXes for each quadrant global spine are located in the north and south sides of the die. Access to
the top and bottom quadrant global spines is available from the north and south sides of the die. Since
the MUXes for quadrant spines are located in the north and south sides of the die, you should not try to
drive T1 and B1 quadrant spines from the same signal.
Revision 5
43
Global Resources in Low Power Flash Devices
Using Clock Aggregation
Clock aggregation allows for multi-spine clock domains to be assigned using hardwired connections,
without adding any extra skew. A MUX tree, shown in Figure 3-8, provides the necessary flexibility to
allow long lines, local resources, or I/Os to access domains of one, two, or four global spines. Signal
access to the clock aggregation system is achieved through long-line resources in the central rib in the
center of the die, and also through local resources in the north and south ribs, allowing I/Os to feed
directly into the clock system. As Figure 3-9 indicates, this access system is contiguous.
There is no break in the middle of the chip for the north and south I/O VersaNet access. This is different
from the quadrant clocks located in these ribs, which only reach the middle of the rib.
Internal/External
Signals
Internal/External
Signals
Tree Node MUX
Tree Node MUX
Internal/External
Signal
Tree Node MUX
Global Rib
Internal/External
Signal
Global Driver MUX
Spine
Figure 3-8 • Spine Selection MUX of Global Tree
Global Spine
Global Rib
Global Driver and MUX
Tree Node MUX
I/O Access
Internal Signal Access
Global Signal Access
Figure 3-9 • Clock Aggregation Tree Architecture
44
R e visio n 5
I/O Tiles
ProASIC3 nano FPGA Fabric User’s Guide
Clock Aggregation Architecture
This clock aggregation feature allows a balanced clock tree, which improves clock skew. The physical
regions for clock aggregation are defined from left to right and shift by one spine. For chip global
networks, there are three types of clock aggregation available, as shown in Figure 3-10:
•
Long lines that can drive up to four adjacent spines (A)
•
Long lines that can drive up to two adjacent spines (B)
•
Long lines that can drive one spine (C)
There are three types of clock aggregation available for the quadrant spines, as shown in Figure 3-10:
•
I/Os or local resources that can drive up to four adjacent spines
•
I/Os or local resources that can drive up to two adjacent spines
•
I/Os or local resources that can drive one spine
As an example, A3PE600 and AFS600 devices have twelve spine locations: T1, T2, T3, T4, T5, T6, B1,
B2, B3, B4, B5, and B6. Table 3-7 shows the clock aggregation you can have in A3PE600 and
AFS600.
A
B
C
Tn
Tn + 1
Tn + 2
Tn + 3
Tn + 4
Figure 3-10 • Four Spines Aggregation
Table 3-7 • Spine Aggregation in A3PE600 or AFS600
Clock Aggregation
Spine
1 spine
T1, T2, T3, T4, T5, T6, B1, B2, B3, B4, B5, B6
2 spines
T1:T2, T2:T3, T3:T4, T4:T5, T5:T6, B1:B2, B2:B3, B3:B4, B4:B5, B5:B6
4 spines
B1:B4, B2:B5, B3:B6, T1:T4, T2:T5, T3:T6
The clock aggregation for the quadrant spines can cross over from the left to right quadrant, but not from
top to bottom. The quadrant spine assignment T1:T4 is legal, but the quadrant spine assignment T1:B1
is not legal. Note that this clock aggregation is hardwired. You can always assign signals to spine T1 and
B2 by instantiating a buffer, but this may add skew in the signal.
Revision 5
45
Global Resources in Low Power Flash Devices
Design Recommendations
The following sections provide design flow recommendations for using a global network in a design.
•
"Global Macros and I/O Standards"
•
"Global Macro and Placement Selections" on page 48
•
"Using Global Macros in Synplicity" on page 50
•
"Global Promotion and Demotion Using PDC" on page 51
•
"Spine Assignment" on page 52
•
"Designer Flow for Global Assignment" on page 53
•
"Simple Design Example" on page 55
•
"Global Management in PLL Design" on page 57
•
"Using Spines of Occupied Global Networks" on page 58
Global Macros and I/O Standards
The larger low power flash devices have six chip global networks and four quadrant global networks.
However, the same clock macros are used for assigning signals to chip globals and quadrant globals.
Depending on the clock macro placement or assignment in the Physical Design Constraint (PDC) file or
MultiView Navigator (MVN), the signal will use the chip global network or quadrant network. Table 3-8
lists the clock macros available for low power flash devices. Refer to the IGLOO, ProASIC3,
SmartFusion, and Fusion Macro Library Guide for details.
Table 3-8 • Clock Macros
Macro Name
CLKBUF
Description
Symbol
Input macro for Clock Network
CLKBUF
Y
PAD
CLKBUF_x
Input macro for Clock Network
with specific I/O standard
CLKBUF_X
CLKBUF_LVDS/LVPECL LVDS or LVPECL input macro PADP
for
Clock
Network
(not
supported for IGLOO nano or
CLKBUF_LVDS
ProASIC3 nano devices)
PADN
CLKINT
Y
PAD
PADP
Y
CLKBUF_LVPECL
Y
PADN
Macro for internal clock interface
A
Y
CLKINT
CLKBIBUF
Bidirectional macro with input
dedicated to routed Clock
Network
D
Y
E
PAD
CLKBIBUF
Use these available macros to assign a signal to the global network. In addition to these global macros,
PLL and CLKDLY macros can also drive the global networks. Use I/O–standard–specific clock macros
(CLKBUF_x) to instantiate a specific I/O standard for the global signals. Table 3-9 on page 47 shows the
list of these I/O–standard–specific macros. Note that if you use these I/O–standard–specific clock
macros, you cannot change the I/O standard later in the design stage. If you use the regular CLKBUF
macro, you can use MVN or the PDC file in Designer to change the I/O standard. The default I/O
46
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
standard for CLKBUF is LVTTL in the current Microsemi Libero® System-on-Chip (SoC) and Designer
software.
Table 3-9 • I/O Standards within CLKBUF
Name
Description
CLKBUF_LVCMOS5
LVCMOS clock buffer with 5.0 V CMOS voltage level
CLKBUF_LVCMOS33
LVCMOS clock buffer with 3.3 V CMOS voltage level
CLKBUF_LVCMOS25
LVCMOS clock buffer with 2.5 V CMOS voltage level1
CLKBUF_LVCMOS18
LVCMOS clock buffer with 1.8 V CMOS voltage level
CLKBUF_LVCMOS15
LVCMOS clock buffer with 1.5 V CMOS voltage level
CLKBUF_LVCMOS12
LVCMOS clock buffer with 1.2 V CMOS voltage level
CLKBUF_PCI
PCI clock buffer
CLKBUF_PCIX
PCIX clock buffer
CLKBUF_GTL25
GTL clock buffer with 2.5 V CMOS voltage level1
CLKBUF_GTL33
GTL clock buffer with 3.3 V CMOS voltage level1
CLKBUF_GTLP25
GTL+ clock buffer with 2.5 V CMOS voltage level1
CLKBUF_GTLP33
GTL+ clock buffer with 3.3 V CMOS voltage level1
CLKBUF_ HSTL _I
HSTL Class I clock buffer1
CLKBUF_ HSTL _II
HSTL Class II clock buffer1
CLKBUF_SSTL2_I
SSTL2 Class I clock buffer1
CLKBUF_SSTL2_II
SSTL2 Class II clock buffer1
CLKBUF_SSTL3_I
SSTL3 Class I clock buffer1
CLKBUF_SSTL3_II
SSTL3 Class II clock buffer1
Notes:
1. Supported in only the IGLOOe, ProASIC3E, AFS600, and AFS1500 devices
2. By default, the CLKBUF macro uses the 3.3 V LVTTL I/O technology.
The current synthesis tool libraries only infer the CLKBUF or CLKINT macros in the netlist. All other
global macros must be instantiated manually into your HDL code. The following is an example of
CLKBUF_LVCMOS25 global macro instantiations that you can copy and paste into your code:
VHDL
component clkbuf_lvcmos25
port (pad : in std_logic; y : out std_logic);
end component
begin
-- concurrent statements
u2 : clkbuf_lvcmos25 port map (pad =>
end
ext_clk, y => int_clk);
Verilog
module design (______);
input _____;
output ______;
clkbuf_lvcmos25 u2 (.y(int_clk), .pad(ext_clk);
endmodule
Revision 5
47
Global Resources in Low Power Flash Devices
Global Macro and Placement Selections
Low power flash devices provide the flexibility of choosing one of the three global input pad locations
available to connect to a global / quadrant global network. For 60K gate devices and above, if the
single-ended I/O standard is chosen, there is flexibility to choose one of the global input pads (the first,
second, and fourth input). Once chosen, the other I/O locations are used as regular I/Os. If the differential
I/O standard is chosen, the first and second inputs are considered as paired, and the third input is paired
with a regular I/O. The user then has the choice of selecting one of the two sets to be used as the global
input source. There is also the option to allow an internal clock signal to feed the global network. A
multiplexer tree selects the appropriate global input for routing to the desired location. Note that the
global I/O pads do not need to feed the global network; they can also be used as regular I/O pads.
Hardwired I/O Clock Source
Hardwired I/O refers to global input pins that are hardwired to the multiplexer tree, which directly
accesses the global network. These global input pins have designated pin locations and are indicated
with the I/O naming convention Gmn (m refers to any one of the positions where the global buffers is
available, and n refers to any one of the three global input MUXes and the pin number of the associated
global location, m). Choosing this option provides the benefit of directly connecting to the global buffers,
which provides less delay. See Figure 3-11 for an example illustration of the connections, shown in red. If
a CLKBUF macro is initiated, the clock input can be placed at one of nine dedicated global input pin
locations: GmA0, GmA1, GmA2, GmB0, GmB1, GmB2, GmC0, GmC1, or GmC2. Note that the
placement of the global will determine whether you are using chip global or quadrant global. For
example, if the CLKBIF is placed in one of the GF pin locations, it will use the chip global network; if the
CLKBIF is placed in one of the GA pin locations, it will use quadrant global network. This is shown in
Figure 3-12 on page 49 and Figure 3-13 on page 49.
To Core
GFA0
GFA1
+
To global network
GFA2
+
From FPGA core
Figure 3-11 • CLKBUF Macro
48
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Bankx
Bankx
Location B
Bankx
Bankx
Location C
Chip Global Region
Bankx
Location F
Bankx
CLKBUF placed at one of the GF pin locations
Location A
Bankx
Location E
Bankx
Location D
Figure 3-12 • Chip Global Region
CLKBUF placed at one of the GA pin locations
Bankx
Bankx
Location B
Bankx
Bankx
Location A
Quadrant Global Region
Location E
Bankx
Location C
Bankx
Location F
Bankx
Bankx
Location D
Figure 3-13 • Quadrant Global Region
Revision 5
49
Global Resources in Low Power Flash Devices
External I/O or Local signal as Clock Source
External I/O refers to regular I/O pins are labeled with the I/O convention IOuxwByVz. You can allow the
external I/O or internal signal to access the global. To allow the external I/O or internal signal to access
the global network, you need to instantiate the CLKINT macro. Refer to Figure 3-4 on page 35 for an
example illustration of the connections. Instead of using CLKINT, you can also use PDC to promote
signals from external I/O or internal signal to the global network. However, it may cause layout issues
because of synthesis logic replication. Refer to the "Global Promotion and Demotion Using PDC" section
on page 51 for details.
CLKINT
INBUF
To Core
GFA0
GFA1
+
To global network
GFA2
+
From FPGA core
INBUF
Figure 3-14 • CLKINT Macro
Using Global Macros in Synplicity
The Synplify® synthesis tool automatically inserts global buffers for nets with high fanout during
synthesis. By default, Synplicity® puts six global macros (CLKBUF or CLKINT) in the netlist, including
any global instantiation or PLL macro. Synplify always honors your global macro instantiation. If you have
a PLL (only primary output is used) in the design, Synplify adds five more global buffers in the netlist.
Synplify uses the following global counting rule to add global macros in the netlist:
1. CLKBUF: 1 global buffer
2. CLKINT: 1 global buffer
3. CLKDLY: 1 global buffer
4. PLL: 1 to 3 global buffers
50
–
GLA, GLB, GLC, YB, and YC are counted as 1 buffer.
–
GLB or YB is used or both are counted as 1 buffer.
–
GLC or YC is used or both are counted as 1 buffer.
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
CLKA
EXTFB
POWERDOWN
OADIVRST
GLA
LOCK
GLB
YB
GLC
YC
Note: OAVDIVRST exists only in the Fusion PLL.
Figure 3-15 • PLLs in Low Power Flash Devices
You can use the syn_global_buffers attribute in Synplify to specify a maximum number of global macros
to be inserted in the netlist. This can also be used to restrict the number of global buffers inserted. In the
Synplicity 8.1 version or newer, a new attribute, syn_global_minfanout, has been added for low power
flash devices. This enables you to promote only the high-fanout signal to global. However, be aware that
you can only have six signals assigned to chip global networks, and the rest of the global signals should
be assigned to quadrant global networks. So, if the netlist has 18 global macros, the remaining 12 global
macros should have fanout that allows the instances driven by these globals to be placed inside a
quadrant.
Global Promotion and Demotion Using PDC
The HDL source file or schematic is the preferred place for defining which signals should be assigned to
a clock network using clock macro instantiation. This method is preferred because it is guaranteed to be
honored by the synthesis tools and Designer software and stop any replication on this net by the
synthesis tool. Note that a signal with fanout may have logic replication if it is not promoted to global
during synthesis. In that case, the user cannot promote that signal to global using PDC. See Synplicity
Help for details on using this attribute. To help you with global management, Designer allows you to
promote a signal to a global network or demote a global macro to a regular macro from the user netlist
using the compile options and/or PDC commands.
The following are the PDC constraints you can use to promote a signal to a global network:
1. PDC syntax to promote a regular net to a chip global clock:
assign_global_clock –net netname
The following will happen during promotion of a regular signal to a global network:
–
If the net is external, the net will be driven by a CLKINT inserted automatically by Compile.
–
The I/O macro will not be changed to CLKBUF macros.
–
If the net is an internal net, the net will be driven by a CLKINT inserted automatically by
Compile.
2. PDC syntax to promote a net to a quadrant clock:
assign_local_clock –net netname –type quadrant UR|UL|LR|LL
This follows the same rule as the chip global clock network.
The following PDC command demotes the clock nets to regular nets.
unassign_global_clock -net netname
Revision 5
51
Global Resources in Low Power Flash Devices
The following will happen during demotion of a global signal to regular nets:
•
CLKBUF_x becomes INBUF_x; CLKINT is removed from the netlist.
•
The essential global macro, such as the output of the Clock Conditioning Circuit, cannot be
demoted.
•
No automatic buffering will happen.
Since no automatic buffering happens when a signal is demoted, this net may have a high delay due to
large fanout. This may have a negative effect on the quality of the results. Microsemi recommends that
the automatic global demotion only be used on small-fanout nets. Use clock networks for high-fanout
nets to improve timing and routability.
Spine Assignment
The low power flash device architecture allows the global networks to be segmented and used as clock
spines. These spines, also called local clock networks, enable the use of PDC or MVN to assign a signal
to a spine.
PDC syntax to promote a net to a spine/local clock:
assign_local_clock –net netname –type [quadrant|chip] Tn|Bn|Tn:Bm
If the net is driven by a clock macro, Designer automatically demotes the clock net to a regular net before
it is assigned to a spine. Nets driven by a PLL or CLKDLY macro cannot be assigned to a local clock.
When assigning a signal to a spine or quadrant global network using PDC (pre-compile), the Designer
software will legalize the shared instances. The number of shared instances to be legalized can be
controlled by compile options. If these networks are created in MVN (only quadrant globals can be
created), no legalization is done (as it is post-compile). Designer does not do legalization between nonclock nets.
As an example, consider two nets, net_clk and net_reset, driving the same flip-flop. The following PDC
constraints are used:
assign_local_clock –net net_clk –type chip T3
assign_local_clock –net net_reset –type chip T1:T2
During Compile, Designer adds a buffer in the reset net and places it in the T1 or T2 region, and places
the flip-flop in the T3 spine region (Figure 3-16).
After Compile
Before Compile
D
D
net_clk
CLK
CLK
net_clk
CLR
CLR
net_reset
net_reset
T1
T2
Added
buffer
T3
assign_local_clock -net net_clk -type chip T3
assign_local_clock -net net_reset -type chip T1:T2
Figure 3-16 • Adding a Buffer for Shared Instances
52
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
You can control the maximum number of shared instances allowed for the legalization to take place using
the Compile Option dialog box shown in Figure 3-17. Refer to Libero SoC / Designer online help for
details on the Compile Option dialog box. A large number of shared instances most likely indicates a
floorplanning problem that you should address.
Figure 3-17 • Shared Instances in the Compile Option Dialog Box
Designer Flow for Global Assignment
To achieve the desired result, pay special attention to global management during synthesis and placeand-route. The current Synplify tool does not insert more than six global buffers in the netlist by default.
Thus, the default flow will not assign any signal to the quadrant global network. However, you can use
attributes in Synplify and increase the default global macro assignment in the netlist. Designer v6.2
supports automatic quadrant global assignment, which was not available in Designer v6.1. Layout will
make the choice to assign the correct signals to global. However, you can also utilize PDC and perform
manual global assignment to overwrite any automatic assignment. The following step-by-step
suggestions guide you in the layout of your design and help you improve timing in Designer:
1. Run Compile and check the Compile report. The Compile report has global information in the
"Device Utilization" section that describes the number of chip and quadrant signals in the design.
A "Net Report" section describes chip global nets, quadrant global nets, local clock nets, a list of
nets listed by fanout, and net candidates for local clock assignment. Review this information. Note
that YB or YC are counted as global only when they are used in isolation; if you use YB only and
not GLB, this net is not shown in the global/quadrant nets report. Instead, it appears in the Global
Utilization report.
2. If some signals have a very high fanout and are candidates for global promotion, promote those
signals to global using the compile options or PDC commands. Figure 3-18 on page 54 shows the
Globals Management section of the compile options. Select Promote regular nets whose
fanout is greater than and enter a reasonable value for fanouts.
Revision 5
53
Global Resources in Low Power Flash Devices
Figure 3-18 • Globals Management GUI in Designer
3. Occasionally, the synthesis tool assigns a global macro to clock nets, even though the fanout is
significantly less than other asynchronous signals. Select Demote global nets whose fanout is
less than and enter a reasonable value for fanouts. This frees up some global networks from the
signals that have very low fanouts. This can also be done using PDC.
4. Use a local clock network for the signals that do not need to go to the whole chip but should have
low skew. This local clock network assignment can only be done using PDC.
5. Assign the I/O buffer using MVN if you have fixed I/O assignment. As shown in Figure 3-10 on
page 45, there are three sets of global pins that have a hardwired connection to each global
network. Do not try to put multiple CLKBUF macros in these three sets of global pins. For
example, do not assign two CLKBUFs to GAA0x and GAA2x pins.
6. You must click Commit at the end of MVN assignment. This runs the pre-layout checker and
checks the validity of global assignment.
7. Always run Compile with the Keep existing physical constraints option on. This uses the
quadrant clock network assignment in the MVN assignment and checks if you have the desired
signals on the global networks.
8. Run Layout and check the timing.
54
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Simple Design Example
Consider a design consisting of six building blocks (shift registers) and targeted for an A3PE600-PQ208
(Figure 3-16 on page 52). The example design consists of two PLLs (PLL1 has GLA only; PLL2 has both
GLA and GLB), a global reset (ACLR), an enable (EN_ALL), and three external clock domains (QCLK1,
QCLK2, and QCLK3) driving the different blocks of the design. Note that the PQ208 package only has
two PLLs (which access the chip global network). Because of fanout, the global reset and enable signals
need to be assigned to the chip global resources. There is only one free chip global for the remaining
global (QCLK1, QCLK2, QCLK3). Place two of these signals on the quadrant global resource. The
design example demonstrates manually assignment of QCLK1 and QCLK2 to the quadrant global using
the PDC command.
PDOWN
PLLZ_CLKA
PLL2
POWER-DOWN LOCK
GLA
CLKA
GLB
\$116
reg256_behave
Shhl_In
Shhl_In Shhl_out
Adr
Clock
REG_PLLCLK2GLA
reg256_behave
Shhl_In
Shhl_In Shhl_out
Adr
Clock
DATA_QCLK1
QCLK1
REG_PLLCLK2GLA_OUT
REG_QCLK1_OUT
REG_QCLK1
DATA_PLLCQCLK2
EN_ALL
DATA_QCLK2
ACLR
QCLK2
reg256_behave
Shhl_In
Shhl_In Shhl_out
Adr
Clock
REG_QCLK2
REG_QCLK2_OUT
reg256_behave
Shhl_In
Shhl_In Shhl_out
Adr
Clock
REG_PLLCLK2GLB_OUT
REG_PLLCLK2GLB
reg256_behave
Shhl_In
Shhl_In Shhl_out
Adr
Clock
DATA_QCLK3
QCLK3
REG_QCLK3_OUT
REG_QCLK3
DATA_PLLCLK1
PLL1_CLKA
PLL1
POWER-DOWN LOCK
GLA
CLKA
\$115
reg256_behave
Shhl_In
Shhl_In Shhl_out
Adr
Clock
REG_PLLCLK1_OUT
REG_PLLCLK1
Figure 3-19 • Block Diagram of the Global Management Example Design
Revision 5
55
Global Resources in Low Power Flash Devices
Step 1
Run Synthesis with default options. The Synplicity log shows the following device utilization:
Cell usage:
cell count
area
count*area
DFN1E1C1
1536
2.0
3072.0
BUFF
278
1.0
278.0
INBUF
10
0.0
0.0
VCC
9
0.0
0.0
GND
9
0.0
0.0
OUTBUF
6
0.0
0.0
CLKBUF
3
0.0
0.0
PLL
2
0.0
0.0
TOTAL
3350.0
1853
Step 2
Run Compile with the Promote regular nets whose fanout is greater than option selected in Designer;
you will see the following in the Compile report:
Device utilization report:
==========================
CORE
Used:
1536 Total: 13824
(11.11%)
IO (W/ clocks)
Used:
19 Total:
147
(12.93%)
Differential IO
Used:
0 Total:
65
(0.00%)
GLOBAL
Used:
8 Total:
18
(44.44%)
PLL
Used:
2 Total:
2 (100.00%)
RAM/FIFO
Used:
0 Total:
24
(0.00%)
FlashROM
Used:
0 Total:
1
(0.00%)
……………………
The following nets have been assigned to a global resource:
Fanout Type
Name
-------------------------1536
INT_NET
Net
: EN_ALL_c
Driver: EN_ALL_pad_CLKINT
Source: AUTO PROMOTED
1536
SET/RESET_NET Net
: ACLR_c
Driver: ACLR_pad_CLKINT
Source: AUTO PROMOTED
256
CLK_NET
Net
: QCLK1_c
Driver: QCLK1_pad_CLKINT
Source: AUTO PROMOTED
256
CLK_NET
Net
: QCLK2_c
Driver: QCLK2_pad_CLKINT
Source: AUTO PROMOTED
256
CLK_NET
Net
: QCLK3_c
Driver: QCLK3_pad_CLKINT
Source: AUTO PROMOTED
256
CLK_NET
Net
: $1N14
Driver: $1I5/Core
Source: ESSENTIAL
256
CLK_NET
Net
: $1N12
Driver: $1I6/Core
Source: ESSENTIAL
256
CLK_NET
Net
: $1N10
Driver: $1I6/Core
Source: ESSENTIAL
Designer will promote five more signals to global due to high fanout. There are eight signals assigned to
global networks.
56
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
During Layout, Designer will assign two of the signals to quadrant global locations.
Step 3 (optional)
You can also assign the QCLK1_c and QCLK2_c nets to quadrant regions using the following PDC
commands:
assign_local_clock –net QCLK1_c
assign_local_clock –net QCLK2_c
–type quadrant UL
–type quadrant LL
Step 4
Import this PDC with the netlist and run Compile again. You will see the following in the Compile report:
The following nets have been assigned to a global resource:
Fanout Type
Name
-------------------------1536
INT_NET
Net
: EN_ALL_c
Driver: EN_ALL_pad_CLKINT
Source: AUTO PROMOTED
1536
SET/RESET_NET Net
: ACLR_c
Driver: ACLR_pad_CLKINT
Source: AUTO PROMOTED
256
CLK_NET
Net
: QCLK3_c
Driver: QCLK3_pad_CLKINT
Source: AUTO PROMOTED
256
CLK_NET
Net
: $1N14
Driver: $1I5/Core
Source: ESSENTIAL
256
CLK_NET
Net
: $1N12
Driver: $1I6/Core
Source: ESSENTIAL
256
CLK_NET
Net
: $1N10
Driver: $1I6/Core
Source: ESSENTIAL
The following nets have been assigned to a quadrant clock resource using PDC:
Fanout Type
Name
-------------------------256
CLK_NET
Net
: QCLK1_c
Driver: QCLK1_pad_CLKINT
Region: quadrant_UL
256
CLK_NET
Net
: QCLK2_c
Driver: QCLK2_pad_CLKINT
Region: quadrant_LL
Step 5
Run Layout.
Global Management in PLL Design
This section describes the legal global network connections to PLLs in the low power flash devices. For
detailed information on using PLLs, refer to "Clock Conditioning Circuits in Low Power Flash Devices and
Mixed Signal FPGAs" section on page 61. Microsemi recommends that you use the dedicated global
pins to directly drive the reference clock input of the associated PLL for reduced propagation delays and
clock distortion. However, low power flash devices offer the flexibility to connect other signals to
reference clock inputs. Each PLL is associated with three global networks (Figure 3-5 on page 36). There
are some limitations, such as when trying to use the global and PLL at the same time:
•
If you use a PLL with only primary output, you can still use the remaining two free global
networks.
•
If you use three globals associated with a PLL location, you cannot use the PLL on that location.
•
If the YB or YC output is used standalone, it will occupy one global, even though this signal does
not go to the global network.
Revision 5
57
Global Resources in Low Power Flash Devices
Using Spines of Occupied Global Networks
When a signal is assigned to a global network, the flash switches are programmed to set the MUX select
lines (explained in the "Clock Aggregation Architecture" section on page 45) to drive the spines of that
network with the global net. However, if the global net is restricted from reaching into the scope of a
spine, the MUX drivers of that spine are available for other high-fanout or critical signals (Figure 3-20).
For example, if you want to limit the CLK1_c signal to the left half of the chip and want to use the right
side of the same global network for CLK2_c, you can add the following PDC commands:
define_region -name region1 -type inclusive 0 0 34 29
assign_net_macros region1 CLK1_c
assign_local_clock –net CLK2_c –type chip B2
Figure 3-20 • Design Example Using Spines of Occupied Global Networks
Conclusion
IGLOO, Fusion, and ProASIC3 devices contain 18 global networks: 6 chip global networks and 12
quadrant global networks. These global networks can be segmented into local low-skew networks called
spines. The spines provide low-skew networks for the high-fanout signals of a design. These allow you
up to 252 different internal/external clocks in an A3PE3000 device. This document describes the
architecture for the global network, plus guidelines and methodologies in assigning signals to globals and
spines.
Related Documents
User’s Guides
IGLOO, ProASIC3, SmartFusion, and Fusion Macro Library Guide
http://www.microsemi.com/soc/documents/pa3_libguide_ug.pdf
58
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
July 2010
v1.4
(December 2008)
v1.3
(October 2008)
v1.2
(June 2008)
Changes
Page
This chapter is no longer published separately with its own part number and
version but is now part of several FPGA fabric user’s guides.
N/A
Notes were added where appropriate to point out that IGLOO nano and
ProASIC3 nano devices do not support differential inputs (SAR 21449).
N/A
The "Global Architecture" section and "VersaNet Global Network Distribution"
section were revised for clarity (SARs 20646, 24779).
31, 33
The "I/O Banks and Global I/Os" section was moved earlier in the document,
renamed to "Chip and Quadrant Global I/Os", and revised for clarity. Figure 3-4 •
Global Connections Details, Figure 3-6 • Global Inputs, Table 3-2 • Chip Global
Pin Name, and Table 3-3 • Quadrant Global Pin Name are new (SARs 20646,
24779).
35
The "Clock Aggregation Architecture" section was revised (SARs 20646, 24779).
41
Figure 3-7 • Chip Global Aggregation was revised (SARs 20646, 24779).
43
The "Global Macro and Placement Selections" section is new (SARs 20646,
24779).
48
The "Global Architecture" section was updated to include 10 k devices, and to
include information about VersaNet global support for IGLOO nano devices.
31
The Table 3-1 • Flash-Based FPGAs was updated to include IGLOO nano and
ProASIC3 nano devices.
32
The "VersaNet Global Network Distribution" section was updated to include 10 k
devices and to note an exception in global lines for nano devices.
33
Figure 3-2 • Simplified VersaNet Global Network (30 k gates and below) is new.
34
The "Spine Architecture" section was updated to clarify support for 10 k and nano
devices.
41
Table 3-4 • Globals/Spines/Rows for IGLOO and ProASIC3 Devices was updated
to include IGLOO nano and ProASIC3 nano devices.
41
The figure in the CLKBUF_LVDS/LVPECL row of Table 3-8 • Clock Macros was
updated to change CLKBIBUF to CLKBUF.
46
A third bullet was added to the beginning of the "Global Architecture" section: In
Fusion devices, the west CCC also contains a PLL core. In the two larger devices
(AFS600 and AFS1500), the west and east CCCs each contain a PLL.
31
The "Global Resource Support in Flash-Based Devices" section was revised to
include new families and make the information more concise.
32
Table 3-4 • Globals/Spines/Rows for IGLOO and ProASIC3 Devices was updated
to include A3PE600/L in the device column.
41
Table note 1 was revised in Table 3-9 • I/O Standards within CLKBUF to include
AFS600 and AFS1500.
47
The following changes were made to the family descriptions in Table 3-1 • FlashBased FPGAs:
32
•
ProASIC3L was updated to include 1.5 V.
•
The number of PLLs for ProASIC3E was changed from five to six.
Revision 5
59
Global Resources in Low Power Flash Devices
Date
Changes
Page
v1.1
(March 2008)
The "Global Architecture" section was updated to include the IGLOO PLUS
family. The bullet was revised to include that the west CCC does not contain a
PLL core in 15 k and 30 k devices. Instances of "A3P030 and AGL030 devices"
were replaced with "15 k and 30 k gate devices."
31
v1.1
(continued)
Table 3-1 • Flash-Based FPGAs and the accompanying text was updated to
include the IGLOO PLUS family. The "IGLOO Terminology" section and
"ProASIC3 Terminology" section are new.
32
The "VersaNet Global Network Distribution" section, "Spine Architecture" section,
the note in Figure 3-1 • Overview of VersaNet Global Network and Device
Architecture, and the note in Figure 3-3 • Simplified VersaNet Global Network
(60 k gates and above) were updated to include mention of 15 k gate devices.
33, 34
Table 3-4 • Globals/Spines/Rows for IGLOO and ProASIC3 Devices was updated
to add the A3P015 device, and to revise the values for clock trees, globals/spines
per tree, and globals/spines per device for the A3P030 and AGL030 devices.
41
Table 3-5 • Globals/Spines/Rows for IGLOO PLUS Devices is new.
42
CLKBUF_LVCMOS12 was added to Table 3-9 • I/O Standards within CLKBUF.
47
The "User’s Guides" section was updated to include the three different I/O
Structures chapters for ProASIC3 and IGLOO device families.
58
Figure 3-3 • Simplified VersaNet Global Network (60 k gates and above) was
updated.
34
The "Naming of Global I/Os" section was updated.
35
The "Using Global Macros in Synplicity" section was updated.
50
The "Global Promotion and Demotion Using PDC" section was updated.
51
The "Designer Flow for Global Assignment" section was updated.
53
The "Simple Design Example" section was updated.
55
Table 3-4 • Globals/Spines/Rows for IGLOO and ProASIC3 Devices was
updated.
41
v1.0
(January 2008)
51900087-0/1.05
(January 2005)
60
R e visio n 5
4 – Clock Conditioning Circuits in Low Power
Flash Devices and Mixed Signal FPGAs
Introduction
This document outlines the following device information: Clock Conditioning Circuit (CCC) features, PLL
core specifications, functional descriptions, software configuration information, detailed usage
information, recommended board-level considerations, and other considerations concerning clock
conditioning circuits and global networks in low power flash devices or mixed signal FPGAs.
Overview of Clock Conditioning Circuitry
In Fusion, IGLOO, and ProASIC3 devices, the CCCs are used to implement frequency division,
frequency multiplication, phase shifting, and delay operations. The CCCs are available in six chip
locations—each of the four chip corners and the middle of the east and west chip sides. For devicespecific variations, refer to the "Device-Specific Layout" section on page 78.
The CCC is composed of the following:
•
PLL core
•
3 phase selectors
•
6 programmable delays and 1 fixed delay that advances/delays phase
•
5 programmable frequency dividers that provide frequency multiplication/division (not shown in
Figure 4-6 on page 71 because they are automatically configured based on the user's required
frequencies)
•
1 dynamic shift register that provides CCC dynamic reconfiguration capability
Figure 4-1 provides a simplified block diagram of the physical implementation of the building blocks in
each of the CCCs.
3 Global I/Os
3 Global I/Os
3 Global I/Os
To
Core
Multiplexer
Tree
CLKA
To Global Network A
From
Core
To
Core
CLKB
From
Core
To
Core
CCC
Function Block
(with or without PLL)
CLKC
To Global Network B
To Global Network C
From
Core
Multiple Signals
Single Signals
Figure 4-1 • Overview of the CCCs Offered in Fusion, IGLOO, and ProASIC3
Revision 5
61
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Each CCC can implement up to three independent global buffers (with or without programmable delay)
or a PLL function (programmable frequency division/multiplication, phase shift, and delays) with up to
three global outputs. Unused global outputs of a PLL can be used to implement independent global
buffers, up to a maximum of three global outputs for a given CCC.
CCC Programming
The CCC block is fully configurable, either via flash configuration bits set in the programming bitstream or
through an asynchronous interface. This asynchronous dedicated shift register interface is dynamically
accessible from inside the low power flash devices to permit parameter changes, such as PLL divide
ratios and delays, during device operation.
To increase the versatility and flexibility of the clock conditioning system, the CCC configuration is
determined either by the user during the design process, with configuration data being stored in flash
memory as part of the device programming procedure, or by writing data into a dedicated shift register
during normal device operation.
This latter mode allows the user to dynamically reconfigure the CCC without the need for core
programming. The shift register is accessed through a simple serial interface. Refer to the "UJTAG
Applications in Microsemi’s Low Power Flash Devices" section on page 297 or the application note Using
Global Resources in Actel Fusion Devices.
Global Resources
Low power flash and mixed signal devices provide three global routing networks (GLA, GLB, and GLC)
for each of the CCC locations. There are potentially many I/O locations; each global I/O location can be
chosen from only one of three possibilities. This is controlled by the multiplexer tree circuitry in each
global network. Once the I/O location is selected, the user has the option to utilize the CCCs before the
signals are connected to the global networks. The CCC in each location (up to six) has the same
structure, so generating the CCC macros is always done with an identical software GUI. The CCCs in the
corner locations drive the quadrant global networks, and the CCCs in the middle of the east and west
chip sides drive the chip global networks. The quadrant global networks span only a quarter of the
device, while the chip global networks span the entire device. For more details on global resources
offered in low power flash devices, refer to the "Global Resources in Low Power Flash Devices" section
on page 31.
A global buffer can be placed in any of the three global locations (CLKA-GLA, CLKB-GLB, or
CLKC-GLC) of a given CCC. A PLL macro uses the CLKA CCC input to drive its reference clock. It uses
the GLA and, optionally, the GLB and GLC global outputs to drive the global networks. A PLL macro can
also drive the YB and YC regular core outputs. The GLB (or GLC) global output cannot be reused if the
YB (or YC) output is used. Refer to the "PLL Macro Signal Descriptions" section on page 68 for more
information.
Each global buffer, as well as the PLL reference clock, can be driven from one of the following:
62
•
3 dedicated single-ended I/Os using a hardwired connection
•
2 dedicated differential I/Os using a hardwired connection (not supported for IGLOO nano or
ProASIC3 nano devices)
•
The FPGA core
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
CCC Support in Microsemi’s Flash Devices
The flash FPGAs listed in Table 4-1 support the CCC feature and the functions described in this
document.
Table 4-1 • Flash-Based FPGAs
Family*
Series
IGLOO
ProASIC3
Fusion
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
IGLOO nano
The industry’s lowest-power, smallest-size solution
ProASIC3
Low power, high-performance 1.5 V FPGAs
ProASIC3E
Higher density ProASIC3 FPGAs with six PLLs and additional I/O standards
ProASIC3 nano
Lowest-cost solution with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V with Flash*Freeze technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Automotive ProASIC3
ProASIC3 FPGAs qualified for automotive applications
Fusion
Mixed signal FPGA integrating ProASIC3 FPGA fabric, programmable
analog block, support for ARM® Cortex™-M1 soft processors, and flash
memory into a monolithic device
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO devices as listed
in Table 4-1. Where the information applies to only one product line or limited devices, these exclusions
will be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 4-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
Revision 5
63
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Global Buffers with No Programmable Delays
Access to the global / quadrant global networks can be configured directly from the global I/O buffer,
bypassing the CCC functional block (as indicated by the dotted lines in Figure 4-1 on page 61). Internal
signals driven by the FPGA core can use the global / quadrant global networks by connecting via the
routed clock input of the multiplexer tree.
There are many specific CLKBUF macros supporting the wide variety of single-ended I/O inputs
(CLKBUF) and differential I/O standards (CLKBUF_LVDS/LVPECL) in the low power flash families. They
are used when connecting global I/Os directly to the global/quadrant networks.
Note: IGLOO nano and ProASIC nano devices do not support differential inputs.
When an internal signal needs to be connected to the global/quadrant network, the CLKINT macro is
used to connect the signal to the routed clock input of the network's MUX tree.
To utilize direct connection from global I/Os or from internal signals to the global/quadrant networks,
CLKBUF, CLKBUF_LVPECL/LVDS, and CLKINT macros are used (Figure 4-2).
•
The CLKBUF and CLKBUF_LVPECL/LVDS1 macros are composite macros that include an I/O
macro driving a global buffer, which uses a hardwired connection.
•
The CLKBUF, CLKBUF_LVPECL/LVDS1 and CLKINT macros are pass-through clock sources
and do not use the PLL or provide any programmable delay functionality.
•
The CLKINT macro provides a global buffer function driven internally by the FPGA core.
The available CLKBUF macros are described in the IGLOO, ProASIC3, SmartFusion, and Fusion
Macro Library Guide.
Clock Source
CLKBUF_LVDS/LVPECL Macro
PADN
Y
PADP
CLKBIBUF Macro
D
Y
CLKINT Macro
A
Clock Conditioning
Y
E
None
Output
GLA, GLB,
or GLC
PAD
CLKBIBUF
CLKBUF Macro
PAD
Y
Note: IGLOO nano and ProASIC nano devices do not support differential inputs.
Figure 4-2 • CCC Options: Global Buffers with No Programmable Delay
Global Buffer with Programmable Delay
Clocks requiring clock adjustments can utilize the programmable delay cores before connecting to the
global / quadrant global networks. A maximum of 18 CCC global buffers can be instantiated in a device—
three per CCC and up to six CCCs per device.
Each CCC functional block contains a programmable delay element for each of the global networks (up
to three), and users can utilize these features by using the corresponding macro (Figure 4-3 on page 65).
1.
64
B-LVDS and M-LVDS are supported with the LVDS macro.
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Clock Source
Clock Conditioning
Output
Input LVDS/LVPECL Macro
PADN
GLA
CLK
Y
GL
PADP
or
GLB
or
DLYGL[4:0]
INBUF* Macro
Y
PAD
GLC
Notes:
1. For INBUF* driving a PLL macro or CLKDLY macro, the I/O will be hard-routed to the CCC; i.e., will be placed by
software to a dedicated Global I/O.
2. IGLOO nano and ProASIC3 nano devices do not support differential inputs.
Figure 4-3 • CCC Options: Global Buffers with Programmable Delay
The CLKDLY macro is a pass-through clock source that does not use the PLL, but provides the ability to
delay the clock input using a programmable delay. The CLKDLY macro takes the selected clock input
and adds a user-defined delay element. This macro generates an output clock phase shift from the input
clock.
The CLKDLY macro can be driven by an INBUF* macro to create a composite macro, where the I/O
macro drives the global buffer (with programmable delay) using a hardwired connection. In this case, the
software will automatically place the dedicated global I/O in the appropriate locations. Many specific
INBUF macros support the wide variety of single-ended and differential I/O standards supported by the
low power flash family. The available INBUF macros are described in the IGLOO, ProASIC3,
SmartFusion, and Fusion Macro Library Guide.
The CLKDLY macro can be driven directly from the FPGA core. The CLKDLY macro can also be driven
from an I/O that is routed through the FPGA regular routing fabric. In this case, users must instantiate a
special macro, PLLINT, to differentiate the clock input driven by the hardwired I/O connection.
The visual CLKDLY configuration in the SmartGen area of the Microsemi Libero System-on-Chip (SoC)
and Designer tools allows the user to select the desired amount of delay and configures the delay
elements appropriately. SmartGen also allows the user to select the input clock source. SmartGen will
automatically instantiate the special macro, PLLINT, when needed.
CLKDLY Macro Signal Descriptions
The CLKDLY macro supports one input and one output. Each signal is described in Table 4-2.
Table 4-2 • Input and Output Description of the CLKDLY Macro
Signal
Name
I/O
CLK
Reference Clock
Input
GL
Global Output
Description
Reference clock input
Output Primary output clock to respective global/quadrant clock networks
Revision 5
65
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
CLKDLY Macro Usage
When a CLKDLY macro is used in a CCC location, the programmable delay element is used to allow the
clock delays to go to the global network. In addition, the user can bypass the PLL in a CCC location
integrated with a PLL, but use the programmable delay that is associated with the global network by
instantiating the CLKDLY macro. The same is true when using programmable delay elements in a CCC
location with no PLLs (the user needs to instantiate the CLKDLY macro). There is no difference between
the programmable delay elements used for the PLL and the CLKDLY macro. The CCC will be configured
to use the programmable delay elements in accordance with the macro instantiated by the user.
As an example, if the PLL is not used in a particular CCC location, the designer is free to specify up to
three CLKDLY macros in the CCC, each of which can have its own input frequency and delay adjustment
options. If the PLL core is used, assuming output to only one global clock network, the other two global
clock networks are free to be used by either connecting directly from the global inputs or connecting from
one or two CLKDLY macros for programmable delay.
The programmable delay elements are shown in the block diagram of the PLL block shown in Figure 4-6
on page 71. Note that any CCC locations with no PLL present contain only the programmable delay
blocks going to the global networks (labeled "Programmable Delay Type 2"). Refer to the "Clock Delay
Adjustment" section on page 86 for a description of the programmable delay types used for the PLL. Also
refer to Table 4-14 on page 94 for Programmable Delay Type 1 step delay values, and Table 4-15 on
page 94 for Programmable Delay Type 2 step delay values. CCC locations with a PLL present can be
configured to utilize only the programmable delay blocks (Programmable Delay Type 2) going to the
global networks A, B, and C.
Global network A can be configured to use only the programmable delay element (bypassing the PLL) if the
PLL is not used in the design. Figure 4-6 on page 71 shows a block diagram of the PLL, where the
programmable delay elements are used for the global networks (Programmable Delay Type 2).
66
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Global Buffers with PLL Function
Clocks requiring frequency synthesis or clock adjustments can utilize the PLL core before connecting to
the global / quadrant global networks. A maximum of 18 CCC global buffers can be instantiated in a
device—three per CCC and up to six CCCs per device. Each PLL core can generate up to three
global/quadrant clocks, while a clock delay element provides one.
The PLL functionality of the clock conditioning block is supported by the PLL macro.
Clock Source
Clock Conditioning
Input LVDS/LVPECL Macro
Output
PLL Macro
GLA
PADN
Y
PADP
INBUF* Macro
PAD
Y
CLKA
EXTFB
POWERDOWN
OADIVRST1
GLA
LOCK
GLB
YB
GLC
YC
OADIVHALF1
2
OADIV[4:0]
OAMUX[2:0] 2
DLYGLA[4:0] 2
OBDIV[4:0] 2
OBMUX[2:0] 2
2
DLYYB[4:0]
2
DLYGLB[4:0]
OCDIV[4:0] 2
2
OCMUX[2:0]
2
DLYYC[4:0]
DLYGLC[4:0]2
2
FINDIV[6:0]
FBDIV[6:0] 2
FBDLY[4:0]2
FBSEL[1:0]2
XDLYSEL2
VCOSEL[2:0]2
or
GLA and (GLB or YB)
or
GLA and (GLC or YC)
or
GLA and (GLB or YB) and
(GLC or YC)
Notes:
1. For Fusion only.
2. Refer to the IGLOO, ProASIC3, SmartFusion, and Fusion Macro Library Guide for more information.
3. For INBUF* driving a PLL macro or CLKDLY macro, the I/O will be hard-routed to the CCC; i.e., will be placed by
software to a dedicated Global I/O.
4. IGLOO nano and ProASIC3 nano devices do not support differential inputs.
Figure 4-4 • CCC Options: Global Buffers with PLL
The PLL macro provides five derived clocks (three independent) from a single reference clock. The PLL
macro also provides power-down input and lock output signals. The additional inputs shown on the
macro are configuration settings, which are configured through the use of SmartGen. For manual setting
of these bits refer to the IGLOO, ProASIC3, SmartFusion, and Fusion Macro Library Guide for details.
Figure 4-6 on page 71 illustrates the various clock output options and delay elements.
Revision 5
67
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
PLL Macro Signal Descriptions
The PLL macro supports two inputs and up to six outputs. Table 4-3 gives a description of each signal.
Table 4-3 • Input and Output Signals of the PLL Block
Signal
Name
I/O
Description
CLKA
Reference Clock
Input
Reference clock input for PLL core; input clock for primary output
clock, GLA
OADIVRST
Reset Signal for the
Output Divider A
Input
For Fusion only. OADIVRST can be used when you bypass the PLL
core (i.e., OAMUX = 001). The purpose of the OADIVRST signals is
to reset the output of the final clock divider to synchronize it with the
input to that divider when the PLL is bypassed. The signal is active
on a low to high transition. The signal must be low for at least one
divider input. If PLL core is used, this signal is "don't care" and the
internal circuitry will generate the reset signal for the
synchronization purpose.
OADIVHALF
Output A Division by
Half
Input
For Fusion only. Active high. Division by half feature. This feature
can only be used when users bypass the PLL core (i.e., OAMUX =
001) and the RC Oscillator (RCOSC) drives the CLKA input. This
can be used to divide the 100 MHz RC oscillator by a factor of 1.5,
2.5, 3.5, 4.5 ... 14.5). Refer to Table 4-18 on page 95 for more
information.
EXTFB
External Feedback
Input
Allows an external signal to be compared to a reference clock in the
PLL core's phase detector.
Input
Active low input that selects power-down mode and disables the
PLL. With the POWERDOWN signal asserted, the PLL core sends
0 V signals on all of the outputs.
POWERDOWN Power Down
GLA
Primary Output
Output Primary output clock to respective global/quadrant clock networks
GLB
Secondary 1 Output
Output Secondary 1 output clock to respective global/quadrant clock
networks
YB
Core 1 Output
Output Core 1 output clock to local routing network
GLC
Secondary 2 Output
Output Secondary 2 output clock to respective global/quadrant clock
networks
YC
Core 2 Output
Output Core 2 output clock to local routing network
LOCK
PLL Lock Indicator
Output Active high signal indicating that steady-state lock has been
achieved between CLKA and the PLL feedback signal
Input Clock
The inputs to the input reference clock (CLKA) of the PLL can come from global input pins, regular I/O
pins, or internally from the core. For Fusion families, the input reference clock can also be from the
embedded RC oscillator or crystal oscillator.
Global Output Clocks
GLA (Primary), GLB (Secondary 1), and GLC (Secondary 2) are the outputs of Global Multiplexer 1,
Global Multiplexer 2, and Global Multiplexer 3, respectively. These signals (GLx) can be used to drive the
high-speed global and quadrant networks of the low power flash devices.
A global multiplexer block consists of the input routing for selecting the input signal for the GLx clock and
the output multiplexer, as well as delay elements associated with that clock.
Core Output Clocks
YB and YC are known as Core Outputs and can be used to drive internal logic without using global
network resources. This is especially helpful when global network resources must be conserved and
utilized for other timing-critical paths.
68
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
YB and YC are identical to GLB and GLC, respectively, with the exception of a higher selectable final
output delay. The SmartGen PLL Wizard will configure these outputs according to user specifications and
can enable these signals with or without the enabling of Global Output Clocks.
The above signals can be enabled in the following output groupings in both internal and external
feedback configurations of the static PLL:
•
One output – GLA only
•
Two outputs – GLA + (GLB and/or YB)
•
Three outputs – GLA + (GLB and/or YB) + (GLC and/or YC)
PLL Macro Block Diagram
As illustrated, the PLL supports three distinct output frequencies from a given input clock. Two of these
(GLB and GLC) can be routed to the B and C global network access, respectively, and/or routed to the
device core (YB and YC).
There are five delay elements to support phase control on all five outputs (GLA, GLB, GLC, YB, and YC).
There are delay elements in the feedback loop that can be used to advance the clock relative to the
reference clock.
The PLL macro reference clock can be driven in the following ways:
1. By an INBUF* macro to create a composite macro, where the I/O macro drives the global buffer
(with programmable delay) using a hardwired connection. In this case, the I/O must be placed in
one of the dedicated global I/O locations.
2. Directly from the FPGA core.
3. From an I/O that is routed through the FPGA regular routing fabric. In this case, users must
instantiate a special macro, PLLINT, to differentiate from the hardwired I/O connection described
earlier.
During power-up, the PLL outputs will toggle around the maximum frequency of the voltage-controlled
oscillator (VCO) gear selected. Toggle frequencies can range from 40 MHz to 250 MHz. This will
continue as long as the clock input (CLKA) is constant (HIGH or LOW). This can be prevented by LOW
assertion of the POWERDOWN signal.
The visual PLL configuration in SmartGen, a component of the Libero SoC and Designer tools, will derive
the necessary internal divider ratios based on the input frequency and desired output frequencies
selected by the user.
Revision 5
69
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Implementing EXTFB in ProASIC3/E Devices
When the external feedback (EXTFB) signal of the PLL in the ProASIC3/E devices is implemented, the
phase detector of the PLL core receives the reference clock (CLKA) and EXTFB as inputs. EXTFB must
be sourced as an INBUF macro and located at the global/chip clock location associated with the target
PLL by Designer software. EXTFB cannot be sourced from the FPGA fabric.
The following example shows CLKA and EXTFB signals assigned to two global I/Os in the same global
area of ProASIC3E device.
To Core
GxA0
The reference clock,
CLKA, can be assigned
on GxA0 or GxA1.
GxA1
–
+
Source for CCC
(CLKA or CLKB or CLKC)
GxA2
Routed Clok
(from FPGA core)
–
+
To Core
GxB0
External Feedback
(EXTFB) signal is
assigned on GxB1
by Designer automatically.
GxB1
–
+
Source for CCC
(CLKA or CLKB or CLKC)
GxB2
Routed Clok
(from FPGA core)
–
+
x represents global location; can be A, B, C, D, E, or F
Figure 4-5 • CLKA and EXTFB Assigned to Global I/Os
70
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
SmartGen also allows the user to select the various delays and phase shift values necessary to adjust
the phases between the reference clock (CLKA) and the derived clocks (GLA, GLB, GLC, YB, and YC).
SmartGen allows the user to select the input clock source. SmartGen automatically instantiates the
special macro, PLLINT, when needed.
CLKA
PLL Core
Programmable Delay
Four-Phase Output
Phase
Select
Programmable
Delay Type 2
GLA
Programmable
Delay Type 2
GLB
Programmable
Delay Type 1
YB
Programmable
Delay Type 2
GLC
Programmable
Delay Type 1
YC
Programmable
Delay Type 1
EXTFB
Phase
Select
Phase
Select
Note: Clock divider and clock multiplier blocks are not shown in this figure or in SmartGen. They are automatically
configured based on the user's required frequencies.
Figure 4-6 • CCC with PLL Block
Global Input Selections
Low power flash devices provide the flexibility of choosing one of the three global input pad locations
available to connect to a CCC functional block or to a global / quadrant global network. Figure 4-7 on
page 72 and Figure 4-8 on page 72 show the detailed architecture of each global input structure for 30 k
gate devices and below, as well as 60 k gate devices and above, respectively. For 60 k gate devices and
above (Figure 4-7 on page 72), if the single-ended I/O standard is chosen, there is flexibility to choose
one of the global input pads (the first, second, and fourth input). Once chosen, the other I/O locations are
used as regular I/Os. If the differential I/O standard is chosen (not applicable for IGLOO nano and
ProASIC3 nano devices), the first and second inputs are considered as paired, and the third input is
paired with a regular I/O.
The user then has the choice of selecting one of the two sets to be used as the clock input source to the
CCC functional block. There is also the option to allow an internal clock signal to feed the global network
or the CCC functional block. A multiplexer tree selects the appropriate global input for routing to the
desired location. Note that the global I/O pads do not need to feed the global network; they can also be
used as regular I/O pads.
Revision 5
71
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
To Core
Dedicated I/O Pad
Sample Pin Names
Drives the global
network directly
(GLA or GLC)
GEC0/IO37RSB1
Routed Clock
(from FPGA core)
Figure 4-7 • Clock Input Sources (30 k gates devices and below)
Each shaded box represents an
INBUF or INBUF_LVDS/LVPECL
macro, as appropriate.
To Core
Sample Pin Names
GAA0/IO0NDB0V0
1
GAA1/IO00PDB0V0
1
+
Source for CCC
(CLKA or CLKB or CLKC)
GAA2/IO13PDB7V1
1
+
Routed Clock
2
(from FPGA core)
GAA[0:2]: GA represents global in the northwest corner
of the device. A[0:2]: designates specific A clock source.
Notes:
1. Represents the global input pins. Globals have direct access to the clock conditioning block and are
not routed via the FPGA fabric. Refer to the "User I/O Naming Conventions in I/O Structures" chapter
of the appropriate device user’s guide.
2. Instantiate the routed clock source input as follows:
a) Connect the output of a logic element to the clock input of a PLL, CLKDLY, or CLKINT macro.
b) Do not place a clock source I/O (INBUF or INBUF_LVPECL/LVDS/B-LVDS/M-LVDS/DDR) in
a relevant global pin location.
3. IGLOO nano and ProASIC3 nano devices do not support differential inputs.
Figure 4-8 • Clock Input Sources Including CLKBUF, CLKBUF_LVDS/LVPECL, and CLKINT (60 k
gates devices and above)
72
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Each global buffer, as well as the PLL reference clock, can be driven from one of the following:
•
3 dedicated single-ended I/Os using a hardwired connection
•
2 dedicated differential I/Os using a hardwired connection (not applicable for IGLOO nano and
ProASIC3 nano devices)
•
The FPGA core
Since the architecture of the devices varies as size increases, the following list details I/O types
supported for globals:
IGLOO and ProASIC3
•
LVDS-based clock sources are available only on 250 k gate devices and above (IGLOO nano and
ProASIC3 nano devices do not support differential inputs).
•
60 k and 125 k gate devices support single-ended clock sources only.
•
15 k and 30 k gate devices support these inputs for CCC only and do not contain a PLL.
•
nano devices:
–
10 k, 15 k, and 20 k devices do not contain PLLs in the CCCs, and support only CLKBUF and
CLKINT.
–
60 k, 125 k, and 250 k devices support one PLL in the middle left CCC position. In the
absence of the PLL, this CCC can be used by CLKBUF, CLKINT, and CLKDLY macros. The
corner CCCs support CLKBUF, CLKINT, and CLKDLY.
Fusion
•
AFS600 and AFS1500: All single-ended, differential, and voltage-referenced I/O standards (Pro
I/O).
•
AFS090 and AFS250: All single-ended and differential I/O standards.
Clock Sources for PLL and CLKDLY Macros
The input reference clock (CLKA for a PLL macro, CLK for a CLKDLY macro) can be accessed from
different sources via the associated clock multiplexer tree. Each CCC has the option of choosing the
source of the input clock from one of the following:
•
Hardwired I/O
•
External I/O
•
Core Logic
•
RC Oscillator (Fusion only)
•
Crystal Oscillator (Fusion only)
The SmartGen macro builder tool allows users to easily create the PLL and CLKDLY macros with the
desired settings. Microsemi strongly recommends using SmartGen to generate the CCC macros.
Hardwired I/O Clock Source
Hardwired I/O refers to global input pins that are hardwired to the multiplexer tree, which directly
accesses the CCC global buffers. These global input pins have designated pin locations and are
indicated with the I/O naming convention Gmn (m refers to any one of the positions where the PLL core
is available, and n refers to any one of the three global input MUXes and the pin number of the
associated global location, m). Choosing this option provides the benefit of directly connecting to the
CCC reference clock input, which provides less delay. See Figure 4-9 on page 74 for an example
illustration of the connections, shown in red. If a CLKDLY macro is initiated to utilize the programmable
delay element of the CCC, the clock input can be placed at one of nine dedicated global input pin
locations. In other words, if Hardwired I/O is chosen as the input source, the user can decide to place the
input pin in one of the GmA0, GmA1, GmA2, GmB0, GmB1, GmB2, GmC0, GmC1, or GmC2 locations of
the low power flash devices. When a PLL macro is used to utilize the PLL core in a CCC location, the
clock input of the PLL can only be connected to one of three GmA* global pin locations: GmA0, GmA1, or
GmA2.
Revision 5
73
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
To Core
Gmn0
Gmn1
Multiplexer
Tree
_
+
PLL or CLKDLY
Macro
To Global (or local)
Routing Network
CLKA
IOuxwByVz
Gmn2
_
+
Routed Clock
(from FPGA core)
Gmn* = Global Input Pin
IOuxwByVz = Regular I/O Pin
PLLINT
Note: Fusion CCCs have additional source selections (RCOSC, XTAL).
Figure 4-9 • Illustration of Hardwired I/O (global input pins) Usage for IGLOO and ProASIC3 devices 60 k Gates
and Larger
Dedicated I/O Pad
To Core
Sample Pin Names
GEC0/IO37RSB1
Directly Drives Global Network
(GLA or GLC)
Routed Clock
(from the FPGA core)
Figure 4-10 • Illustration of Hardwired I/O (global input pins) Usage for IGLOO and ProASIC3 devices 30 k
Gates and Smaller
74
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
External I/O Clock Source
External I/O refers to regular I/O pins. The clock source is instantiated with one of the various INBUF
options and accesses the CCCs via internal routing. The user has the option of assigning this input to
any of the I/Os labeled with the I/O convention IOuxwByVz. Refer to the "User I/O Naming Conventions
in I/O Structures" chapter of the appropriate device user’s guide, and for Fusion, refer to the Fusion
Family of Mixed Signal FPGAs datasheet for more information. Figure 4-11 gives a brief explanation of
external I/O usage. Choosing this option provides the freedom of selecting any user I/O location but
introduces additional delay because the signal connects to the routed clock input through internal routing
before connecting to the CCC reference clock input.
For the External I/O option, the routed signal would be instantiated with a PLLINT macro before
connecting to the CCC reference clock input. This instantiation is conveniently done automatically by
SmartGen when this option is selected. Microsemi recommends using the SmartGen tool to generate the
CCC macro. The instantiation of the PLLINT macro results in the use of the routed clock input of the I/O
to connect to the PLL clock input. If not using SmartGen, manually instantiate a PLLINT macro before the
PLL reference clock to indicate that the regular I/O driving the PLL reference clock should be used (see
Figure 4-11 for an example illustration of the connections, shown in red).
In the above two options, the clock source must be instantiated with one of the various INBUF macros.
The reference clock pins of the CCC functional block core macros must be driven by regular input
macros (INBUFs), not clock input macros.
To Core
Gmn*
Gmn*
Multiplexer
Tree
_
+
PLL or CLKDLY
Macro
CLKA
To Global (or Local)
Routing Network
IOuxwByVz*
Gmn*
_
+
PLLINT
Routed Clock
(from FPGA Core)
Gmn* = Global Input Pin
IOuxwByVz = Regular I/O Pin
IOuxwByVz*
Figure 4-11 • Illustration of External I/O Usage
For Fusion devices, the input reference clock can also be from the embedded RC oscillator and crystal
oscillator. In this case, the CCC configuration is the same as the hardwired I/O clock source, and users
are required to instantiate the RC oscillator or crystal oscillator macro and connect its output to the input
reference clock of the CCC block.
Revision 5
75
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Core Logic Clock Source
Core logic refers to internal routed nets. Internal routed signals access the CCC via the FPGA Core
Fabric. Similar to the External I/O option, whenever the clock source comes internally from the core itself,
the routed signal is instantiated with a PLLINT macro before connecting to the CCC clock input (see
Figure 4-12 for an example illustration of the connections, shown in red).
To Core
Gmn*
Gmn*
Multiplexer
Tree
_
+
PLL or CLKDLY
Macro
CLKA
To Global (or Local)
Routing Network
IOuxwByVz*
Gmn*
_
+
Routed Clock
(from FPGA Core)
PLLINT
Gmn* = Global Input Pin
IOuxwByVz = Regular I/O Pin
From Internal
Signals
Figure 4-12 • Illustration of Core Logic Usage
For Fusion devices, the input reference clock can also be from the embedded RC oscillator and crystal
oscillator. In this case, the CCC configuration is the same as the hardwired I/O clock source, and users
are required to instantiate the RC oscillator or crystal oscillator macro and connect its output to the input
reference clock of the CCC block.
76
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Available I/O Standards
Table 4-4 • Available I/O Standards within CLKBUF and CLKBUF_LVDS/LVPECL Macros
CLKBUF_LVCMOS5
CLKBUF_LVCMOS33 1
CLKBUF_LVCMOS25 2
CLKBUF_LVCMOS18
CLKBUF_LVCMOS15
CLKBUF_PCI
CLKBUF_PCIX 3
CLKBUF_GTL25 2,3
CLKBUF_GTL33 2,3
CLKBUF_GTLP25 2,3
CLKBUF_GTLP33 2,3
CLKBUF_HSTL_I 2,3
CLKBUF_HSTL_II 2,3
CLKBUF_SSTL3_I 2,3
CLKBUF_SSTL3_II 2,3
CLKBUF_SSTL2_I 2,3
CLKBUF_SSTL2_II 2,3
CLKBUF_LVDS 4,5
CLKBUF_LVPECL5
Notes:
1. By default, the CLKBUF macro uses 3.3 V LVTTL I/O technology. For more details, refer to the
IGLOO, ProASIC3, SmartFusion, and Fusion Macro Library Guide.
2. I/O standards only supported in ProASIC3E and IGLOOe families.
3. I/O standards only supported in the following Fusion devices: AFS600 and AFS1500.
4. B-LVDS and M-LVDS standards are supported by CLKBUF_LVDS.
5. Not supported for IGLOO nano and ProASIC3 nano devices.
Global Synthesis Constraints
The Synplify® synthesis tool, by default, allows six clocks in a design for Fusion, IGLOO, and ProASIC3.
When more than six clocks are needed in the design, a user synthesis constraint attribute,
syn_global_buffers, can be used to control the maximum number of clocks (up to 18) that can be inferred
by the synthesis engine.
High-fanout nets will be inferred with clock buffers and/or internal clock buffers. If the design consists of
CCC global buffers, they are included in the count of clocks in the design.
The subsections below discuss the clock input source (global buffers with no programmable delays) and
the clock conditioning functional block (global buffers with programmable delays and/or PLL function) in
detail.
Revision 5
77
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Device-Specific Layout
Two kinds of CCCs are offered in low power flash devices: CCCs with integrated PLLs, and CCCs
without integrated PLLs (simplified CCCs). Table 4-5 lists the number of CCCs in various devices.
Table 4-5 • Number of CCCs by Device Size and Package
Device
ProASIC3
IGLOO
Package
CCCs with
Integrated
PLLs
A3PN010
AGLN010
All
0
2
A3PN015
AGLN015
All
0
2
A3PN020
AGLN020
All
0
2
AGLN060
CS81
0
6
AGLN060
All other
packages
1
5
AGLN125
CS81
0
6
A3PN125
AGLN125
All other
packages
1
5
AGLN250
CS81
0
6
A3PN250
AGLN250
All other
packages
1
5
AGL015
All
0
2
A3PN060
A3P015
A3P030
CCCs without
Integrated PLLs
(simplified CCC)
AGL030/AGLP030
All
0
2
AGL060/AGLP060
CS121/CS201
0
6
A3P060
AGL060/AGLP060
All other
packages
1
5
A3P125
AGL125/AGLP125
All
1
5
A3P250/L
AGL250
All
1
5
A3P400
AGL400
All
1
5
A3P600/L
AGL600
All
1
5
A3P1000/L
AGL1000
All
1
5
A3PE600
AGLE600
PQ208
2
4
All other
packages
6
0
A3PE600/L
A3PE1500
PQ208
2
4
A3PE1500
All other
packages
6
0
PQ208
2
4
All other
packages
6
0
All
1
5
A3PE3000/L
A3PE3000/L
AGLE3000
Fusion Devices
AFS090
AFS250, M1AFS250
All
1
5
AFS600, M7AFS600, M1AFS600
All
2
4
AFS1500, M1AFS1500
All
2
4
Note: nano 10 k, 15 k, and 20 k offer 6 global MUXes instead of CCCs.
78
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
This section outlines the following device information: CCC features, PLL core specifications, functional
descriptions, software configuration information, detailed usage information, recommended board-level
considerations, and other considerations concerning global networks in low power flash devices.
Clock Conditioning Circuits with Integrated PLLs
Each of the CCCs with integrated PLLs includes the following:
•
1 PLL core, which consists of a phase detector, a low-pass filter, and a four-phase voltagecontrolled oscillator
•
3 global multiplexer blocks that steer signals from the global pads and the PLL core onto the
global networks
•
6 programmable delays and 1 fixed delay for time advance/delay adjustments
•
5 programmable frequency divider blocks to provide frequency synthesis (automatically
configured by the SmartGen macro builder tool)
Clock Conditioning Circuits without Integrated PLLs
There are two types of simplified CCCs without integrated PLLs in low power flash devices.
1. The simplified CCC with programmable delays, which is composed of the following:
–
3 global multiplexer blocks that steer signals from the global pads and the programmable
delay elements onto the global networks
–
3 programmable delay elements to provide time delay adjustments
2. The simplified CCC (referred to as CCC-GL) without programmable delay elements, which is
composed of the following:
–
A global multiplexer block that steer signals from the global pads onto the global networks
Revision 5
79
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
CCC Locations
CCCs located in the middle of the east and west sides of the device access the three VersaNet global
networks on each side (six total networks), while the four CCCs located in the four corners access three
quadrant global networks (twelve total networks). See Figure 4-13.
Northwest Quadrant Global Networks
Quadrant Global Spine
CCC Location A
3
3
3
3
Chip-Wide (main)
Global
Networks
6
6
CCC Location B
3
6
6
CCC Location C
CCC Location F
Global Spine
3
CCC Location E
6
6
3
6
3
6
3
3
Southeast Quadrant Global Networks
CCC Location D
Figure 4-13 • Global Network Architecture for 60 k Gate Devices and Above
The following explains the locations of the CCCs in IGLOO and ProASIC3 devices:
In Figure 4-15 on page 82 through Figure 4-16 on page 82, CCCs with integrated PLLs are indicated in
red, and simplified CCCs are indicated in yellow. There is a letter associated with each location of the
CCC, in clockwise order. The upper left corner CCC is named "A," the upper right is named "B," and so
on. These names finish up at the middle left with letter "F."
80
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
IGLOO and ProASIC3 CCC Locations
In all IGLOO and ProASIC3 devices (except 10 k through 30 k gate devices, which do not contain PLLs),
six CCCs are located in the same positions as the IGLOOe and ProASIC3E CCCs. Only one of the
CCCs has an integrated PLL and is located in the middle of the west (middle left) side of the device. The
other five CCCs are simplified CCCs and are located in the four corners and the middle of the east side
of the device (Figure 4-14).
A
B
Bank 0
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
Bank 1
Bank 3
CCC
I/Os
C
F
Bank 1
Bank 3
VersaTile
ISP AES
Decryption
User Nonvolatile
FlashROM (FROM)
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
Charge Pumps
D
E
Bank 2
= CCC with integrated PLL
= Simplified CCC with programmable delay elements (no PLL)
Figure 4-14 • CCC Locations in IGLOO and ProASIC3 Family Devices
(except 10 k through 30 k gate devices)
Note: The number and architecture of the banks are different for some devices.
10 k through 30 k gate devices do not support PLL features. In these devices, there are two CCC-GLs at
the lower corners (one at the lower right, and one at the lower left). These CCC-GLs do not have
programmable delays.
Revision 5
81
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
IGLOOe and ProASIC3E CCC Locations
IGLOOe and ProASIC3E devices have six CCCs—one in each of the four corners and one each in the
middle of the east and west sides of the device (Figure 4-15).
All six CCCs are integrated with PLLs, except in PQFP-208 package devices. PQFP-208 package
devices also have six CCCs, of which two include PLLs and four are simplified CCCs. The CCCs with
PLLs are implemented in the middle of the east and west sides of the device (middle right and middle
left). The simplified CCCs without PLLs are located in the four corners of the device (Figure 4-16).
CCC
A
B
RAM Block
4,608-Bit Dual-Port SRAM
or FIFO Block
Pro I/Os
C
F
VersaTile
ISP AES
Decryption
User Nonvolatile
FlashRom
Flash*Freeze
Technology
RAM Block
4,608-Bit Dual-Port SRAM
or FIFO Block
Charge
Pumps
E
D
= CCC with integrated PLL
Figure 4-15 • CCC Locations in IGLOOe and ProASIC3E Family Devices (except PQFP-208
package)
Bank 0
A
B
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
Bank 1
Bank 3
CCC
I/Os
C
F
Bank 1
Bank 3
VersaTile
ISP AES
Decryption*
User Nonvolatile
FlashRom
Flash*Freeze
Technology
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
Charge
Pumps
D
E
Bank 2
= CCC with integrated PLL
= Simplified CCC with programmable delay elements (no PLL)
Figure 4-16 • CCC Locations in ProASIC3E Family Devices (PQFP-208 package)
82
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Fusion CCC Locations
Fusion devices have six CCCs: one in each of the four corners and one each in the middle of the east
and west sides of the device (Figure 4-17 and Figure 4-18). The device can have one integrated PLL in
the middle of the west side of the device or two integrated PLLs in the middle of the east and west sides
of the device (middle right and middle left).
A
B
Bank 0
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
Bank 1
Bank 3
CCC
I/Os
C
F
Bank 1
Bank 3
VersaTile
ISP AES
Decryption
User Nonvolatile
FlashROM (FROM)
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
Charge Pumps
D
E
Bank 2
= CCC with integrated PLL
= Simplified CCC with programmable delay elements (no PLL)
Figure 4-17 • CCC Locations in Fusion Family Devices (AFS090, AFS250, M1AFS250)
Bank 0
A
B
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
Bank 1
Bank 3
CCC
I/Os
C
F
Bank 1
Bank 3
VersaTile
ISP AES
Decryption*
User Nonvolatile
FlashRom
Flash*Freeze
Technology
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
Charge
Pumps
D
E
Bank 2
= CCC with integrated PLL
= Simplified CCC with programmable delay elements (no PLL)
Figure 4-18 • CCC Locations in Fusion Family Devices (except AFS090, AFS250, M1AFS250)
Revision 5
83
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
PLL Core Specifications
PLL core specifications can be found in the DC and Switching Characteristics chapter of the appropriate
family datasheet.
Loop Bandwidth
Common design practice for systems with a low-noise input clock is to have PLLs with small loop
bandwidths to reduce the effects of noise sources at the output. Table 4-6 shows the PLL loop
bandwidth, providing a measure of the PLL's ability to track the input clock and jitter.
Table 4-6 • –3 dB Frequency of the PLL
Typical
Maximum
Minimum
(Ta = +125°C, VCCA = 1.4 V) (Ta = +25°C, VCCA = 1.5 V) (Ta = –55°C, VCCA = 1.6 V)
15 kHz
25 kHz
45 kHz
–3 dB
Frequency
PLL Core Operating Principles
This section briefly describes the basic principles of PLL operation. The PLL core is composed of a
phase detector (PD), a low-pass filter (LPF), and a four-phase voltage-controlled oscillator (VCO).
Figure 4-19 illustrates a basic single-phase PLL core with a divider and delay in the feedback path.
Frequency
Reference
Input FIN
Phase
Detector
Voltage
Controlled
Oscillator
Low-Pass
Filter
Divide by M
Counter
Frequency
Output
M × FIN
Delay
Figure 4-19 • Simplified PLL Core with Feedback Divider and Delay
The PLL is an electronic servo loop that phase-aligns the PD feedback signal with the reference input. To
achieve this, the PLL dynamically adjusts the VCO output signal according to the average phase
difference between the input and feedback signals.
The first element is the PD, which produces a voltage proportional to the phase difference between its
inputs. A simple example of a digital phase detector is an Exclusive-OR gate. The second element, the
LPF, extracts the average voltage from the phase detector and applies it to the VCO. This applied voltage
alters the resonant frequency of the VCO, thus adjusting its output frequency.
Consider Figure 4-19 with the feedback path bypassing the divider and delay elements. If the LPF
steadily applies a voltage to the VCO such that the output frequency is identical to the input frequency,
this steady-state condition is known as lock. Note that the input and output phases are also identical. The
PLL core sets a LOCK output signal HIGH to indicate this condition.
Should the input frequency increase slightly, the PD detects the frequency/phase difference between its
reference and feedback input signals. Since the PD output is proportional to the phase difference, the
change causes the output from the LPF to increase. This voltage change increases the resonant
frequency of the VCO and increases the feedback frequency as a result. The PLL dynamically adjusts in
this manner until the PD senses two phase-identical signals and steady-state lock is achieved. The
opposite (decreasing PD output signal) occurs when the input frequency decreases.
Now suppose the feedback divider is inserted in the feedback path. As the division factor M (shown in
Figure 4-20 on page 85) is increased, the average phase difference increases. The average phase
84
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
difference will cause the VCO to increase its frequency until the output signal is phase-identical to the
input after undergoing division. In other words, lock in both frequency and phase is achieved when the
output frequency is M times the input. Thus, clock division in the feedback path results in multiplication at
the output.
A similar argument can be made when the delay element is inserted into the feedback path. To achieve
steady-state lock, the VCO output signal will be delayed by the input period less the feedback delay. For
periodic signals, this is equivalent to time-advancing the output clock by the feedback delay.
Another key parameter of a PLL system is the acquisition time. Acquisition time is the amount of time it
takes for the PLL to achieve lock (i.e., phase-align the feedback signal with the input reference clock).
For example, suppose there is no voltage applied to the VCO, allowing it to operate at its free-running
frequency. Should an input reference clock suddenly appear, a lock would be established within the
maximum acquisition time.
Functional Description
This section provides detailed descriptions of PLL block functionality: clock dividers and multipliers, clock
delay adjustment, phase adjustment, and dynamic PLL configuration.
Clock Dividers and Multipliers
The PLL block contains five programmable dividers. Figure 4-20 shows a simplified PLL block.
n
CLKA
PLL Core
m
Fixed
Delay
System
Delay
90°
180°
270°
0°
u
D1
Feedback
Delay
Output
Delay
D2
Output
Delay
D2
GLB
Secondary 1
v
D1
Output
Delay
Output
Delay
D2
D1 = Programmable Delay Type 1
D2 = Programmable Delay Type 2
GLA
Primary
w
YB
GLC
Secondary 2
D1
Output
Delay
YC
Figure 4-20 • PLL Block Diagram
Revision 5
85
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Dividers n and m (the input divider and feedback divider, respectively) provide integer frequency division
factors from 1 to 128. The output dividers u, v, and w provide integer division factors from 1 to 32.
Frequency scaling of the reference clock CLKA is performed according to the following formulas:
fGLA = fCLKA × m / (n × u) – GLA Primary PLL Output Clock
EQ 4-1
fGLB = fYB = fCLKA × m / (n × v) – GLB Secondary 1 PLL Output Clock(s)
EQ 4-2
fGLC = fYC = fCLKA × m / (n × w) – GLC Secondary 2 PLL Output Clock(s)
EQ 4-3
SmartGen provides a user-friendly method of generating the configured PLL netlist, which includes
automatically setting the division factors to achieve the closest possible match to the requested
frequencies. Since the five output clocks share the n and m dividers, the achievable output frequencies
are interdependent and related according to the following formula:
fGLA = fGLB × (v / u) = fGLC × (w / u)
EQ 4-4
Clock Delay Adjustment
There are a total of seven configurable delay elements implemented in the PLL architecture.
Two of the delays are located in the feedback path, entitled System Delay and Feedback Delay. System
Delay provides a fixed delay of 2 ns (typical), and Feedback Delay provides selectable delay values from
0.6 ns to 5.56 ns in 160 ps increments (typical). For PLLs, delays in the feedback path will effectively
advance the output signal from the PLL core with respect to the reference clock. Thus, the System and
Feedback delays generate negative delay on the output clock. Additionally, each of these delays can be
independently bypassed if necessary.
The remaining five delays perform traditional time delay and are located at each of the outputs of the
PLL. Besides the fixed global driver delay of 0.755 ns for each of the global networks, the global
multiplexer outputs (GLA, GLB, and GLC) each feature an additional selectable delay value, as given in
Table 4-7.
Table 4-7 • Delay Values in Libero SoC Software per Device Family
Device
Typical
Starting Values
Increments
Ending Value
ProASIC3
200 ps
0 to 735 ps
200 ps
6.735 ns
IGLOO/ProASIC3L 1.5 V
360 ps
0 to 1.610 ns
360 ps
12.410 ns
IGLOO/ProASIC3L 1.2 V
580 ps
0 to 2.880 ns
580 ps
20.280 ns
The additional YB and YC signals have access to a selectable delay from 0.6 ns to 5.56 ns in 160 ps
increments (typical). This is the same delay value as the CLKDLY macro. It is similar to CLKDLY, which
bypasses the PLL core just to take advantage of the phase adjustment option with the delay value.
The following parameters must be taken into consideration to achieve minimum delay at the outputs
(GLA, GLB, GLC, YB, and YC) relative to the reference clock: routing delays from the PLL core to CCC
outputs, core outputs and global network output delays, and the feedback path delay. The feedback path
delay acts as a time advance of the input clock and will offset any delays introduced beyond the PLL core
output. The routing delays are determined from back-annotated simulation and are configurationdependent.
86
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Phase Adjustment
The four phases available (0, 90, 180, 270) are phases with respect to VCO (PLL output). The
VCO is divided to achieve the user's CCC required output frequency (GLA, YB/GLB, YC/GLC). The
division happens after the selection of the VCO phase. The effective phase shift is actually the VCO
phase shift divided by the output divider. This is why the visual CCC shows both the actual achievable
phase and more importantly the actual delay that is equivalent to the phase shift that can be
achieved.
Dynamic PLL Configuration
The CCCs can be configured both statically and dynamically.
In addition to the ports available in the Static CCC, the Dynamic CCC has the dynamic shift register
signals that enable dynamic reconfiguration of the CCC. With the Dynamic CCC, the ports CLKB and
CLKC are also exposed. All three clocks (CLKA, CLKB, and CLKC) can be configured independently.
The CCC block is fully configurable. The following two sources can act as the CCC configuration bits.
Flash Configuration Bits
The flash configuration bits are the configuration bits associated with programmed flash switches. These
bits are used when the CCC is in static configuration mode. Once the device is programmed, these bits
cannot be modified. They provide the default operating state of the CCC.
Dynamic Shift Register Outputs
This source does not require core reprogramming and allows core-driven dynamic CCC reconfiguration.
When the dynamic register drives the configuration bits, the user-defined core circuit takes full control
over SDIN, SDOUT, SCLK, SSHIFT, and SUPDATE. The configuration bits can consequently be
dynamically changed through shift and update operations in the serial register interface. Access to the
logic core is accomplished via the dynamic bits in the specific tiles assigned to the PLLs.
Figure 4-21 illustrates a simplified block diagram of the MUX architecture in the CCCs.
SDIN
SDOUT
SCLK
SSHIFT
SUPDATE
*
Dynamic Shift
Register
RESET_ENABLE
Flash
Programming
Configuration
Bits
*
MODE
Configuration Bits
Note: *For Fusion, bit is also needed.
Figure 4-21 • The CCC Configuration MUX Architecture
The selection between the flash configuration bits and the bits from the configuration register is made
using the MODE signal shown in Figure 4-21. If the MODE signal is logic HIGH, the dynamic shift
register configuration bits are selected. There are 81 control bits to configure the different functions of the
CCC.
Revision 5
87
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Each group of control bits is assigned a specific location in the configuration shift register. For a list of the
81 configuration bits (C[80:0]) in the CCC and a description of each, refer to "PLL Configuration Bits
Description" on page 90. The configuration register can be serially loaded with the new configuration
data and programmed into the CCC using the following ports:
•
SDIN: The configuration bits are serially loaded into a shift register through this port. The LSB of
the configuration data bits should be loaded first.
•
SDOUT: The shift register contents can be shifted out (LSB first) through this port using the shift
operation.
•
SCLK: This port should be driven by the shift clock.
•
SSHIFT: The active-high shift enable signal should drive this port. The configuration data will be
shifted into the shift register if this signal is HIGH. Once SSHIFT goes LOW, the data shifting will
be halted.
•
SUPDATE: The SUPDATE signal is used to configure the CCC with the new configuration bits
when shifting is complete.
To access the configuration ports of the shift register (SDIN, SDOUT, SSHIFT, etc.), the user should
instantiate the CCC macro in his design with appropriate ports. Microsemi recommends that users
choose SmartGen to generate the CCC macros with the required ports for dynamic reconfiguration.
Users must familiarize themselves with the architecture of the CCC core and its input, output, and
configuration ports to implement the desired delay and output frequency in the CCC structure.
Figure 4-22 shows a model of the CCC with configurable blocks and switches.
88
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
CLKA
/n
C
90°
180°
270°
0°
PLL
Core
/m
(7)
(6)
(5)
(4)
M
U
X
A
D
/u
C
(2)
Internal
C
C
(1)
(0)
GLA
C
(2)
D
(1)
D
M
U
X
B
C
C
C
D
/v
YB
C
C
D
GLB
CLKB
C
Internal
M
U
X
C
D
/w
YC
C
C
C
D
GLC
CLKC
C
Internal
Figure 4-22 • CCC Block Control Bits – Graphical Representation of Assignments
Revision 5
89
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Loading the Configuration Register
The most important part of CCC dynamic configuration is to load the shift register properly with the
configuration bits. There are different ways to access and load the configuration shift register:
•
JTAG interface
•
Logic core
•
Specific I/O tiles
JTAG Interface
The JTAG interface requires no additional I/O pins. The JTAG TAP controller is used to control the
loading of the CCC configuration shift register.
Low power flash devices provide a user interface macro between the JTAG pins and the device core
logic. This macro is called UJTAG. A user should instantiate the UJTAG macro in his design to access the
configuration register ports via the JTAG pins.
For more information on CCC dynamic reconfiguration using UJTAG, refer to the "UJTAG Applications in
Microsemi’s Low Power Flash Devices" section on page 297.
Logic Core
If the logic core is employed, the user must design a module to provide the configuration data and control
the shifting and updating of the CCC configuration shift register. In effect, this is a user-designed TAP
controller, which requires additional chip resources.
Specific I/O Tiles
If specific I/O tiles are used for configuration, the user must provide the external equivalent of a TAP
controller. This does not require additional core resources but does use pins.
Shifting the Configuration Data
To enter a new configuration, all 81 bits must shift in via SDIN. After all bits are shifted, SSHIFT must go
LOW and SUPDATE HIGH to enable the new configuration. For simulation purposes, bits and
are "don't care."
The SUPDATE signal must be LOW during any clock cycle where SSHIFT is active. After SUPDATE is
asserted, it must go back to the LOW state until a new update is required.
PLL Configuration Bits Description
Table 4-8 • Configuration Bit Descriptions for the CCC Blocks
Config.
Bits
Signal
Name
1
GLMUXCFG [1:0] NGMUX configuration
Description
The configuration bits specify the input clocks
to the NGMUX (refer to Table 4-17 on
page 94).2
86
OCDIVHALF1
Division by half
When the PLL is bypassed, the 100 MHz RC
oscillator can be divided by the divider factor
in Table 4-18 on page 95.
85
OBDIVHALF1
Division by half
When the PLL is bypassed, the 100 MHz RC
oscillator can be divided by a 0.5 factor (refer
to Table 4-18 on page 95).
84
OADIVHALF1
Division by half
When the PLL is bypassed, the 100 MHz RC
oscillator can be divided by certain 0.5 factor
(refer to Table 4-16 on page 94).
Notes:
1. The configuration bits are only for the Fusion dynamic CCC.
2. This value depends on the input clock source, so Layout must complete before these bits can be set.
After completing Layout in Designer, generate the "CCC_Configuration" report by choosing Tools >
Report > CCC_Configuration. The report contains the appropriate settings for these bits.
90
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Table 4-8 • Configuration Bit Descriptions for the CCC Blocks (continued)
Config.
Bits
Signal
Name
Description
83
RXCSEL1
CLKC input selection
Select the CLKC input clock source between
RC oscillator and crystal oscillator (refer to
Table 4-16 on page 94).2
82
RXBSEL1
CLKB input selection
Select the CLKB input clock source between
RC oscillator and crystal oscillator (refer to
Table 4-16 on page 94).2
81
RXASEL1
CLKA input selection
Select the CLKA input clock source between
RC oscillator and crystal oscillator (refer to
Table 4-16 on page 94).2
80
RESETEN
Reset Enable
Enables (active high) the synchronization of
PLL output dividers after dynamic
reconfiguration (SUPDATE). The Reset
Enable signal is READ-ONLY.
79
DYNCSEL
Clock Input C Dynamic
Select
Configures clock input C to be sent to GLC for
dynamic control.2
78
DYNBSEL
Clock Input B Dynamic
Select
Configures clock input B to be sent to GLB for
dynamic control.2
77
DYNASEL
Clock Input A Dynamic
Select
Configures clock input A for dynamic PLL
configuration.2
VCO Gear Control
Three-bit VCO Gear Control for four frequency
ranges (refer to Table 4-19 on page 95 and
Table 4-20 on page 95).
VCOSEL[2:0]
73
STATCSEL
MUX Select on Input C
MUX selection for clock input C2
72
STATBSEL
MUX Select on Input B
MUX selection for clock input B2
71
STATASEL
MUX Select on Input A
MUX selection for clock input A2
DLYC[4:0]
YC Output Delay
Sets the output delay value for YC.
DLYB[4:0]
YB Output Delay
Sets the output delay value for YB.
DLYGLC[4:0]
GLC Output Delay
Sets the output delay value for GLC.
DLYGLB[4:0]
GLB Output Delay
Sets the output delay value for GLB.
DLYGLA[4:0]
Primary Output Delay
Primary GLA output delay
XDLYSEL
System Delay Select
When selected, inserts System Delay in the
feedback path in Figure 4-20 on page 85.
Sets the feedback delay value for the
feedback element in Figure 4-20 on page 85.
45
FBDLY[4:0]
Feedback Delay
FBSEL[1:0]
Primary Feedback Delay Controls the feedback MUX: no delay, include
Select
programmable delay element, or use external
feedback.
OCMUX[2:0]
Secondary 2 Output
Select
Selects from the VCO’s four phase outputs for
GLC/YC.
OBMUX[2:0]
Secondary 1 Output
Select
Selects from the VCO’s four phase outputs for
GLB/YB.
Notes:
1. The configuration bits are only for the Fusion dynamic CCC.
2. This value depends on the input clock source, so Layout must complete before these bits can be set.
After completing Layout in Designer, generate the "CCC_Configuration" report by choosing Tools >
Report > CCC_Configuration. The report contains the appropriate settings for these bits.
Revision 5
91
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Table 4-8 • Configuration Bit Descriptions for the CCC Blocks (continued)
Config.
Bits
Signal
OAMUX[2:0]
GLA Output Select
Selects from the VCO’s four phase outputs for
GLA.
OCDIV[4:0]
Secondary 2 Output
Divider
Sets the divider value for the GLC/YC outputs.
Also known as divider w in Figure 4-20 on
page 85. The divider value will be OCDIV[4:0]
+ 1.
OBDIV[4:0]
Secondary 1 Output
Divider
Sets the divider value for the GLB/YB outputs.
Also known as divider v in Figure 4-20 on
page 85. The divider value will be OBDIV[4:0]
+ 1.
OADIV[4:0]
Primary Output Divider
Sets the divider value for the GLA output. Also
known as divider u in Figure 4-20 on page 85.
The divider value will be OADIV[4:0] + 1.
FBDIV[6:0]
Feedback Divider
Sets the divider value for the PLL core
feedback. Also known as divider m in
Figure 4-20 on page 85. The divider value will
be FBDIV[6:0] + 1.
FINDIV[6:0]
Input Divider
Input Clock Divider (/n). Sets the divider value
for the input delay on CLKA. The divider value
will be FINDIV[6:0] + 1.
Name
Description
Notes:
1. The configuration bits are only for the Fusion dynamic CCC.
2. This value depends on the input clock source, so Layout must complete before these bits can be set.
After completing Layout in Designer, generate the "CCC_Configuration" report by choosing Tools >
Report > CCC_Configuration. The report contains the appropriate settings for these bits.
92
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Table 4-9 to Table 4-15 on page 94 provide descriptions of the configuration data for the configuration
bits.
Table 4-9 • Input Clock Divider, FINDIV[6:0] (/n)
FINDIV State
Divisor
New Frequency Factor
0
1
1.00000
1
2
0.50000
…
…
…
127
128
0.0078125
Divisor
New Frequency Factor
0
1
1
1
2
2
Table 4-10 • Feedback Clock Divider, FBDIV[6:0] (/m)
FBDIV State
…
…
…
127
128
128
Divisor
New Frequency Factor
0
1
1.00000
1
2
0.50000
Table 4-11 • Output Frequency Dividers
A Output Divider, OADIV (/u);
B Output Divider, OBDIV (/v);
C Output Divider, OCDIV (/w)
OADIV; OBDIV;
CDIV State
…
…
…
31
32
0.03125
Table 4-12 • MUXA, MUXB, MUXC
OAMUX; OBMUX; OCMUX State
MUX Input Selected
0
None. Six-input MUX and PLL are bypassed.
Clock passes only through global MUX and goes directly
into HC ribs.
1
Not available
2
PLL feedback delay line output
3
Not used
4
PLL VCO 0° phase shift
5
PLL VCO 270° phase shift
6
PLL VCO 180° phase shift
7
PLL VCO 90° phase shift
Revision 5
93
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Table 4-13 • 2-Bit Feedback MUX
FBSEL State
MUX Input Selected
0
Ground. Used for power-down mode in power-down logic
block.
1
PLL VCO 0° phase shift
2
PLL delayed VCO 0° phase shift
3
N/A
Table 4-14 • Programmable Delay Selection for Feedback Delay and Secondary Core Output Delays
FBDLY; DLYYB; DLYYC State
Delay Value
0
Typical delay = 600 ps
1
Typical delay = 760 ps
2
Typical delay = 920 ps
…
…
31
Typical delay = 5.56 ns
Table 4-15 • Programmable Delay Selection for Global Clock Output Delays
DLYGLA; DLYGLB; DLYGLC State
Delay Value
0
Typical delay = 225 ps
1
Typical delay = 760 ps
2
Typical delay = 920 ps
…
…
31
Typical delay = 5.56 ns
Table 4-16 • Fusion Dynamic CCC Clock Source Selection
DYNASEL
Source of CLKA
1
RXASEL
0
RC Oscillator
1
1
Crystal Oscillator
DYNBSEL
Source of CLKB
RXBSEL
1
0
RC Oscillator
1
1
Crystal Oscillator
DYNCSEL
Source of CLKC
1
0
RC Oscillator
1
1
Crystal Oscillator
RXBSEL
Table 4-17 • Fusion Dynamic CCC NGMUX Configuration
GLMUXCFG
NGMUX Select Signal
Supported Input Clocks to NGMUX
0
GLA
1
GLC
01
0
GLA
1
GLINT
10
0
GLC
1
GLINT
00
94
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Table 4-18 • Fusion Dynamic CCC Division by Half Configuration
OADIVHALF /
OBDIVHALF /
OCDIVHALF
OADIV /
OBDIV /
OCDIV
(in decimal)
Divider Factor
2
1.5
4
2.5
6
3.5
28.6
8
4.5
22.2
10
5.5
18.2
12
6.5
15.4
14
7.5
13.3
16
8.5
11.8
18
9.5
10.5
20
10.5
9.5
22
11.5
8.7
24
12.5
8.0
26
13.5
7.4
28
14.5
6.9
0–31
1–32
1
0
Input Clock
Frequency
Output Clock
Frequency (MHz)
100 MHz RC
Oscillator
66.7
Other Clock Sources
40.0
Depends on other
divider settings
Table 4-19 • Configuration Bit / VCOSEL Selection for All Families
VCOSEL[2:1]
00
Min.
(MHz)
Voltage
01
10
11
Max.
(MHz)
Min.
(MHz)
Max.
(MHz)
Min.
(MHz)
Max.
(MHz)
Min.
(MHz)
Max.
(MHz)
IGLOO and IGLOO PLUS
1.2 V ± 5%
24
35
30
70
60
140
135
160
1.5 V ± 5%
24
43.75
30
87.5
60
175
135
250
ProASIC3L, RT ProASIC3, and Military ProASIC3/L
1.2 V ± 5%
24
35
30
70
60
140
135
250
1.5 V ± 5%
24
43.75
30
70
60
175
135
350
24
43.75
33.75
87.5
67.5
175
135
350
ProASIC3 and Fusion
1.5 V ± 5%
Table 4-20 • Configuration Bit / VCOSEL Selection for All Families
VCOSEL[0]
Description
0
Fast PLL lock acquisition time with high tracking jitter. Refer to the corresponding datasheet for specific
value and definition.
1
Slow PLL lock acquisition time with low tracking jitter. Refer to the corresponding datasheet for specific
value and definition.
Revision 5
95
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Software Configuration
SmartGen automatically generates the desired CCC functional block by configuring the control bits, and
allows the user to select two CCC modes: Static PLL and Delayed Clock (CLKDLY).
Static PLL Configuration
The newly implemented Visual PLL Configuration Wizard feature provides the user a quick and easy way
to configure the PLL with the desired settings (Figure 4-23). The user can invoke SmartGen to set the
parameters and generate the netlist file with the appropriate flash configuration bits set for the CCCs. As
mentioned in "PLL Macro Block Diagram" on page 69, the input reference clock CLKA can be configured
to be driven by Hardwired I/O, External I/O, or Core Logic. The user enters the desired settings for all the
parameters (output frequency, output selection, output phase adjustment, clock delay, feedback delay,
and system delay). Notice that the actual values (divider values, output frequency, delay values, and
phase) are shown to aid the user in reaching the desired design frequency in real time. These values are
typical-case data. Best- and worst-case data can be observed through static timing analysis in
SmartTime within Designer.
For dynamic configuration, the CCC parameters are defined using either the external JTAG port or an
internally defined serial interface via the built-in dynamic shift register. This feature provides the ability to
compensate for changes in the external environment.
VCO Clock Frequency
Programmable Output Delay Elements
Input
Selection
Fixed System Delay
Output
Selection
Feedback Selection (Feedback MUX)
Figure 4-23 • Visual PLL Configuration Wizard
96
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
Feedback Configuration
The PLL provides both internal and external feedback delays. Depending on the configuration, various
combinations of feedback delays can be achieved.
Internal Feedback Configuration
This configuration essentially sets the feedback multiplexer to route the VCO output of the PLL core as
the input to the feedback of the PLL. The feedback signal can be processed with the fixed system and
the adjustable feedback delay, as shown in Figure 4-24. The dividers are automatically configured by
SmartGen based on the user input.
Indicated below is the System Delay pull-down menu. The System Delay can be bypassed by setting it to
0. When set, it adds a 2 ns delay to the feedback path (which results in delay advancement of the output
clock by 2 ns).
Figure 4-24 • Internal Feedback with Selectable System Delay
Figure 4-25 shows the controllable Feedback Delay. If set properly in conjunction with the fixed System
Delay, the total output delay can be advanced significantly.
Figure 4-25 • Internal Feedback with Selectable Feedback Delay
Revision 5
97
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
External Feedback Configuration
For certain applications, such as those requiring generation of PCB clocks that must be matched with
existing board delays, it is useful to implement an external feedback, EXTFB. The Phase Detector of the
PLL core will receive CLKA and EXTFB as inputs. EXTFB may be processed by the fixed System Delay
element as well as the M divider element. The EXTFB option is currently not supported.
After setting all the required parameters, users can generate one or more PLL configurations with HDL or
EDIF descriptions by clicking the Generate button. SmartGen gives the option of saving session results
and messages in a log file:
****************
Macro Parameters
****************
Name
Family
Output Format
Type
Input Freq(MHz)
CLKA Source
Feedback Delay Value Index
Feedback Mux Select
XDLY Mux Select
Primary Freq(MHz)
Primary PhaseShift
Primary Delay Value Index
Primary Mux Select
Secondary1 Freq(MHz)
Use GLB
Use YB
GLB Delay Value Index
YB Delay Value Index
Secondary1 PhaseShift
Secondary1 Mux Select
Secondary2 Freq(MHz)
Use GLC
Use YC
GLC Delay Value Index
YC Delay Value Index
Secondary2 PhaseShift
Secondary2 Mux Select
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
test_pll
ProASIC3E
VHDL
Static PLL
10.000
Hardwired I/O
1
2
No
33.000
0
1
4
66.000
YES
YES
1
1
0
4
101.000
YES
NO
1
1
0
4
…
…
…
Primary Clock frequency 33.333
Primary Clock Phase Shift 0.000
Primary Clock Output Delay from CLKA 0.180
Secondary1
Secondary1
Secondary1
Secondary1
Clock
Clock
Clock
Clock
frequency 66.667
Phase Shift 0.000
Global Output Delay from CLKA 0.180
Core Output Delay from CLKA 0.625
Secondary2 Clock frequency 100.000
Secondary2 Clock Phase Shift 0.000
Secondary2 Clock Global Output Delay from CLKA 0.180
Below is an example Verilog HDL description of a legal PLL core configuration generated by SmartGen:
module test_pll(POWERDOWN,CLKA,LOCK,GLA);
input POWERDOWN, CLKA;
output LOCK, GLA;
98
R e visio n 5
ProASIC3 nano FPGA Fabric User’s Guide
wire VCC, GND;
VCC VCC_1_net(.Y(VCC));
GND GND_1_net(.Y(GND));
PLL Core(.CLKA(CLKA), .EXTFB(GND), .POWERDOWN(POWERDOWN),
.GLA(GLA), .LOCK(LOCK), .GLB(), .YB(), .GLC(), .YC(),
.OADIV0(GND), .OADIV1(GND), .OADIV2(GND), .OADIV3(GND),
.OADIV4(GND), .OAMUX0(GND), .OAMUX1(GND), .OAMUX2(VCC),
.DLYGLA0(GND), .DLYGLA1(GND), .DLYGLA2(GND), .DLYGLA3(GND)
, .DLYGLA4(GND), .OBDIV0(GND), .OBDIV1(GND), .OBDIV2(GND),
.OBDIV3(GND), .OBDIV4(GND), .OBMUX0(GND), .OBMUX1(GND),
.OBMUX2(GND), .DLYYB0(GND), .DLYYB1(GND), .DLYYB2(GND),
.DLYYB3(GND), .DLYYB4(GND), .DLYGLB0(GND), .DLYGLB1(GND),
.DLYGLB2(GND), .DLYGLB3(GND), .DLYGLB4(GND), .OCDIV0(GND),
.OCDIV1(GND), .OCDIV2(GND), .OCDIV3(GND), .OCDIV4(GND),
.OCMUX0(GND), .OCMUX1(GND), .OCMUX2(GND), .DLYYC0(GND),
.DLYYC1(GND), .DLYYC2(GND), .DLYYC3(GND), .DLYYC4(GND),
.DLYGLC0(GND), .DLYGLC1(GND), .DLYGLC2(GND), .DLYGLC3(GND)
, .DLYGLC4(GND), .FINDIV0(VCC), .FINDIV1(GND), .FINDIV2(
VCC), .FINDIV3(GND), .FINDIV4(GND), .FINDIV5(GND),
.FINDIV6(GND), .FBDIV0(VCC), .FBDIV1(GND), .FBDIV2(VCC),
.FBDIV3(GND), .FBDIV4(GND), .FBDIV5(GND), .FBDIV6(GND),
.FBDLY0(GND), .FBDLY1(GND), .FBDLY2(GND), .FBDLY3(GND),
.FBDLY4(GND), .FBSEL0(VCC), .FBSEL1(GND), .XDLYSEL(GND),
.VCOSEL0(GND), .VCOSEL1(GND), .VCOSEL2(GND));
defparam Core.VCOFREQUENCY = 33.000;
endmodule
The "PLL Configuration Bits Description" section on page 90 provides descriptions of the PLL
configuration bits for completeness. The configuration bits are shown as busses only for purposes of
illustration. They will actually be broken up into individual pins in compilation libraries and all simulation
models. For example, the FBSEL[1:0] bus will actually appear as pins FBSEL1 and FBSEL0. The setting
of these select lines for the static PLL configuration is performed by the software and is completely
transparent to the user.
Revision 5
99
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Dynamic PLL Configuration
To generate a dynamically reconfigurable CCC, the user should select Dynamic CCC in the
configuration section of the SmartGen GUI (Figure 4-26). This will generate both the CCC core and the
configuration shift register / control bit MUX.
Figure 4-26 • SmartGen GUI
Even if dynamic configuration is selected in SmartGen, the user must still specify the static configuration
data for the CCC (Figure 4-27). The specified static configuration is used whenever the MODE signal is
set to LOW and the CCC is required to function in the static mode. The static configuration data can be
used as the default behavior of the CCC where required.
Figure 4-27 • Dynamic CCC Configuration in SmartGen
100
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
When SmartGen is used to define the configuration that will be shifted in via the serial interface,
SmartGen prints out the values of the 81 configuration bits. For ease of use, several configuration bits
are automatically inferred by SmartGen when the dynamic PLL core is generated; however,
(STATASEL, STATBSEL, STATCSEL) and (DYNASEL, DYNBSEL, DYNCSEL) depend on the
input clock source of the corresponding CCC. Users must first run Layout in Designer to determine the
exact setting for these ports. After Layout is complete, generate the "CCC_Configuration" report by
choosing Tools > Reports > CCC_Configuration in the Designer software. Refer to "PLL Configuration
Bits Description" on page 90 for descriptions of the PLL configuration bits. For simulation purposes, bits
and are "don't care." Therefore, it is strongly suggested that SmartGen be used to
generate the correct configuration bit settings for the dynamic PLL core.
After setting all the required parameters, users can generate one or more PLL configurations with HDL or
EDIF descriptions by clicking the Generate button. SmartGen gives the option of saving session results
and messages in a log file:
****************
Macro Parameters
****************
Name
Family
Output Format
Type
Input Freq(MHz)
CLKA Source
Feedback Delay Value Index
Feedback Mux Select
XDLY Mux Select
Primary Freq(MHz)
Primary PhaseShift
Primary Delay Value Index
Primary Mux Select
Secondary1 Freq(MHz)
Use GLB
Use YB
GLB Delay Value Index
YB Delay Value Index
Secondary1 PhaseShift
Secondary1 Mux Select
Secondary1 Input Freq(MHz)
CLKB Source
Secondary2 Freq(MHz)
Use GLC
Use YC
GLC Delay Value Index
YC Delay Value Index
Secondary2 PhaseShift
Secondary2 Mux Select
Secondary2 Input Freq(MHz)
CLKC Source
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
dyn_pll_hardio
ProASIC3E
VERILOG
Dynamic CCC
30.000
Hardwired I/O
1
1
No
33.000
0
1
4
40.000
YES
NO
1
1
0
0
40.000
Hardwired I/O
50.000
YES
NO
1
1
0
0
50.000
Hardwired I/O
Configuration Bits:
FINDIV[6:0]
0000101
FBDIV[6:0]
0100000
OADIV[4:0]
00100
OBDIV[4:0]
00000
OCDIV[4:0]
00000
OAMUX[2:0]
100
OBMUX[2:0]
000
OCMUX[2:0]
000
FBSEL[1:0]
01
FBDLY[4:0]
00000
XDLYSEL
0
DLYGLA[4:0]
00000
DLYGLB[4:0]
00000
Revision 5
101
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
DLYGLC[4:0]
DLYYB[4:0]
DLYYC[4:0]
VCOSEL[2:0]
00000
00000
00000
100
Primary Clock Frequency 33.000
Primary Clock Phase Shift 0.000
Primary Clock Output Delay from CLKA 1.695
Secondary1 Clock Frequency 40.000
Secondary1 Clock Phase Shift 0.000
Secondary1 Clock Global Output Delay from CLKB 0.200
Secondary2 Clock Frequency 50.000
Secondary2 Clock Phase Shift 0.000
Secondary2 Clock Global Output Delay from CLKC 0.200
######################################
# Dynamic Stream Data
######################################
-------------------------------------|NAME
|SDIN
|VALUE
|TYPE
|
-------------------------------------|FINDIV |[6:0]
|0000101 |EDIT
|
|FBDIV
|[13:7]
|0100000 |EDIT
|
|OADIV
|[18:14] |00100
|EDIT
|
|OBDIV
|[23:19] |00000
|EDIT
|
|OCDIV
|[28:24] |00000
|EDIT
|
|OAMUX
|[31:29] |100
|EDIT
|
|OBMUX
|[34:32] |000
|EDIT
|
|OCMUX
|[37:35] |000
|EDIT
|
|FBSEL
|[39:38] |01
|EDIT
|
|FBDLY
|[44:40] |00000
|EDIT
|
|XDLYSEL |[45]
|0
|EDIT
|
|DLYGLA |[50:46] |00000
|EDIT
|
|DLYGLB |[55:51] |00000
|EDIT
|
|DLYGLC |[60:56] |00000
|EDIT
|
|DLYYB
|[65:61] |00000
|EDIT
|
|DLYYC
|[70:66] |00000
|EDIT
|
|STATASEL|[71]
|X
|MASKED
|
|STATBSEL|[72]
|X
|MASKED
|
|STATCSEL|[73]
|X
|MASKED
|
|VCOSEL |[76:74] |100
|EDIT
|
|DYNASEL |[77]
|X
|MASKED
|
|DYNBSEL |[78]
|X
|MASKED
|
|DYNCSEL |[79]
|X
|MASKED
|
|RESETEN |[80]
|1
|READONLY |
Below is the resultant Verilog HDL description of a legal dynamic PLL core configuration generated by
SmartGen:
module dyn_pll_macro(POWERDOWN, CLKA, LOCK, GLA, GLB, GLC, SDIN, SCLK, SSHIFT, SUPDATE,
MODE, SDOUT, CLKB, CLKC);
input POWERDOWN, CLKA;
output LOCK, GLA, GLB, GLC;
input SDIN, SCLK, SSHIFT, SUPDATE, MODE;
output SDOUT;
input CLKB, CLKC;
wire VCC, GND;
VCC VCC_1_net(.Y(VCC));
GND GND_1_net(.Y(GND));
102
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
DYNCCC Core(.CLKA(CLKA), .EXTFB(GND), .POWERDOWN(POWERDOWN), .GLA(GLA), .LOCK(LOCK),
.CLKB(CLKB), .GLB(GLB), .YB(), .CLKC(CLKC), .GLC(GLC), .YC(), .SDIN(SDIN),
.SCLK(SCLK), .SSHIFT(SSHIFT), .SUPDATE(SUPDATE), .MODE(MODE), .SDOUT(SDOUT),
.OADIV0(GND), .OADIV1(GND), .OADIV2(VCC), .OADIV3(GND), .OADIV4(GND), .OAMUX0(GND),
.OAMUX1(GND), .OAMUX2(VCC), .DLYGLA0(GND), .DLYGLA1(GND), .DLYGLA2(GND),
.DLYGLA3(GND), .DLYGLA4(GND), .OBDIV0(GND), .OBDIV1(GND), .OBDIV2(GND),
.OBDIV3(GND), .OBDIV4(GND), .OBMUX0(GND), .OBMUX1(GND), .OBMUX2(GND), .DLYYB0(GND),
.DLYYB1(GND), .DLYYB2(GND), .DLYYB3(GND), .DLYYB4(GND), .DLYGLB0(GND),
.DLYGLB1(GND), .DLYGLB2(GND), .DLYGLB3(GND), .DLYGLB4(GND), .OCDIV0(GND),
.OCDIV1(GND), .OCDIV2(GND), .OCDIV3(GND), .OCDIV4(GND), .OCMUX0(GND), .OCMUX1(GND),
.OCMUX2(GND), .DLYYC0(GND), .DLYYC1(GND), .DLYYC2(GND), .DLYYC3(GND), .DLYYC4(GND),
.DLYGLC0(GND), .DLYGLC1(GND), .DLYGLC2(GND), .DLYGLC3(GND), .DLYGLC4(GND),
.FINDIV0(VCC), .FINDIV1(GND), .FINDIV2(VCC), .FINDIV3(GND), .FINDIV4(GND),
.FINDIV5(GND), .FINDIV6(GND), .FBDIV0(GND), .FBDIV1(GND), .FBDIV2(GND),
.FBDIV3(GND), .FBDIV4(GND), .FBDIV5(VCC), .FBDIV6(GND), .FBDLY0(GND), .FBDLY1(GND),
.FBDLY2(GND), .FBDLY3(GND), .FBDLY4(GND), .FBSEL0(VCC), .FBSEL1(GND),
.XDLYSEL(GND), .VCOSEL0(GND), .VCOSEL1(GND), .VCOSEL2(VCC));
defparam Core.VCOFREQUENCY = 165.000;
endmodule
Delayed Clock Configuration
The CLKDLY macro can be generated with the desired delay and input clock source (Hardwired I/O,
External I/O, or Core Logic), as in Figure 4-28.
Figure 4-28 • Delayed Clock Configuration Dialog Box
After setting all the required parameters, users can generate one or more PLL configurations with HDL or
EDIF descriptions by clicking the Generate button. SmartGen gives the option of saving session results
and messages in a log file:
****************
Macro Parameters
****************
Name
Family
Output Format
Type
Delay Index
CLKA Source
:
:
:
:
:
:
delay_macro
ProASIC3
Verilog
Delayed Clock
2
Hardwired I/O
Total Clock Delay = 0.935 ns.
The resultant CLKDLY macro Verilog netlist is as follows:
module delay_macro(GL,CLK);
output GL;
input CLK;
Revision 5
103
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
wire VCC, GND;
VCC VCC_1_net(.Y(VCC));
GND GND_1_net(.Y(GND));
CLKDLY Inst1(.CLK(CLK), .GL(GL), .DLYGL0(VCC), .DLYGL1(GND), .DLYGL2(VCC),
.DLYGL3(GND), .DLYGL4(GND));
endmodule
Detailed Usage Information
Clock Frequency Synthesis
Deriving clocks of various frequencies from a single reference clock is known as frequency synthesis.
The PLL has an input frequency range from 1.5 to 350 MHz. This frequency is automatically divided
down to a range between 1.5 MHz and 5.5 MHz by input dividers (not shown in Figure 4-19 on page 84)
between PLL macro inputs and PLL phase detector inputs. The VCO output is capable of an output
range from 24 to 350 MHz. With dividers before the input to the PLL core and following the VCO outputs,
the VCO output frequency can be divided to provide the final frequency range from 0.75 to 350 MHz.
Using SmartGen, the dividers are automatically set to achieve the closest possible matches to the
specified output frequencies.
Users should be cautious when selecting the desired PLL input and output frequencies and the I/O buffer
standard used to connect to the PLL input and output clocks. Depending on the I/O standards used for
the PLL input and output clocks, the I/O frequencies have different maximum limits. Refer to the family
datasheets for specifications of maximum I/O frequencies for supported I/O standards. Desired PLL input
or output frequencies will not be achieved if the selected frequencies are higher than the maximum I/O
frequencies allowed by the selected I/O standards. Users should be careful when selecting the I/O
standards used for PLL input and output clocks. Performing post-layout simulation can help detect this
type of error, which will be identified with pulse width violation errors. Users are strongly encouraged to
perform post-layout simulation to ensure the I/O standard used can provide the desired PLL input or
output frequencies. Users can also choose to cascade PLLs together to achieve the high frequencies
needed for their applications. Details of cascading PLLs are discussed in the "Cascading CCCs" section
on page 109.
In SmartGen, the actual generated frequency (under typical operating conditions) will be displayed
beside the requested output frequency value. This provides the ability to determine the exact frequency
that can be generated by SmartGen, in real time. The log file generated by SmartGen is a useful tool in
determining how closely the requested clock frequencies match the user specifications. For example,
assume a user specifies 101 MHz as one of the secondary output frequencies. If the best output
frequency that could be achieved were 100 MHz, the log file generated by SmartGen would indicate the
actual generated frequency.
Simulation Verification
The integration of the generated PLL and CLKDLY modules is similar to any VHDL component or Verilog
module instantiation in a larger design; i.e., there is no special requirement that users need to take into
account to successfully synthesize their designs.
For simulation purposes, users need to refer to the VITAL or Verilog library that includes the functional
description and associated timing parameters. Refer to the Software Tools section of the Microsemi SoC
Products Group website to obtain the family simulation libraries. If Designer is installed, these libraries
are stored in the following locations:
\lib\vtl\95\proasic3.vhd
\lib\vtl\95\proasic3e.vhd
\lib\vlog\proasic3.v
\lib\vlog\proasic3e.v
For Libero users, there is no need to compile the simulation libraries, as they are conveniently precompiled in the ModelSim® Microsemi simulation tool.
104
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
The following is an example of a PLL configuration utilizing the clock frequency synthesis and clock delay
adjustment features. The steps include generating the PLL core with SmartGen, performing simulation
for verification with ModelSim, and performing static timing analysis with SmartTime in Designer.
Parameters of the example PLL configuration:
Input Frequency – 20 MHz
Primary Output Requirement – 20 MHz with clock advancement of 3.02 ns
Secondary 1 Output Requirement – 40 MHz with clock delay of 2.515 ns
Figure 4-29 shows the SmartGen settings. Notice that the overall delays are calculated automatically,
allowing the user to adjust the delay elements appropriately to obtain the desired delays.
Figure 4-29 • SmartGen Settings
After confirming the correct settings, generate a structural netlist of the PLL and verify PLL core settings
by checking the log file:
Name
Family
Output Format
Type
Input Freq(MHz)
CLKA Source
Feedback Delay Value Index
Feedback Mux Select
XDLY Mux Select
Primary Freq(MHz)
Primary PhaseShift
Primary Delay Value Index
Primary Mux Select
Secondary1 Freq(MHz)
Use GLB
Use YB
…
…
…
Primary Clock frequency 20.000
Primary Clock Phase Shift 0.000
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
test_pll_delays
ProASIC3E
VHDL
Static PLL
20.000
Hardwired I/O
21
2
No
20.000
0
1
4
40.000
YES
NO
Revision 5
105
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Primary Clock Output Delay from CLKA -3.020
Secondary1 Clock frequency 40.000
Secondary1 Clock Phase Shift 0.000
Secondary1 Clock Global Output Delay from CLKA 2.515
Next, perform simulation in ModelSim to verify the correct delays. Figure 4-30 shows the simulation
results. The delay values match those reported in the SmartGen PLL Wizard.
Primary Clock Output Time
Advancement from CLKA
Secondary1 Clock Global
Output Delay from CLKA
Figure 4-30 • ModelSim Simulation Results
The timing can also be analyzed using SmartTime in Designer. The user should import the synthesized
netlist to Designer, perform Compile and Layout, and then invoke SmartTime. Go to Tools > Options
and change the maximum delay operating conditions to Typical Case. Then expand the Clock-to-Out
paths of GLA and GLB and the individual components of the path delays are shown. The path of GLA is
shown in Figure 4-31 on page 107 displaying the same delay value.
106
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Figure 4-31 • Static Timing Analysis Using SmartTime
Place-and-Route Stage Considerations
Several considerations must be noted to properly place the CCC macros for layout.
For CCCs with clock inputs configured with the Hardwired I/O–Driven option:
•
PLL macros must have the clock input pad coming from one of the GmA* locations.
•
CLKDLY macros must have the clock input pad coming from one of the Global I/Os.
If a PLL with a Hardwired I/O input is used at a CCC location and a Hardwired I/O–Driven CLKDLY
macro is used at the same CCC location, the clock input of the CLKDLY macro must be chosen from one
of the GmB* or GmC* pin locations. If the PLL is not used or is an External I/O–Driven or Core Logic–
Driven PLL, the clock input of the CLKDLY macro can be sourced from the GmA*, GmB*, or GmC* pin
locations.
For CCCs with clock inputs configured with the External I/O–Driven option, the clock input pad can be
assigned to any regular I/O location (IO******** pins). Note that since global I/O pins can also be used as
regular I/Os, regardless of CCC function (CLKDLY or PLL), clock inputs can also be placed in any of
these I/O locations.
By default, the Designer layout engine will place global nets in the design at one of the six chip globals.
When the number of globals in the design is greater than six, the Designer layout engine will
automatically assign additional globals to the quadrant global networks of the low power flash devices. If
the user wishes to decide which global signals should be assigned to chip globals (six available) and
which to the quadrant globals (three per quadrant for a total of 12 available), the assignment can be
achieved with PinEditor, ChipPlanner, or by importing a placement constraint file. Layout will fail if the
Revision 5
107
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
global assignments are not allocated properly. See the "Physical Constraints for Quadrant Clocks"
section for information on assigning global signals to the quadrant clock networks.
Promoted global signals will be instantiated with CLKINT macros to drive these signals onto the global
network. This is automatically done by Designer when the Auto-Promotion option is selected. If the user
wishes to assign the signals to the quadrant globals instead of the default chip globals, this can done by
using ChipPlanner, by declaring a physical design constraint (PDC), or by importing a PDC file.
Physical Constraints for Quadrant Clocks
If it is necessary to promote global clocks (CLKBUF, CLKINT, PLL, CLKDLY) to quadrant clocks, the user
can define PDCs to execute the promotion. PDCs can be created using PDC commands (pre-compile) or
the MultiView Navigator (MVN) interface (post-compile). The advantage of using the PDC flow over the
MVN flow is that the Compile stage is able to automatically promote any regular net to a global net before
assigning it to a quadrant. There are three options to place a quadrant clock using PDC commands:
•
Place a clock core (not hardwired to an I/O) into a quadrant clock location.
•
Place a clock core (hardwired to an I/O) into an I/O location (set_io) or an I/O module location
(set_location) that drives a quadrant clock location.
•
Assign a net driven by a regular net or a clock net to a quadrant clock using the following
command:
assign_local_clock -net -type quadrant
where
is the name of the net assigned to the local user clock region.
defines which quadrant the net should be assigned to. Quadrant
clock regions are defined as UL (upper left), UR (upper right), LL (lower left), and LR (lower right).
Note: If the net is a regular net, the software inserts a CLKINT buffer on the net.
For example:
assign_local_clock -net localReset -type quadrant UR
Keep in mind the following when placing quadrant clocks using MultiView Navigator:
Hardwired I/O–Driven CCCs
•
Find the associated clock input port under the Ports tab, and place the input port at one of the
Gmn* locations using PinEditor or I/O Attribute Editor, as shown in Figure 4-32.
Figure 4-32 • Port Assignment for a CCC with Hardwired I/O Clock Input
108
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
•
Use quadrant global region assignments by finding the clock net associated with the CCC macro
under the Nets tab and creating a quadrant global region for the net, as shown in Figure 4-33.
Figure 4-33 • Quadrant Clock Assignment for a Global Net
External I/O–Driven CCCs
The above-mentioned recommendation for proper layout techniques will ensure the correct assignment.
It is possible that, especially with External I/O–Driven CCC macros, placement of the CCC macro in a
desired location may not be achieved. For example, assigning an input port of an External I/O–Driven
CCC near a particular CCC location does not guarantee global assignments to the desired location. This
is because the clock inputs of External I/O–Driven CCCs can be assigned to any I/O location; therefore,
it is possible that the CCC connected to the clock input will be routed to a location other than the one
closest to the I/O location, depending on resource availability and placement constraints.
Clock Placer
The clock placer is a placement engine for low power flash devices that places global signals on the chip
global and quadrant global networks. Based on the clock assignment constraints for the chip global and
quadrant global clocks, it will try to satisfy all constraints, as well as creating quadrant clock regions when
necessary. If the clock placer fails to create the quadrant clock regions for the global signals, it will report
an error and stop Layout.
The user must ensure that the constraints set to promote clock signals to quadrant global networks are
valid.
Cascading CCCs
The CCCs in low power flash devices can be cascaded. Cascading CCCs can help achieve more
accurate PLL output frequency results than those achievable with a single CCC. In addition, this
technique is useful when the user application requires the output clock of the PLL to be a multiple of the
reference clock by an integer greater than the maximum feedback divider value of the PLL (divide by
128) to achieve the desired frequency.
For example, the user application may require a 280 MHz output clock using a 2 MHz input reference
clock, as shown in Figure 4-34 on page 110.
Revision 5
109
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Figure 4-34 • Cascade PLL Configuration
Using internal feedback, we know from EQ 4-1 on page 86 that the maximum achievable output
frequency from the primary output is
fGLA = fCLKA × m / (n × u) = 2 MHz × 128 / (1 × 1) = 256 MHz
EQ 4-5
Figure 4-35 shows the settings of the initial PLL. When configuring the initial PLL, specify the input to be
either Hardwired I/O–Driven or External I/O–Driven. This generates a netlist with the initial PLL routed
from an I/O. Do not specify the input to be Core Logic–Driven, as this prohibits the connection from the
I/O pin to the input of the PLL.
Figure 4-35 • First-Stage PLL Showing Input of 2 MHz and Output of 256 MHz
A second PLL can be connected serially to achieve the required frequency. EQ 4-1 on page 86 to EQ 4-3
on page 86 are extended as follows:
fGLA2 = fGLA × m2 / (n2 × u2) = fCLKA1 × m1 × m2 / (n1 × u1 × n2 × u2) – Primary PLL Output Clock
EQ 4-6
fGLB2 = fYB2 = fCLKA1 × m1 × m2 / (n1 × n2 × v1 × v2) – Secondary 1 PLL Output Clock(s)
EQ 4-7
fGLC2 = fYC2 = fCLKA1 × m1 × m2 / (n1 × n2 × w1 × w2) – Secondary 2 PLL Output Clock(s)
EQ 4-8
In the example, the final output frequency (foutput) from the primary output of the second PLL will be as
follows (EQ 4-9):
foutput = fGLA2 = fGLA × m2 / (n2 × u2) = 256 MHz × 70 / (64 × 1) = 280 MHz
EQ 4-9
Figure 4-36 on page 111 shows the settings of the second PLL. When configuring the second PLL (or
any subsequent-stage PLLs), specify the input to be Core Logic–Driven. This generates a netlist with the
second PLL routed internally from the core. Do not specify the input to be Hardwired I/O–Driven or
External I/O–Driven, as these options prohibit the connection from the output of the first PLL to the input
of the second PLL.
110
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Figure 4-36 • Second-Stage PLL Showing Input of 256 MHz from First Stage and Final Output of 280 MHz
Figure 4-37 shows the simulation results, where the first PLL’s output period is 3.9 ns (~256 MHz), and
the stage 2 (final) output period is 3.56 ns (~280 MHz).
Stage 1 Output Clock Period
Stage 2 Output Clock Period
Figure 4-37 • ModelSim Simulation Results
Revision 5
111
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Recommended Board-Level Considerations
The power to the PLL core is supplied by VCCPLA/B/C/D/E/F (VCCPLx), and the associated ground
connections are supplied by VCOMPLA/B/C/D/E/F (VCOMPLx). When the PLLs are not used, the
Designer place-and-route tool automatically disables the unused PLLs to lower power consumption. The
user should tie unused VCCPLx and VCOMPLx pins to ground. Optionally, the PLL can be turned on/off
during normal device operation via the POWERDOWN port (see Table 4-3 on page 68).
PLL Power Supply Decoupling Scheme
The PLL core is designed to tolerate noise levels on the PLL power supply as specified in the datasheets.
When operated within the noise limits, the PLL will meet the output peak-to-peak jitter specifications
specified in the datasheets. User applications should always ensure the PLL power supply is powered
from a noise-free or low-noise power source.
However, in situations where the PLL power supply noise level is higher than the tolerable limits, various
decoupling schemes can be designed to suppress noise to the PLL power supply. An example is
provided in Figure 4-38. The VCCPLx and VCOMPLx pins correspond to the PLL analog power supply
and ground.
Microsemi strongly recommends that two ceramic capacitors (10 nF in parallel with 100 nF) be placed
close to the power pins (less than 1 inch away). A third generic 10 µF electrolytic capacitor is
recommended for low-frequency noise and should be placed farther away due to its large physical size.
Microsemi recommends that a 6.8 µH inductor be placed between the supply source and the capacitors
to filter out any low-/medium- and high-frequency noise. In addition, the PCB layers should be controlled
so the VCCPLx and VCOMPLx planes have the minimum separation possible, thus generating a goodquality RF capacitor.
For more recommendations, refer to the Board-Level Considerations application note.
Recommended 100 nF capacitor:
•
Producer BC Components, type X7R, 100 nF, 16 V
•
BC Components part number: 0603B104K160BT
•
Digi-Key part number: BC1254CT-ND
•
Digi-Key part number: BC1254TR-ND
Recommended 10 nF capacitor:
•
Surface-mount ceramic capacitor
•
Producer BC Components, type X7R, 10 nF, 50 V
•
BC Components part number: 0603B103K500BT
•
Digi-Key part number: BC1252CT-ND
•
Digi-Key part number: BC1252TR-ND
VCCPLx
IGLOO/e or
ProASIC3/E
Device
10 nF
100 nF
10 μF
Power
Supply
VCOMPLx
Figure 4-38 • Decoupling Scheme for One PLL (should be replicated for each PLL used)
112
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Conclusion
The advanced CCCs of the IGLOO and ProASIC3 devices are ideal for applications requiring precise
clock management. They integrate easily with the internal low-skew clock networks and provide flexible
frequency synthesis, clock deskewing, and/or time-shifting operations.
Related Documents
Application Notes
Board-Level Considerations
http://www.microsemi.com/soc/documents/ALL_AC276_AN.pdf
Datasheets
Fusion Family of Mixed Signal FPGAs
http://www.microsemi.com/soc/documents/Fusion_DS.pdf
User’s Guides
IGLOO, ProASIC3, SmartFusion, and Fusion Macro Library Guide
http://www.microsemi.com/soc/documents/pa3_libguide_ug.pdf
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
August 2012
Changes
Page
The "Implementing EXTFB in ProASIC3/E Devices" section is new (SAR 36647).
70
Table 4-7 • Delay Values in Libero SoC Software per Device Family was added to the
"Clock Delay Adjustment" section (SAR 22709).
86
The "Phase Adjustment" section was rewritten to explain better why the visual
CCC shows both the actual phase and the actual delay that is equivalent to this phase
87
shift (SAR 29647).
The hyperlink for the Board-Level Considerations application note was corrected (SAR 112, 113
36663)
December 2011
Figure 4-20 • PLL Block Diagram, Figure 4-22 • CCC Block Control Bits – Graphical
Representation of Assignments, and Table 4-12 • MUXA, MUXB, MUXC were revised
to change the phase shift assignments for PLLs 4 through 7 (SAR 33791).
June 2011
The description for RESETEN in Table 4-8 • Configuration Bit Descriptions for the
CCC Blocks was revised. The phrase "and should not be modified via dynamic
configuration" was deleted because RESETEN is read only (SAR 25949).
90
July 2010
This chapter is no longer published separately with its own part number and version
but is now part of several FPGA fabric user’s guides.
N/A
Notes were added where appropriate to point out that IGLOO nano and ProASIC3
nano devices do not support differential inputs (SAR 21449).
N/A
Revision 5
85, 89,
93
113
Clock Conditioning Circuits in Low Power Flash Devices and Mixed Signal FPGAs
Date
v1.4
(December 2008)
v1.3
(October 2008)
Changes
Page
The"CCC Support in Microsemi’s Flash Devices" section was updated to include
IGLOO nano and ProASIC3 nano devices.
63
Figure 4-2 • CCC Options: Global Buffers with No Programmable Delay was revised to
add the CLKBIBUF macro.
64
The description of the reference clock was revised in Table 4-2 • Input and Output
Description of the CLKDLY Macro.
65
Figure 4-7 • Clock Input Sources (30 k gates devices and below) is new. Figure 4-8 •
Clock Input Sources Including CLKBUF, CLKBUF_LVDS/LVPECL, and CLKINT (60 k
gates devices and above) applies to 60 k gate devices and above.
72
The "IGLOO and ProASIC3" section was updated to include information for IGLOO
nano devices.
73
A note regarding Fusion CCCs was added to Figure 4-9 • Illustration of Hardwired I/O
(global input pins) Usage for IGLOO and ProASIC3 devices 60 k Gates and Larger
and the name of the figure was changed from Figure 4-8 • Illustration of Hardwired I/O
(global input pins) Usage. Figure 4-10 • Illustration of Hardwired I/O (global input pins)
Usage for IGLOO and ProASIC3 devices 30 k Gates and Smaller is new.
74
Table 4-5 • Number of CCCs by Device Size and Package was updated to include
IGLOO nano and ProASIC3 nano devices. Entries were added to note differences for
the CS81, CS121, and CS201 packages.
78
The "Clock Conditioning Circuits without Integrated PLLs" section was rewritten.
79
The "IGLOO and ProASIC3 CCC Locations" section was updated for nano devices.
81
Figure 4-13 • CCC Locations in the 15 k and 30 k Gate Devices was deleted.
4-20
This document was updated to include Fusion and RT ProASIC3 device information.
Please review the document very carefully.
N/A
The "CCC Support in Microsemi’s Flash Devices" section was updated.
63
In the "Global Buffer with Programmable Delay" section, the following sentence was
changed from:
64
"In this case, the I/O must be placed in one of the dedicated global I/O locations."
To
"In this case, the software will automatically place the dedicated global I/O in the
appropriate locations."
114
Figure 4-4 • CCC Options: Global Buffers with PLL was updated to include OADIVRST
and OADIVHALF.
67
In Figure 4-6 • CCC with PLL Block "fixed delay" was changed to "programmable
delay".
67
Table 4-3 • Input and Output Signals of the PLL Block was updated to include
OADIVRST and OADIVHALF descriptions.
68
Table 4-8 • Configuration Bit Descriptions for the CCC Blocks was updated to include
configuration bits 88 to 81. Note 2 is new. In addition, the description for bit
was updated.
90
Table 4-16 • Fusion Dynamic CCC Clock Source Selection and Table 4-17 • Fusion
Dynamic CCC NGMUX Configuration are new.
94
Table 4-18 • Fusion Dynamic CCC Division by Half Configuration and Table 4-19 •
Configuration Bit / VCOSEL Selection for All Families are new.
95
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Date
v1.2
(June 2008)
Changes
Page
The following changes were made to the family descriptions in Figure 4-1 • Overview
of the CCCs Offered in Fusion, IGLOO, and ProASIC3:
61
• ProASIC3L was updated to include 1.5 V.
• The number of PLLs for ProASIC3E was changed from five to six.
v1.1
(March 2008)
v1.0
(January 2008)
Table 4-1 • Flash-Based FPGAs and the associated text were updated to include the
IGLOO PLUS family. The "IGLOO Terminology" section and "ProASIC3 Terminology"
section are new.
63
The "Global Input Selections" section was updated to include 15 k gate devices as
supported I/O types for globals, for CCC only.
71
Table 4-5 • Number of CCCs by Device Size and Package was revised to include
ProASIC3L, IGLOO PLUS, A3P015, AGL015, AGLP030, AGLP060, and AGLP125.
78
The "IGLOO and ProASIC3 CCC Locations" section was revised to include 15 k gate
devices in the exception statements, as they do not contain PLLs.
81
Information about unlocking the PLL was removed from the "Dynamic PLL
Configuration" section.
87
In the "Dynamic PLL Configuration" section, information was added about running
Layout and determining the exact setting of the ports.
100
In Table 4-8 • Configuration Bit Descriptions for the CCC Blocks, the following bits
were updated to delete "transport to the user" and reference the footnote at the bottom
of the table: 79 to 71.
90
Revision 5
115
5 – FlashROM in Microsemi’s Low Power Flash
Devices
Introduction
The Fusion, IGLOO, and ProASIC3 families of low power flash-based devices have a dedicated
nonvolatile FlashROM memory of 1,024 bits, which provides a unique feature in the FPGA market. The
FlashROM can be read, modified, and written using the JTAG (or UJTAG) interface. It can be read but
not modified from the FPGA core. Only low power flash devices contain on-chip user nonvolatile memory
(NVM).
Architecture of User Nonvolatile FlashROM
Low power flash devices have 1 kbit of user-accessible nonvolatile flash memory on-chip that can be
read from the FPGA core fabric. The FlashROM is arranged in eight banks of 128 bits (16 bytes) during
programming. The 128 bits in each bank are addressable as 16 bytes during the read-back of the
FlashROM from the FPGA core. Figure 5-1 shows the FlashROM logical structure.
The FlashROM can only be programmed via the IEEE 1532 JTAG port. It cannot be programmed directly
from the FPGA core. When programming, each of the eight 128-bit banks can be selectively
reprogrammed. The FlashROM can only be reprogrammed on a bank boundary. Programming involves
an automatic, on-chip bank erase prior to reprogramming the bank. The FlashROM supports
synchronous read. The address is latched on the rising edge of the clock, and the new output data is
stable after the falling edge of the same clock cycle. For more information, refer to the timing diagrams in
the DC and Switching Characteristics chapter of the appropriate datasheet. The FlashROM can be read
on byte boundaries. The upper three bits of the FlashROM address from the FPGA core define the bank
being accessed. The lower four bits of the FlashROM address from the FPGA core define which of the 16
bytes in the bank is being accessed.
Byte Number in Bank
Bank Number 3 MSB of
ADDR (READ)
15
14
13
12
11
4 LSB of ADDR (READ)
10
9
8
7
6
5
4
3
2
1
0
7
6
5
4
3
2
1
0
Figure 5-1 • FlashROM Architecture
Revision 5
117
FlashROM in Microsemi’s Low Power Flash Devices
FlashROM Support in Flash-Based Devices
The flash FPGAs listed in Table 5-1 support the FlashROM feature and the functions described in this
document.
Table 5-1 • Flash-Based FPGAs
Family*
Series
IGLOO
ProASIC3
Fusion
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO nano
The industry’s lowest-power, smallest-size solution
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
ProASIC3
Low power, high-performance 1.5 V FPGAs
ProASIC3E
Higher density ProASIC3 FPGAs with six PLLs and additional I/O standards
ProASIC3 nano
Lowest-cost solution with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V with Flash*Freeze technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Automotive ProASIC3
ProASIC3 FPGAs qualified for automotive applications
Fusion
Mixed signal FPGA integrating ProASIC3 FPGA fabric, programmable
analog block, support for ARM® Cortex™-M1 soft processors, and flash
memory into a monolithic device
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO devices as listed
in Table 5-1. Where the information applies to only one product line or limited devices, these exclusions
will be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 5-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
118
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Bank 0
Bank 1
CCC
SRAM Block
4,608-Bit Dual-Port SRAM
or FIFO Block
OSC
I/Os
CCC/PLL
Bank 2
Bank 4
VersaTile
User Nonvolatile
FlashROM
ISP AES
Decryption
Flash Memory Blocks
Analog
Quad
Analog
Quad
Analog
Quad
Charge Pumps
ADC
Analog
Quad
Analog
Quad
SRAM Block
4,608-Bit Dual-Port SRAM
or FIFO Block
Flash Memory Blocks
Analog
Quad
Analog
Quad
Analog
Quad
Analog
Quad
Analog
Quad
CCC
Bank 3
Figure 5-2 • Fusion Device Architecture Overview (AFS600)
CCC
RAM Block
4,608-Bit Dual-Port SRAM
or FIFO Block
I/Os
VersaTile
RAM Block
4,608-Bit Dual-Port SRAM
or FIFO Block
ISP AES Decryption
Nonvolatile Memory
FlashROM
Charge Pumps
Figure 5-3 • ProASIC3 and IGLOO Device Architecture
Revision 5
119
FlashROM in Microsemi’s Low Power Flash Devices
FlashROM Applications
The SmartGen core generator is used to configure FlashROM content. You can configure each page
independently. SmartGen enables you to create and modify regions within a page; these regions can be
1 to 16 bytes long (Figure 5-4).
Page Number
15
14
13
12
Byte Number in Page
11 10
9
8
7
6
5
4
3
2
1
0
7
6
5
4
3
2
1
0
Figure 5-4 • FlashROM Configuration
The FlashROM content can be changed independently of the FPGA core content. It can be easily
accessed and programmed via JTAG, depending on the security settings of the device. The SmartGen
core generator enables each region to be independently updated (described in the "Programming and
Accessing FlashROM" section on page 122). This enables you to change the FlashROM content on a
per-part basis while keeping some regions "constant" for all parts. These features allow the FlashROM to
be used in diverse system applications. Consider the following possible uses of FlashROM:
120
•
Internet protocol (IP) addressing (wireless or fixed)
•
System calibration settings
•
Restoring configuration after unpredictable system power-down
•
Device serialization and/or inventory control
•
Subscription-based business models (e.g., set-top boxes)
•
Secure key storage
•
Asset management tracking
•
Date stamping
•
Version management
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
FlashROM Security
Low power flash devices have an on-chip Advanced Encryption Standard (AES) decryption core,
combined with an enhanced version of the Microsemi flash-based lock technology (FlashLock®).
Together, they provide unmatched levels of security in a programmable logic device. This security
applies to both the FPGA core and FlashROM content. These devices use the 128-bit AES (Rijndael)
algorithm to encrypt programming files for secure transmission to the on-chip AES decryption core. The
same algorithm is then used to decrypt the programming file. This key size provides approximately 3.4 ×
1038 possible 128-bit keys. A computing system that could find a DES key in a second would take
approximately 149 trillion years to crack a 128-bit AES key. The 128-bit FlashLock feature in low power
flash devices works via a FlashLock security Pass Key mechanism, where the user locks or unlocks the
device with a user-defined key. Refer to the "Security in Low Power Flash Devices" section on page 235.
If the device is locked with certain security settings, functions such as device read, write, and erase are
disabled. This unique feature helps to protect against invasive and noninvasive attacks. Without the
correct Pass Key, access to the FPGA is denied. To gain access to the FPGA, the device first must be
unlocked using the correct Pass Key. During programming of the FlashROM or the FPGA core, you can
generate the security header programming file, which is used to program the AES key and/or FlashLock
Pass Key. The security header programming file can also be generated independently of the FlashROM
and FPGA core content. The FlashLock Pass Key is not stored in the FlashROM.
Low power flash devices with AES-based security allow for secure remote field updates over public
networks such as the Internet, and ensure that valuable intellectual property (IP) remains out of the
hands of IP thieves. Figure 5-5 shows this flow diagram.
Programming
Data
Flash Device
FlashROM
FPGA Core
AES
Encryption
Same AES Key
AES-128
Decryption
Core
Untrusted
Medium
Encrypted Data
Encrypted Data
Figure 5-5 • Programming FlashROM Using AES
Revision 5
121
FlashROM in Microsemi’s Low Power Flash Devices
Programming and Accessing FlashROM
The FlashROM content can only be programmed via JTAG, but it can be read back selectively through
the JTAG programming interface, the UJTAG interface, or via direct FPGA core addressing. The pages of
the FlashROM can be made secure to prevent read-back via JTAG. In that case, read-back on these
secured pages is only possible by the FPGA core fabric or via UJTAG.
A 7-bit address from the FPGA core defines which of the eight pages (three MSBs) is being read, and
which of the 16 bytes within the selected page (four LSBs) are being read. The FlashROM content can
be read on a random basis; the access time is 10 ns for a device supporting commercial specifications.
The FPGA core will be powered down during writing of the FlashROM content. FPGA power-down during
FlashROM programming is managed on-chip, and FPGA core functionality is not available during
programming of the FlashROM. Table 5-2 summarizes various FlashROM access scenarios.
Table 5-2 • FlashROM Read/Write Capabilities by Access Mode
Access Mode
FlashROM Read
FlashROM Write
JTAG
Yes
Yes
UJTAG
Yes
No
FPGA core
Yes
No
Figure 5-6 shows the accessing of the FlashROM using the UJTAG macro. This is similar to FPGA core
access, where the 7-bit address defines which of the eight pages (three MSBs) is being read and which
of the 16 bytes within the selected page (four LSBs) are being read. Refer to the "UJTAG Applications in
Microsemi’s Low Power Flash Devices" section on page 297 for details on using the UJTAG macro to
read the FlashROM.
Figure 5-7 on page 123 and Figure 5-8 on page 123 show the FlashROM access from the JTAG port.
The FlashROM content can be read on a random basis. The three-bit address defines which page is
being read or updated.
UJTAG
Address Generation and
Data Serialization
UIREG [7:0]
Enable
FlashROM
RESET
TDO
TDI
URSTB
UDRUPD
UDRCK
TMS
TCK
TRST
UDRCAP
UDRSH
Control
CLK
SDI
Addr [6:0]
Addr [6:0]
Data [7:0]
Data[7:0]
SDO
UTDI
UTDO
Figure 5-6 • Block Diagram of Using UJTAG to Read FlashROM Contents
122
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
7-Bit Address from Core
1110000
7
6
5
3 MSB of
ADDR (READ)
3-Bit Page Address
Word Number in Page
Page Number
111
4-Bit Word Address
0000
4
3
2
1
0
15
14
13
12
11
4 LSB of ADDR (READ)
10
9
8
7
6
5
4
3
2
1
0
8-Bit Data
8-Bit Data
to FPGA Core
8-Bit Data from Page 7 Word 0
Figure 5-7 • Accessing FlashROM Using FPGA Core
...........................00001:128 Bit Data
To/From JTAG Interface
4-Bit Page Address
from JTAG Interface
Page Number
Word Number in Page
15
14
13
12
11
4 LSB of ADDR (READ)
10
9
8
7
6
5
4
3
2
1
0
3 MSB of
ADDR (READ)
7
6
5
4
3
2
1
0
Figure 5-8 • Accessing FlashROM Using JTAG Port
Revision 5
123
FlashROM in Microsemi’s Low Power Flash Devices
FlashROM Design Flow
The Microsemi Libero System-on-Chip (SoC) software has extensive FlashROM support, including
FlashROM generation, instantiation, simulation, and programming. Figure 5-9 shows the user flow
diagram. In the design flow, there are three main steps:
1. FlashROM generation and instantiation in the design
2. Simulation of FlashROM design
3. Programming file generation for FlashROM design
SmartGen
UFC
File
FlashROM
Netlist
User
Design
MEM
File
Simulator
Synthesis
User
Netlist
BackAnnotated
Netlist
Designer
Core
Map
Security
Header
Options
FlashPoint
Programmer
Programming
Files
Figure 5-9 • FlashROM Design Flow
124
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
FlashROM Generation and Instantiation in the Design
The SmartGen core generator, available in Libero SoC and Designer, is the only tool that can be used to
generate the FlashROM content. SmartGen has several user-friendly features to help generate the
FlashROM contents. Instead of selecting each byte and assigning values, you can create a region within
a page, modify the region, and assign properties to that region. The FlashROM user interface, shown in
Figure 5-10, includes the configuration grid, existing regions list, and properties field. The properties field
specifies the region-specific information and defines the data used for that region. You can assign values
to the following properties:
1. Static Fixed Data—Enables you to fix the data so it cannot be changed during programming time.
This option is useful when you have fixed data stored in this region, which is required for the
operation of the design in the FPGA. Key storage is one example.
2. Static Modifiable Data—Select this option when the data in a particular region is expected to be
static data (such as a version number, which remains the same for a long duration but could
conceivably change in the future). This option enables you to avoid changing the value every time
you enter new data.
3. Read from File—This provides the full flexibility of FlashROM usage to the customer. If you have
a customized algorithm for generating the FlashROM data, you can specify this setting. You can
then generate a text file with data for as many devices as you wish to program, and load that into
the FlashPoint programming file generation software to get programming files that include all the
data. SmartGen will optionally pass the location of the file where the data is stored if the file is
specified in SmartGen. Each text file has only one type of data format (binary, decimal, hex, or
ASCII text). The length of each data file must be shorter than or equal to the selected region
length. If the data is shorter than the selected region length, the most significant bits will be
padded with 0s. For multiple text files for multiple regions, the first lines are for the first device. In
SmartGen, Load Sim. Value From File allows you to load the first device data in the MEM file for
simulation.
4. Auto Increment/Decrement—This scenario is useful when you specify the contents of FlashROM
for a large number of devices in a series. You can specify the step value for the serial number and
a maximum value for inventory control. During programming file generation, the actual number of
devices to be programmed is specified and a start value is fed to the software.
Figure 5-10 • SmartGen GUI of the FlashROM
Revision 5
125
FlashROM in Microsemi’s Low Power Flash Devices
SmartGen allows you to generate the FlashROM netlist in VHDL, Verilog, or EDIF format. After the
FlashROM netlist is generated, the core can be instantiated in the main design like other SmartGen
cores. Note that the macro library name for FlashROM is UFROM. The following is a sample FlashROM
VHDL netlist that can be instantiated in the main design:
library ieee;
use ieee.std_logic_1164.all;
library fusion;
entity FROM_a is
port( ADDR : in std_logic_vector(6 downto 0); DOUT : out std_logic_vector(7 downto 0));
end FROM_a;
architecture DEF_ARCH of
FROM_a is
component UFROM
generic (MEMORYFILE:string);
port(DO0, DO1, DO2, DO3, DO4, DO5, DO6, DO7 : out std_logic;
ADDR0, ADDR1, ADDR2, ADDR3, ADDR4, ADDR5, ADDR6 : in std_logic := 'U') ;
end component;
component GND
port( Y : out std_logic);
end component;
signal U_7_PIN2 : std_logic ;
begin
GND_1_net : GND port map(Y => U_7_PIN2);
UFROM0 : UFROM
generic map(MEMORYFILE => "FROM_a.mem")
port map(DO0 => DOUT(0), DO1 => DOUT(1), DO2 => DOUT(2), DO3 => DOUT(3), DO4 => DOUT(4),
DO5 => DOUT(5), DO6 => DOUT(6), DO7 => DOUT(7), ADDR0 => ADDR(0), ADDR1 => ADDR(1),
ADDR2 => ADDR(2), ADDR3 => ADDR(3), ADDR4 => ADDR(4), ADDR5 => ADDR(5),
ADDR6 => ADDR(6));
end DEF_ARCH;
SmartGen generates the following files along with the netlist. These are located in the SmartGen folder
for the Libero SoC project.
1. MEM (Memory Initialization) file
2. UFC (User Flash Configuration) file
3. Log file
The MEM file is used for simulation, as explained in the "Simulation of FlashROM Design" section on
page 127. The UFC file, generated by SmartGen, has the FlashROM configuration for single or multiple
devices and is used during STAPL generation. It contains the region properties and simulation values.
Note that any changes in the MEM file will not be reflected in the UFC file. Do not modify the UFC to
change FlashROM content. Instead, use the SmartGen GUI to modify the FlashROM content. See the
"Programming File Generation for FlashROM Design" section on page 127 for a description of how the
UFC file is used during the programming file generation. The log file has information regarding the file
type and file location.
126
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Simulation of FlashROM Design
The MEM file has 128 rows of 8 bits, each representing the contents of the FlashROM used for
simulation. For example, the first row represents page 0, byte 0; the next row is page 0, byte 1; and so
the pattern continues. Note that the three MSBs of the address define the page number, and the four
LSBs define the byte number. So, if you send address 0000100 to FlashROM, this corresponds to the
page 0 and byte 4 location, which is the fifth row in the MEM file. SmartGen defaults to 0s for any
unspecified location of the FlashROM. Besides using the MEM file generated by SmartGen, you can
create a binary file with 128 rows of 8 bits each and use this as a MEM file. Microsemi recommends that
you use different file names if you plan to generate multiple MEM files. During simulation, Libero SoC
passes the MEM file used as the generic file in the netlist, along with the design files and testbench. If
you want to use different MEM files during simulation, you need to modify the generic file reference in the
netlist.
…………………
UFROM0: UFROM
--generic map(MEMORYFILE => "F:\Appsnotes\FROM\test_designs\testa\smartgen\FROM_a.mem")
--generic map(MEMORYFILE => "F:\Appsnotes\FROM\test_designs\testa\smartgen\FROM_b.mem")
…………………….
The VITAL and Verilog simulation models accept the generics passed by the netlist, read the MEM file,
and perform simulation with the data in the file.
Programming File Generation for FlashROM Design
FlashPoint is the programming software used to generate the programming files for flash devices.
Depending on the applications, you can use the FlashPoint software to generate a STAPL file with
different FlashROM contents. In each case, optional AES decryption is available. To generate a STAPL
file that contains the same FPGA core content and different FlashROM contents, the FlashPoint software
needs an Array Map file for the core and UFC file(s) for the FlashROM. This final STAPL file represents
the combination of the logic of the FPGA core and FlashROM content.
FlashPoint generates the STAPL files you can use to program the desired FlashROM page and/or FPGA
core of the FPGA device contents. FlashPoint supports the encryption of the FlashROM content and/or
FPGA Array configuration data. In the case of using the FlashROM for device serialization, a sequence
of unique FlashROM contents will be generated. When generating a programming file with multiple
unique FlashROM contents, you can specify in FlashPoint whether to include all FlashROM content in a
single STAPL file or generate a different STAPL file for each FlashROM (Figure 5-11). The programming
software (FlashPro) handles the single STAPL file that contains the FlashROM content from multiple
devices. It enables you to program the FlashROM content into a series of devices sequentially
(Figure 5-11). See the FlashPro User’s Guide for information on serial programming.
UFC File for
Single FlashROM
Contents
FPGA Array
Map File
UFC File for
Multiple FlashROM
Contents
FPGA Array
Map File
FlashPoint
Security Settings
FlashPoint
Security Settings
Single
STAPL
File
Single
STAPL
File
Single
STAPL
File
Figure 5-11 • Single or Multiple Programming File Generation
Revision 5
127
FlashROM in Microsemi’s Low Power Flash Devices
Figure 5-12 shows the programming file generator, which enables different STAPL file generation
methods. When you select Program FlashROM and choose the UFC file, the FlashROM Settings
window appears, as shown in Figure 5-13. In this window, you can select the FlashROM page you want
to program and the data value for the configured regions. This enables you to use a different page for
different programming files.
Figure 5-12 • Programming File Generator
Figure 5-13 • Setting FlashROM during Programming File Generation
The programming hardware and software can load the FlashROM with the appropriate STAPL file.
Programming software handles the single STAPL file that contains multiple FlashROM contents for
multiple devices, and programs the FlashROM in sequential order (e.g., for device serialization). This
feature is supported in the programming software. After programming with the STAPL file, you can run
DEVICE_INFO to check the FlashROM content.
128
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
DEVICE_INFO displays the FlashROM content, serial number, Design Name, and checksum, as shown
below:
EXPORT IDCODE[32] = 123261CF
EXPORT SILSIG[32] = 00000000
User information :
CHECKSUM: 61A0
Design Name:
TOP
Programming Method: STAPL
Algorithm Version: 1
Programmer: UNKNOWN
=========================================
FlashROM Information :
EXPORT Region_7_0[128] = FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
=========================================
Security Setting :
Encrypted FlashROM Programming Enabled.
Encrypted FPGA Array Programming Enabled.
=========================================
The Libero SoC file manager recognizes the UFC and MEM files and displays them in the appropriate
view. Libero SoC also recognizes the multiple programming files if you choose the option to generate
multiple files for multiple FlashROM contents in Designer. These features enable a user-friendly flow for
the FlashROM generation and programming in Libero SoC.
Custom Serialization Using FlashROM
You can use FlashROM for device serialization or inventory control by using the Auto Inc region or Read
From File region. FlashPoint will automatically generate the serial number sequence for the Auto Inc
region with the Start Value, Max Value, and Step Value provided. If you have a unique serial number
generation scheme that you prefer, the Read From File region allows you to import the file with your
serial number scheme programmed into the region. See the FlashPro User's Guide for custom
serialization file format information.
The following steps describe how to perform device serialization or inventory control using FlashROM:
1. Generate FlashROM using SmartGen. From the Properties section in the FlashROM Settings
dialog box, select the Auto Inc or Read From File region. For the Auto Inc region, specify the
desired step value. You will not be able to modify this value in the FlashPoint software.
2. Go through the regular design flow and finish place-and-route.
3. Select Programming File in Designer and open Generate Programming File (Figure 5-12 on
page 128).
4. Click Program FlashROM, browse to the UFC file, and click Next. The FlashROM Settings
window appears, as shown in Figure 5-13 on page 128.
5. Select the FlashROM page you want to program and the data value for the configured regions.
The STAPL file generated will contain only the data that targets the selected FlashROM page.
6. Modify properties for the serialization.
–
For the Auto Inc region, specify the Start and Max values.
–
For the Read From File region, select the file name of the custom serialization file.
7. Select the FlashROM programming file type you want to generate from the two options below:
–
Single programming file for all devices: generates one programming file with all FlashROM
values.
–
One programming file per device: generates a separate programming file for each FlashROM
value.
8. Enter the number of devices you want to program and generate the required programming file.
9. Open the programming software and load the programming file. The programming software,
FlashPro3 and Silicon Sculptor II, supports the device serialization feature. If, for some reason,
the device fails to program a part during serialization, the software allows you to reuse or skip the
serial data. Refer to the FlashPro User’s Guide for details.
Revision 5
129
FlashROM in Microsemi’s Low Power Flash Devices
Conclusion
The Fusion, IGLOO, and ProASIC3 families are the only FPGAs that offer on-chip FlashROM support.
This document presents information on the FlashROM architecture, possible applications, programming,
access through the JTAG and UJTAG interface, and integration into your design. In addition, the Libero
tool set enables easy creation and modification of the FlashROM content.
The nonvolatile FlashROM block in the FPGA can be customized, enabling multiple applications.
Additionally, the security offered by the low power flash devices keeps both the contents of FlashROM
and the FPGA design safe from system over-builders, system cloners, and IP thieves.
Related Documents
User’s Guides
FlashPro User’s Guide
http://www.microsemi.com/documents/FlashPro_UG.pdf
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
Changes
Page
July 2010
This chapter is no longer published separately with its own part number and version
but is now part of several FPGA fabric user’s guides.
N/A
v1.4
(December 2008)
IGLOO nano and ProASIC3 nano devices were added to Table 5-1 • Flash-Based
FPGAs.
118
v1.3
(October 2008)
The "FlashROM Support in Flash-Based Devices" section was revised to include
new families and make the information more concise.
118
Figure 5-2 • Fusion Device Architecture Overview (AFS600) was replaced. 119, 121
Figure 5-5 • Programming FlashROM Using AES was revised to change "Fusion" to
"Flash Device."
v1.2
(June 2008)
v1.1
(March 2008)
130
The FlashPoint User’s Guide was removed from the "User’s Guides" section, as its
content is now part of the FlashPro User’s Guide.
130
The following changes were made to the family descriptions in Table 5-1 • FlashBased FPGAs:
118
•
ProASIC3L was updated to include 1.5 V.
•
The number of PLLs for ProASIC3E was changed from five to six.
The chapter was updated to include the IGLOO PLUS family and information
regarding 15 k gate devices. The "IGLOO Terminology" section and "ProASIC3
Terminology" section are new.
R e vi s i o n 5
N/A
6 – SRAM and FIFO Memories in Microsemi's Low
Power Flash Devices
Introduction
As design complexity grows, greater demands are placed upon an FPGA's embedded memory. Fusion,
IGLOO, and ProASIC3 devices provide the flexibility of true dual-port and two-port SRAM blocks. The
embedded memory, along with built-in, dedicated FIFO control logic, can be used to create cascading
RAM blocks and FIFOs without using additional logic gates.
IGLOO, IGLOO PLUS, and ProASIC3L FPGAs contain an additional feature that allows the device to be
put in a low power mode called Flash*Freeze. In this mode, the core draws minimal power (on the order
of 2 to 127 µW) and still retains values on the embedded SRAM/FIFO and registers. Flash*Freeze
technology allows the user to switch to Active mode on demand, thus simplifying power management
and the use of SRAM/FIFOs.
Device Architecture
The low power flash devices feature up to 504 kbits of RAM in 4,608-bit blocks (Figure 6-1 on page 132
and Figure 6-2 on page 133). The total embedded SRAM for each device can be found in the
datasheets. These memory blocks are arranged along the top and bottom of the device to allow better
access from the core and I/O (in some devices, they are only available on the north side of the device).
Every RAM block has a flexible, hardwired, embedded FIFO controller, enabling the user to implement
efficient FIFOs without sacrificing user gates.
In the IGLOO and ProASIC3 families of devices, the following memories are supported:
•
30 k gate devices and smaller do not support SRAM and FIFO.
•
60 k and 125 k gate devices support memories on the north side of the device only.
•
250 k devices and larger support memories on the north and south sides of the device.
In Fusion devices, the following memories are supported:
•
AFS090 and AFS250 support memories on the north side of the device only.
•
AFS600 and AFS1500 support memories on the north and south sides of the device.
Revision 5
131
SRAM and FIFO Memories in Microsemi's Low Power Flash Devices
Bank 0
Bank 1
Bank 3
CCC
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
I/Os
Bank 1
Bank 3
VersaTile
ISP AES
Decryption 1
User Nonvolatile
FlashRom
Flash*Freeze
Technology 2
Charge
Pumps
Bank 2
Notes:
1. AES decryption not supported in 30 k gate devices and smaller.
2. Flash*Freeze is supported in all IGLOO devices and the ProASIC3L devices.
Figure 6-1 • IGLOO and ProASIC3 Device Architecture Overview
132
R e vi s i o n 5
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
ProASIC3 nano FPGA Fabric User’s Guide
Bank 1
Bank 0
CCC/PLL
RAM Block
4,608-Bit Dual-Port SRAM
or FIFO Block
OSC
I/Os
CCC
Bank 2
Bank 4
VersaTile
ISP AES
Decryption
User Nonvolatile
FlashROM (FROM)
Flash Array
Analog
Quad
Analog
Quad
Analog
Quad
Charge Pumps
ADC
Analog
Quad
Analog
Quad
Analog
Quad
RAM Block
4,608-Bit Dual-Port SRAM
or FIFO Block
Flash Array
Analog
Quad
Analog
Quad
Analog
Quad
Analog
Quad
Bank 3
Figure 6-2 • Fusion Device Architecture Overview (AFS600)
Revision 5
133
SRAM and FIFO Memories in Microsemi's Low Power Flash Devices
SRAM/FIFO Support in Flash-Based Devices
The flash FPGAs listed in Table 6-1 support SRAM and FIFO blocks and the functions described in this
document.
Table 6-1 • Flash-Based FPGAs
Family*
Series
IGLOO
ProASIC3
Fusion
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO nano
The industry’s lowest-power, smallest-size solution
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
ProASIC3
Low power, high-performance 1.5 V FPGAs
ProASIC3E
Higher density ProASIC3 FPGAs with six PLLs and additional I/O standards
ProASIC3 nano
Lowest-cost solution with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V with Flash*Freeze technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Automotive ProASIC3
ProASIC3 FPGAs qualified for automotive applications
Fusion
Mixed signal FPGA integrating ProASIC3 FPGA fabric, programmable
analog block, support for ARM® Cortex™-M1 soft processors, and flash
memory into a monolithic device
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO devices as listed
in Table 6-1. Where the information applies to only one product line or limited devices, these exclusions
will be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 6-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
134
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
SRAM and FIFO Architecture
To meet the needs of high-performance designs, the memory blocks operate strictly in synchronous
mode for both read and write operations. The read and write clocks are completely independent, and
each can operate at any desired frequency up to 250 MHz.
•
4k×1, 2k×2, 1k×4, 512×9 (dual-port RAM—2 read / 2 write or 1 read / 1 write)
•
512×9, 256×18 (2-port RAM—1 read / 1 write)
•
Sync write, sync pipelined / nonpipelined read
Automotive ProASIC3 devices support single-port SRAM capabilities or dual-port SRAM only under
specific conditions. Dual-port mode is supported if the clocks to the two SRAM ports are the same and
180° out of phase (i.e., the port A clock is the inverse of the port B clock). The Libero SoC software
macro libraries support a dual-port macro only. For use of this macro as a single-port SRAM, the inputs
and clock of one port should be tied off (grounded) to prevent errors during design compile. For use in
dual-port mode, the same clock with an inversion between the two clock pins of the macro should be
used in the design to prevent errors during compile.
The memory block includes dedicated FIFO control logic to generate internal addresses and external flag
logic (FULL, EMPTY, AFULL, AEMPTY).
Simultaneous dual-port read/write and write/write operations at the same address are allowed when
certain timing requirements are met.
During RAM operation, addresses are sourced by the user logic, and the FIFO controller is ignored. In
FIFO mode, the internal addresses are generated by the FIFO controller and routed to the RAM array by
internal MUXes.
The low power flash device architecture enables the read and write sizes of RAMs to be organized
independently, allowing for bus conversion. For example, the write size can be set to 256×18 and the
read size to 512×9.
Both the write width and read width for the RAM blocks can be specified independently with the WW
(write width) and RW (read width) pins. The different D×W configurations are 256×18, 512×9, 1k×4,
2k×2, and 4k×1. When widths of one, two, or four are selected, the ninth bit is unused. For example,
when writing nine-bit values and reading four-bit values, only the first four bits and the second four bits of
each nine-bit value are addressable for read operations. The ninth bit is not accessible.
Conversely, when writing four-bit values and reading nine-bit values, the ninth bit of a read operation will
be undefined. The RAM blocks employ little-endian byte order for read and write operations.
Memory Blocks and Macros
Memory blocks can be configured with many different aspect ratios, but are generically supported in the
macro libraries as one of two memory elements: RAM4K9 or RAM512X18. The RAM4K9 is configured
as a true dual-port memory block, and the RAM512X18 is configured as a two-port memory block. Dualport memory allows the RAM to both read from and write to either port independently. Two-port memory
allows the RAM to read from one port and write to the other using a common clock or independent read
and write clocks. If needed, the RAM4K9 blocks can be configured as two-port memory blocks. The
memory block can be configured as a FIFO by combining the basic memory block with dedicated FIFO
controller logic. The FIFO macro is named FIFO4KX18 (Figure 6-3 on page 136).
Clocks for the RAM blocks can be driven by the VersaNet (global resources) or by regular nets. When
using local clock segments, the clock segment region that encompasses the RAM blocks can drive the
RAMs. In the dual-port configuration (RAM4K9), each memory block port can be driven by either risingedge or falling-edge clocks. Each port can be driven by clocks with different edges. Though only a risingedge clock can drive the physical block itself, the Microsemi Designer software will automatically bubblepush the inversion to properly implement the falling-edge trigger for the RAM block.
Revision 5
135
SRAM and FIFO Memories in Microsemi's Low Power Flash Devices
RAM512x18
RAM4K9
ADDRA11
ADDRA10
DOUTA8
DOUTA7
RADDR8
RADDR7
RD17
RD16
ADDRA0
DINA8
DINA7
DOUTA0
RADDR0
RD0
RW1
RW0
DINA0
WIDTHA1
WIDTHA0
PIPEA
WMODEA
BLKA
WENA
CLKA
ADDRB0
DOUTB0
DINB8
DINB7
WIDTHB1
WIDTHB0
PIPEB
WMODEB
BLKB
WENB
CLKB
RESET
RD0
FULL
AFULL
EMPTY
AEMPTY
AFVAL11
AFVAL10
WADDR8
WADDR7
WADDR0
WD17
WD16
WD0
DINB0
RD17
RD16
AEVAL0
REN
RCLK
DOUTB8
DOUTB7
RW2
RW1
RW0
WW2
WW1
WW0
ESTOP
FSTOP
AEVAL11
AEVAL10
PIPE
ADDRB11
ADDRB10
FIFO4K18
AFVAL0
REN
RBLK
RCLK
WD17
WD16
WD0
WW1
WW0
WEN
WBLK
WCLK
RPIPE
WEN
WCLK
RESET
RESET
Notes:
1. Automotive ProASIC3 devices restrict RAM4K9 to a single port or to dual ports with the same clock 180° out of
phase (inverted) between clock pins. In single-port mode, inputs to port B should be tied to ground to prevent
errors during compile. This warning applies only to automotive ProASIC3 parts of certain revisions and earlier.
Contact Technical Support at soc_tech@microsemi.com for information on the revision number for a particular lot
and date code.
2. For FIFO4K18, the same clock 180° out of phase (inverted) between clock pins should be used.
Figure 6-3 • Supported Basic RAM Macros
136
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
SRAM Features
RAM4K9 Macro
RAM4K9 is the dual-port configuration of the RAM block (Figure 6-4). The RAM4K9 nomenclature refers
to both the deepest possible configuration and the widest possible configuration the dual-port RAM block
can assume, and does not denote a possible memory aspect ratio. The RAM block can be configured to
the following aspect ratios: 4,096×1, 2,048×2, 1,024×4, and 512×9. RAM4K9 is fully synchronous and
has the following features:
•
Two ports that allow fully independent reads and writes at different frequencies
•
Selectable pipelined or nonpipelined read
•
Active-low block enables for each port
•
Toggle control between read and write mode for each port
•
Active-low asynchronous reset
•
Pass-through write data or hold existing data on output. In pass-through mode, the data written to
the write port will immediately appear on the read port.
•
Designer software will automatically facilitate falling-edge clocks by bubble-pushing the inversion
to previous stages.
DINA
DOUTA
ADDRA
BLKA
WENA
CLKA
Write Data
Write Data
Read Data
Read Data
Address
BLK
Address
RAM4K9
BLK
WEN
WEN
CLK
CLK
DINB
DOUTB
ADDRB
BLKB
WENB
CLKB
Reset
Note: For timing diagrams of the RAM signals, refer to the appropriate family datasheet.
Figure 6-4 • RAM4K9 Simplified Configuration
Signal Descriptions for RAM4K9
Note: Automotive ProASIC3 devices support single-port SRAM capabilities, or dual-port SRAM
only under specific conditions. Dual-port mode is supported if the clocks to the two SRAM
ports are the same and 180° out of phase (i.e., the port A clock is the inverse of the port B
clock). Since Libero SoC macro libraries support a dual-port macro only, certain
modifications must be made. These are detailed below.
The following signals are used to configure the RAM4K9 memory element:
WIDTHA and WIDTHB
These signals enable the RAM to be configured in one of four allowable aspect ratios (Table 6-2 on
page 138).
Note: When using the SRAM in single-port mode for Automotive ProASIC3 devices, WIDTHB
should be tied to ground.
Revision 5
137
SRAM and FIFO Memories in Microsemi's Low Power Flash Devices
Table 6-2 • Allowable Aspect Ratio Settings for WIDTHA[1:0]
WIDTHA[1:0]
WIDTHB[1:0]
D×W
00
00
4k×1
01
01
2k×2
10
10
1k×4
11
11
512×9
Note: The aspect ratio settings are constant and cannot be changed on the fly.
BLKA and BLKB
These signals are active-low and will enable the respective ports when asserted. When a BLKx signal is
deasserted, that port’s outputs hold the previous value.
Note: When using the SRAM in single-port mode for Automotive ProASIC3 devices, BLKB should
be tied to ground.
WENA and WENB
These signals switch the RAM between read and write modes for the respective ports. A LOW on these
signals indicates a write operation, and a HIGH indicates a read.
Note: When using the SRAM in single-port mode for Automotive ProASIC3 devices, WENB should
be tied to ground.
CLKA and CLKB
These are the clock signals for the synchronous read and write operations. These can be driven
independently or with the same driver.
Note: For Automotive ProASIC3 devices, dual-port mode is supported if the clocks to the two
SRAM ports are the same and 180° out of phase (i.e., the port A clock is the inverse of the
port B clock). For use of this macro as a single-port SRAM, the inputs and clock of one port
should be tied off (grounded) to prevent errors during design compile.
PIPEA and PIPEB
These signals are used to specify pipelined read on the output. A LOW on PIPEA or PIPEB indicates a
nonpipelined read, and the data appears on the corresponding output in the same clock cycle. A HIGH
indicates a pipelined read, and data appears on the corresponding output in the next clock cycle.
Note: When using the SRAM in single-port mode for Automotive ProASIC3 devices, PIPEB should
be tied to ground. For use in dual-port mode, the same clock with an inversion between the
two clock pins of the macro should be used in the design to prevent errors during compile.
WMODEA and WMODEB
These signals are used to configure the behavior of the output when the RAM is in write mode. A LOW
on these signals makes the output retain data from the previous read. A HIGH indicates pass-through
behavior, wherein the data being written will appear immediately on the output. This signal is overridden
when the RAM is being read.
Note: When using the SRAM in single-port mode for Automotive ProASIC3 devices, WMODEB
should be tied to ground.
RESET
This active-low signal resets the control logic, forces the output hold state registers to zero, disables
reads and writes from the SRAM block, and clears the data hold registers when asserted. It does not
reset the contents of the memory array.
While the RESET signal is active, read and write operations are disabled. As with any asynchronous
reset signal, care must be taken not to assert it too close to the edges of active read and write clocks.
ADDRA and ADDRB
These are used as read or write addresses, and they are 12 bits wide. When a depth of less than 4 k is
specified, the unused high-order bits must be grounded (Table 6-3 on page 139).
138
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Note: When using the SRAM in single-port mode for Automotive ProASIC3 devices, ADDRB
should be tied to ground.
Table 6-3 • Address Pins Unused/Used for Various Supported Bus Widths
ADDRx
D×W
Unused
Used
4k×1
None
[11:0]
2k×2
[11]
[10:0]
1k×4
[11:10]
[9:0]
512×9
[11:9]
[8:0]
Note: The "x" in ADDRx implies A or B.
DINA and DINB
These are the input data signals, and they are nine bits wide. Not all nine bits are valid in all
configurations. When a data width less than nine is specified, unused high-order signals must be
grounded (Table 6-4).
Note: When using the SRAM in single-port mode for Automotive ProASIC3 devices, DINB should
be tied to ground.
DOUTA and DOUTB
These are the nine-bit output data signals. Not all nine bits are valid in all configurations. As with DINA
and DINB, high-order bits may not be used (Table 6-4). The output data on unused pins is undefined.
Table 6-4 • Unused/Used Input and Output Data Pins for Various Supported Bus Widths
DINx/DOUTx
D×W
Unused
Used
4k×1
[8:1]
[0]
2k×2
[8:2]
[1:0]
1k×4
[8:4]
[3:0]
512×9
None
[8:0]
Note: The "x" in DINx or DOUTx implies A or B.
RAM512X18 Macro
RAM512X18 is the two-port configuration of the same RAM block (Figure 6-5 on page 140). Like the
RAM4K9 nomenclature, the RAM512X18 nomenclature refers to both the deepest possible configuration
and the widest possible configuration the two-port RAM block can assume. In two-port mode, the RAM
block can be configured to either the 512×9 aspect ratio or the 256×18 aspect ratio. RAM512X18 is also
fully synchronous and has the following features:
•
Dedicated read and write ports
•
Active-low read and write enables
•
Selectable pipelined or nonpipelined read
•
Active-low asynchronous reset
•
Designer software will automatically facilitate falling-edge clocks by bubble-pushing the inversion
to previous stages.
Revision 5
139
SRAM and FIFO Memories in Microsemi's Low Power Flash Devices
Write Data
WD
WADDR
Read Data
Write Address
WEN
WCLK
Read Address
RD
RADDR
RAM512X18
Write Enable
Read Enable
Write CLK
Read CLK
REN
RCLK
Reset
Note: For timing diagrams of the RAM signals, refer to the appropriate family datasheet.
Figure 6-5 • 512X18 Two-Port RAM Block Diagram
Signal Descriptions for RAM512X18
RAM512X18 has slightly different behavior from RAM4K9, as it has dedicated read and write ports.
WW and RW
These signals enable the RAM to be configured in one of the two allowable aspect ratios (Table 6-5).
Table 6-5 • Aspect Ratio Settings for WW[1:0]
WW[1:0]
RW[1:0]
D×W
01
01
512×9
10
10
256×18
00, 11
Reserved
00, 11
WD and RD
These are the input and output data signals, and they are 18 bits wide. When a 512×9 aspect ratio is
used for write, WD[17:9] are unused and must be grounded. If this aspect ratio is used for read, RD[17:9]
are undefined.
WADDR and RADDR
These are read and write addresses, and they are nine bits wide. When the 256×18 aspect ratio is used
for write or read, WADDR[8] and RADDR[8] are unused and must be grounded.
WCLK and RCLK
These signals are the write and read clocks, respectively. They can be clocked on the rising or falling
edge of WCLK and RCLK.
WEN and REN
These signals are the write and read enables, respectively. They are both active-low by default. These
signals can be configured as active-high.
RESET
This active-low signal resets the control logic, forces the output hold state registers to zero, disables
reads and writes from the SRAM block, and clears the data hold registers when asserted. It does not
reset the contents of the memory array.
While the RESET signal is active, read and write operations are disabled. As with any asynchronous
reset signal, care must be taken not to assert it too close to the edges of active read and write clocks.
PIPE
This signal is used to specify pipelined read on the output. A LOW on PIPE indicates a nonpipelined
read, and the data appears on the output in the same clock cycle. A HIGH indicates a pipelined read, and
data appears on the output in the next clock cycle.
140
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
SRAM Usage
The following descriptions refer to the usage of both RAM4K9 and RAM512X18.
Clocking
The dual-port SRAM blocks are only clocked on the rising edge. SmartGen allows falling-edge-triggered
clocks by adding inverters to the netlist, hence achieving dual-port SRAM blocks that are clocked on
either edge (rising or falling). For dual-port SRAM, each port can be clocked on either edge and by
separate clocks by port. Note that for Automotive ProASIC3, the same clock, with an inversion between
the two clock pins of the macro, should be used in design to prevent errors during compile.
Low power flash devices support inversion (bubble-pushing) throughout the FPGA architecture, including
the clock input to the SRAM modules. Inversions added to the SRAM clock pin on the design schematic
or in the HDL code will be automatically accounted for during design compile without incurring additional
delay in the clock path.
The two-port SRAM can be clocked on the rising or falling edge of WCLK and RCLK.
If negative-edge RAM and FIFO clocking is selected for memory macros, clock edge inversion
management (bubble-pushing) is automatically used within the development tools, without performance
penalty.
Modes of Operation
There are two read modes and one write mode:
•
Read Nonpipelined (synchronous—1 clock edge): In the standard read mode, new data is driven
onto the RD bus in the same clock cycle following RA and REN valid. The read address is
registered on the read port clock active edge, and data appears at RD after the RAM access time.
Setting PIPE to OFF enables this mode.
•
Read Pipelined (synchronous—2 clock edges): The pipelined mode incurs an additional clock
delay from address to data but enables operation at a much higher frequency. The read address
is registered on the read port active clock edge, and the read data is registered and appears at
RD after the second read clock edge. Setting PIPE to ON enables this mode.
•
Write (synchronous—1 clock edge): On the write clock active edge, the write data is written into
the SRAM at the write address when WEN is HIGH. The setup times of the write address, write
enables, and write data are minimal with respect to the write clock.
RAM Initialization
Each SRAM block can be individually initialized on power-up by means of the JTAG port using the UJTAG
mechanism. The shift register for a target block can be selected and loaded with the proper bit
configuration to enable serial loading. The 4,608 bits of data can be loaded in a single operation.
FIFO Features
The FIFO4KX18 macro is created by merging the RAM block with dedicated FIFO logic (Figure 6-6 on
page 142). Since the FIFO logic can only be used in conjunction with the memory block, there is no
separate FIFO controller macro. As with the RAM blocks, the FIFO4KX18 nomenclature does not refer to
a possible aspect ratio, but rather to the deepest possible data depth and the widest possible data width.
FIFO4KX18 can be configured into the following aspect ratios: 4,096×1, 2,048×2, 1,024×4, 512×9, and
256×18. In addition to being fully synchronous, the FIFO4KX18 also has the following features:
•
Four FIFO flags: Empty, Full, Almost-Empty, and Almost-Full
•
Empty flag is synchronized to the read clock
•
Full flag is synchronized to the write clock
•
Both Almost-Empty and Almost-Full flags have programmable thresholds
•
Active-low asynchronous reset
•
Active-low block enable
•
Active-low write enable
•
Active-high read enable
•
Ability to configure the FIFO to either stop counting after the empty or full states are reached or to
allow the FIFO counters to continue
Revision 5
141
SRAM and FIFO Memories in Microsemi's Low Power Flash Devices
•
Designer software will automatically facilitate falling-edge clocks by bubble-pushing the inversion
to previous stages.
Write Data
WD
Full Flag
FULL
AFULL
Read Data
Almost-Full Flag
FIFO4KX18
AEMPTY
Read Enable
Write Clock
WCLK
EMPTY
Almost-Empty Flag
Write Enable
WEN
RD
Empty Flag
REN
Read Clock
RCLK
Reset
Figure 6-6 • FIFO4KX18 Block Diagram
RD[17:0]
RCLK
WCLK
RCLK
WCLK
RADD[J:0]
WADD[J:0]
FREN
RBLK
REN
REN
WEN
FWEN
RD
RAM
RW[2:0]
WW[2:0]
WD[17:0]
RPIPE
WD
CNT 12
E
=
ESTOP
AFVAL
FULL
AFULL
AEMPTY
WBLK
WEN
CNT 12
SUB 12
AEVAL
E
=
FSTOP
EMPTY
Reset
Figure 6-7 • RAM Block with Embedded FIFO Controller
The FIFOs maintain a separate read and write address. Whenever the difference between the write
address and the read address is greater than or equal to the almost-full value (AFVAL), the Almost-Full
flag is asserted. Similarly, the Almost-Empty flag is asserted whenever the difference between the write
address and read address is less than or equal to the almost-empty value (AEVAL).
Due to synchronization between the read and write clocks, the Empty flag will deassert after the second
read clock edge from the point that the write enable asserts. However, since the Empty flag is
synchronized to the read clock, it will assert after the read clock reads the last data in the FIFO. Also,
since the Full flag is dependent on the actual hardware configuration, it will assert when the actual
physical implementation of the FIFO is full.
For example, when a user configures a 128×18 FIFO, the actual physical implementation will be a
256×18 FIFO element. Since the actual implementation is 256×18, the Full flag will not trigger until the
142
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
256×18 FIFO is full, even though a 128×18 FIFO was requested. For this example, the Almost-Full flag
can be used instead of the Full flag to signal when the 128th data word is reached.
To accommodate different aspect ratios, the almost-full and almost-empty values are expressed in terms
of data bits instead of data words. SmartGen translates the user’s input, expressed in data words, into
data bits internally. SmartGen allows the user to select the thresholds for the Almost-Empty and AlmostFull flags in terms of either the read data words or the write data words, and makes the appropriate
conversions for each flag.
After the empty or full states are reached, the FIFO can be configured so the FIFO counters either stop or
continue counting. For timing numbers, refer to the appropriate family datasheet.
Signal Descriptions for FIFO4K18
The following signals are used to configure the FIFO4K18 memory element:
WW and RW
These signals enable the FIFO to be configured in one of the five allowable aspect ratios (Table 6-6).
Table 6-6 • Aspect Ratio Settings for WW[2:0]
WW[2:0]
RW[2:0]
D×W
000
000
4k×1
001
001
2k×2
010
010
1k×4
011
011
512×9
100
100
256×18
101, 110, 111
Reserved
101, 110, 111
WBLK and RBLK
These signals are active-low and will enable the respective ports when LOW. When the RBLK signal is
HIGH, that port’s outputs hold the previous value.
WEN and REN
Read and write enables. WEN is active-low and REN is active-high by default. These signals can be
configured as active-high or -low.
WCLK and RCLK
These are the clock signals for the synchronous read and write operations. These can be driven
independently or with the same driver.
Note: For the Automotive ProASIC3 FIFO4K18, for the same clock, 180° out of phase (inverted)
between clock pins should be used.
RPIPE
This signal is used to specify pipelined read on the output. A LOW on RPIPE indicates a nonpipelined
read, and the data appears on the output in the same clock cycle. A HIGH indicates a pipelined read, and
data appears on the output in the next clock cycle.
RESET
This active-low signal resets the control logic and forces the output hold state registers to zero when
asserted. It does not reset the contents of the memory array (Table 6-7 on page 144).
While the RESET signal is active, read and write operations are disabled. As with any asynchronous
RESET signal, care must be taken not to assert it too close to the edges of active read and write clocks.
WD
This is the input data bus and is 18 bits wide. Not all 18 bits are valid in all configurations. When a data
width less than 18 is specified, unused higher-order signals must be grounded (Table 6-7 on page 144).
Revision 5
143
SRAM and FIFO Memories in Microsemi's Low Power Flash Devices
RD
This is the output data bus and is 18 bits wide. Not all 18 bits are valid in all configurations. Like the WD
bus, high-order bits become unusable if the data width is less than 18. The output data on unused pins is
undefined (Table 6-7).
Table 6-7 • Input Data Signal Usage for Different Aspect Ratios
D×W
WD/RD Unused
4k×1
WD[17:1], RD[17:1]
2k×2
WD[17:2], RD[17:2]
1k×4
WD[17:4], RD[17:4]
512×9
WD[17:9], RD[17:9]
256×18
–
ESTOP, FSTOP
ESTOP is used to stop the FIFO read counter from further counting once the FIFO is empty (i.e., the
EMPTY flag goes HIGH). A HIGH on this signal inhibits the counting.
FSTOP is used to stop the FIFO write counter from further counting once the FIFO is full (i.e., the FULL
flag goes HIGH). A HIGH on this signal inhibits the counting.
For more information on these signals, refer to the "ESTOP and FSTOP Usage" section.
FULL, EMPTY
When the FIFO is full and no more data can be written, the FULL flag asserts HIGH. The FULL flag is
synchronous to WCLK to inhibit writing immediately upon detection of a full condition and to prevent
overflows. Since the write address is compared to a resynchronized (and thus time-delayed) version of
the read address, the FULL flag will remain asserted until two WCLK active edges after a read operation
eliminates the full condition.
When the FIFO is empty and no more data can be read, the EMPTY flag asserts HIGH. The EMPTY flag
is synchronous to RCLK to inhibit reading immediately upon detection of an empty condition and to
prevent underflows. Since the read address is compared to a resynchronized (and thus time-delayed)
version of the write address, the EMPTY flag will remain asserted until two RCLK active edges after a
write operation removes the empty condition.
For more information on these signals, refer to the "FIFO Flag Usage Considerations" section on
page 145.
AFULL, AEMPTY
These are programmable flags and will be asserted on the threshold specified by AFVAL and AEVAL,
respectively.
When the number of words stored in the FIFO reaches the amount specified by AEVAL while reading,
the AEMPTY output will go HIGH. Likewise, when the number of words stored in the FIFO reaches the
amount specified by AFVAL while writing, the AFULL output will go HIGH.
AFVAL, AEVAL
The AEVAL and AFVAL pins are used to specify the almost-empty and almost-full threshold values. They
are 12-bit signals. For more information on these signals, refer to the "FIFO Flag Usage Considerations"
section on page 145.
FIFO Usage
ESTOP and FSTOP Usage
The ESTOP pin is used to stop the read counter from counting any further once the FIFO is empty (i.e.,
the EMPTY flag goes HIGH). Likewise, the FSTOP pin is used to stop the write counter from counting
any further once the FIFO is full (i.e., the FULL flag goes HIGH).
The FIFO counters in the device start the count at zero, reach the maximum depth for the configuration
(e.g., 511 for a 512×9 configuration), and then restart at zero. An example application for ESTOP, where
the read counter keeps counting, would be writing to the FIFO once and reading the same content over
and over without doing another write.
144
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
FIFO Flag Usage Considerations
The AEVAL and AFVAL pins are used to specify the 12-bit AEMPTY and AFULL threshold values. The
FIFO contains separate 12-bit write address (WADDR) and read address (RADDR) counters. WADDR is
incremented every time a write operation is performed, and RADDR is incremented every time a read
operation is performed. Whenever the difference between WADDR and RADDR is greater than or equal
to AFVAL, the AFULL output is asserted. Likewise, whenever the difference between WADDR and
RADDR is less than or equal to AEVAL, the AEMPTY output is asserted. To handle different read and
write aspect ratios, AFVAL and AEVAL are expressed in terms of total data bits instead of total data
words. When users specify AFVAL and AEVAL in terms of read or write words, the SmartGen tool
translates them into bit addresses and configures these signals automatically. SmartGen configures the
AFULL flag to assert when the write address exceeds the read address by at least a predefined value. In
a 2k×8 FIFO, for example, a value of 1,500 for AFVAL means that the AFULL flag will be asserted after a
write when the difference between the write address and the read address reaches 1,500 (there have
been at least 1,500 more writes than reads). It will stay asserted until the difference between the write
and read addresses drops below 1,500.
The AEMPTY flag is asserted when the difference between the write address and the read address is
less than a predefined value. In the example above, a value of 200 for AEVAL means that the AEMPTY
flag will be asserted when a read causes the difference between the write address and the read address
to drop to 200. It will stay asserted until that difference rises above 200. Note that the FIFO can be
configured with different read and write widths; in this case, the AFVAL setting is based on the number of
write data entries, and the AEVAL setting is based on the number of read data entries. For aspect ratios
of 512×9 and 256×18, only 4,096 bits can be addressed by the 12 bits of AFVAL and AEVAL. The
number of words must be multiplied by 8 and 16 instead of 9 and 18. The SmartGen tool automatically
uses the proper values. To avoid halfwords being written or read, which could happen if different read
and write aspect ratios were specified, the FIFO will assert FULL or EMPTY as soon as at least one word
cannot be written or read. For example, if a two-bit word is written and a four-bit word is being read, the
FIFO will remain in the empty state when the first word is written. This occurs even if the FIFO is not
completely empty, because in this case, a complete word cannot be read. The same is applicable in the
full state. If a four-bit word is written and a two-bit word is read, the FIFO is full and one word is read. The
FULL flag will remain asserted because a complete word cannot be written at this point.
Variable Aspect Ratio and Cascading
Variable aspect ratio and cascading allow users to configure the memory in the width and depth required.
The memory block can be configured as a FIFO by combining the basic memory block with dedicated
FIFO controller logic. The FIFO macro is named FIFO4KX18. Low power flash device RAM can be
configured as 1, 2, 4, 9, or 18 bits wide. By cascading the memory blocks, any multiple of those widths
can be created. The RAM blocks can be from 256 to 4,096 bits deep, depending on the aspect ratio, and
the blocks can also be cascaded to create deeper areas. Refer to the aspect ratios available for each
macro cell in the "SRAM Features" section on page 137. The largest continuous configurable memory
area is equal to half the total memory available on the device, because the RAM is separated into two
groups, one on each side of the device.
The SmartGen core generator will automatically configure and cascade both RAM and FIFO blocks.
Cascading is accomplished using dedicated memory logic and does not consume user gates for depths
up to 4,096 bits deep and widths up to 18, depending on the configuration. Deeper memory will utilize
some user gates to multiplex the outputs.
Generated RAM and FIFO macros can be created as either structural VHDL or Verilog for easy
instantiation into the design. Users of Libero SoC can create a symbol for the macro and incorporate it
into a design schematic.
Table 6-10 on page 147 shows the number of memory blocks required for each of the supported depth
and width memory configurations, and for each depth and width combination. For example, a 256-bit
deep by 32-bit wide two-port RAM would consist of two 256×18 RAM blocks. The first 18 bits would be
stored in the first RAM block, and the remaining 14 bits would be implemented in the other 256×18 RAM
block. This second RAM block would have four bits of unused storage. Similarly, a dual-port memory
block that is 8,192 bits deep and 8 bits wide would be implemented using 16 memory blocks. The dualport memory would be configured in a 4,096×1 aspect ratio. These blocks would then be cascaded two
deep to achieve 8,192 bits of depth, and eight wide to achieve the eight bits of width.
Revision 5
145
SRAM and FIFO Memories in Microsemi's Low Power Flash Devices
Table 6-8 and Table 6-9 show the maximum potential width and depth configuration for each device. Note
that 15 k and 30 k gate devices do not support RAM or FIFO.
Table 6-8 • Memory Availability per IGLOO and ProASIC3 Device
Maximum Potential Width1
Device
IGLOO
IGLOO nano
IGLOO PLUS
RAM
ProASIC3
ProASIC3 nano Block
s
ProASIC3L
Maximum Potential Depth2
Depth
Width
Depth
Width
AGL060
AGLN060
AGLP060
A3P060
A3PN060
4
256
72 (4×18)
16,384 (4,096×4)
1
AGL125
AGLN125
AGLP125
A3P125
A3PN125
8
256
144 (8×18)
32,768 (4,094×8)
1
AGL250
AGLN250
A3P250/L
A3PN250
8
256
144 (8×18)
32,768 (4,096×8)
1
AGL400
A3P400
12
256
216 (12×18)
49,152 (4,096×12)
1
AGL600
A3P600/L
24
256
432 (24×18)
98,304 (4,096×24)
1
AGL1000
A3P1000/L
32
256
576 (32×18)
131,072 (4,096×32)
1
AGLE600
AGLE3000
A3PE600
24
256
432 (24×18)
98,304 (4,096×24)
1
A3PE1500
60
256
1,080 (60×18)
245,760 (4,096×60)
1
A3PE3000/L
112
256
2,016 (112×18)
458,752 (4,096×112)
1
Notes:
1. Maximum potential width uses the two-port configuration.
2. Maximum potential depth uses the dual-port configuration.
Table 6-9 • Memory Availability per Fusion Device
Maximum Potential Width1
Device
RAM Blocks
Depth
Width
Depth
Width
AFS090
6
256
108 (6×18)
24,576 (4,094×6)
1
AFS250
8
256
144 (8×18)
32,768 (4,094×8)
1
AFS600
24
256
432 (24×18)
98,304 (4,096×24)
1
AFS1500
60
256
1,080 (60×18)
245,760 (4,096×60)
1
Notes:
1. Maximum potential width uses the two-port configuration.
2. Maximum potential depth uses the dual-port configuration.
146
Maximum Potential Depth2
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Table 6-10 • RAM and FIFO Memory Block Consumption
Depth
256
Two-Port
1
2
4
8
9
Width
16
18
32
36
64
72
Note:
Dual-Port
512
1,024
2,048
4,096
8,192
16,384
32,768
65,536
Dual-Port
Dual-Port
Dual-Port
Dual-Port
Dual-Port
Dual-Port
Dual-Port
Dual-Port
Number Block
1
1
1
1
1
1
2
4
8
16 × 1
Configuration
Any
Any
Any
1,024 × 4
2,048 × 2
4,096 × 1
2 × (4,096 × 1)
Cascade Deep
4 × (4,096 × 1)
Cascade Deep
8 × (4,096 × 1)
Cascade Deep
16 × (4,096 × 1)
Cascade Deep
4
8
16
32
Number Block
1
1
1
1
1
2
Configuration
Any
Any
Any
1,024×4
2,048 × 2
2 × (4,096 × 1)
Cascaded Wide
Number Block
1
1
1
1
2
4
Configuration
Any
Any
Any
1,024 × 4
2 × (2,048 × 2)
Cascaded Wide
4 × (4,096 × 1)
Cascaded Wide
Number Block
1
1
1
2
4
8
Configuration
Any
Any
Any
2 × (1,024 × 4)
Cascaded Wide
4 × (2,048 × 2)
Cascaded Wide
8 × (4,096 × 1)
Cascaded Wide
4 × (4,096 × 1)
8 × (4,096 × 1)
16 × (4,096 × 1) 32 × (4,096 × 1)
Cascaded 2 Deep Cascaded 4 Deep Cascaded 8 Deep
Cascaded 16
and 2 Wide
and 2 Wide
and 2 Wide
Deep and 2 Wide
8
16
16
32
1
1
1
2
4
8
16
32
Configuration
Any
Any
Any
2 × (512 × 9)
Cascaded Deep
4 × (512 × 9)
Cascaded Deep
8 × (512 × 9)
Cascaded Deep
16 × (512 × 9)
Cascaded Deep
32 × (512 × 9)
Cascaded Deep
16
32
64
1
1
1
4
8
Configuration
256 × 18
256 × 18
256 × 18
4 × (1,024 × 4)
Cascaded Wide
8 × (2,048 × 2)
Cascaded Wide
4
8
Number Block
1
2
2
Configuration
256 × 8
2 × (512 × 9)
Cascaded Wide
2 × (512 × 9)
Cascaded Wide
2
4
4
8
16
Configuration
2 × (256 × 18)
Cascaded Wide
4 × (512 × 9)
Cascaded Wide
4 × (512 × 9)
Cascaded Wide
8 × (1,024 × 4)
Cascaded Wide
16 × (2,048 × 2)
Cascaded Wide
8
16
2
4
4
Configuration
2 × (256 × 18)
Cascaded Wide
4 × (512 × 9)
Cascaded Wide
4 × (512 × 9)
Cascaded Wide
64
16 × (4,096 × 1) 32 × (4,096 × 1) 32 × (4,096 × 1)
Cascaded Wide Cascaded 2 Deep Cascaded 4 Deep
and 16 Wide
and 16 Wide
18
32
4 × (512 × 9)
8 × (512 × 9)
16 × (512 × 9)
16 × (512 × 9)
Cascaded 2 Deep Cascaded 4 Deep Cascaded 8 Deep
Cascaded 16
and 2 Wide
and 2 Wide
and 2 Wide
Deep and 2 Wide
Number Block
Number Block
64
16 × (4,096 × 1) 32 × (4,096 × 1) 64 × (4,096 × 1)
Cascaded 2 Deep Cascaded 4 Deep Cascaded 8 Deep
and 8 Wide
and 8 Wide
and 8 Wide
Number Block
Number Block
32
4 × (4,096 × 1)
16 × (4,096 × 1) 32 × (4,096 × 1) 64 × (4,096 × 1)
Cascaded 2 Deep Cascaded 4 Deep Cascaded 8 Deep
Cascaded 16
and 4 Wide
and 4 Wide
and 4 Wide
Deep and 4 Wide
32
64
32 × (4,096 × 1) 64 × (4,096 × 1)
Cascaded Wide Cascaded 2 Deep
and 32 Wide
32
4 × (512 × 9)
16 × (512 × 9)
16 × (512 × 9)
Cascaded 2 Deep Cascaded 4 Deep Cascaded 8 Deep
and 4 Wide
and 4 Wide
and 4 Wide
Number Block
4
8
8
16
32
64
Configuration
4 × (256 × 18)
Cascaded Wide
8 × (512 × 9)
Cascaded Wide
8 × (512 × 9)
Cascaded Wide
16 × (1,024 × 4)
Cascaded Wide
32 × (2,048 × 2)
Cascaded Wide
64 × (4,096 × 1)
Cascaded Wide
Number Block
4
8
8
16
32
Configuration
4 × (256 × 18)
Cascaded Wide
8 × (512 × 9)
Cascaded Wide
8 × (512 × 9)
Cascaded Wide
16 × (512 × 9)
Cascaded Wide
16 × (512 × 9)
Cascaded 4 Deep
and 8 Wide
Memory configurations represented by grayed cells are not supported.
R e visi on 5
147
SRAM and FIFO Memories in Microsemi's Low Power Flash Devices
Initializing the RAM/FIFO
The SRAM blocks can be initialized with data to use as a lookup table (LUT). Data initialization can be
accomplished either by loading the data through the design logic or through the UJTAG interface. The
UJTAG macro is used to allow access from the JTAG port to the internal logic in the device. By sending
the appropriate initialization string to the JTAG Test Access Port (TAP) Controller, the designer can put
the JTAG circuitry into a mode that allows the user to shift data into the array logic through the JTAG port
using the UJTAG macro. For a more detailed explanation of the UJTAG macro, refer to the "FlashROM in
Microsemi’s Low Power Flash Devices" section on page 117.
A user interface is required to receive the user command, initialization data, and clock from the UJTAG
macro. The interface must synchronize and load the data into the correct RAM block of the design. The
main outputs of the user interface block are the following:
•
Memory block chip select: Selects a memory block for initialization. The chip selects signals for
each memory block that can be generated from different user-defined pockets or simple logic,
such as a ring counter (see below).
•
Memory block write address: Identifies the address of the memory cell that needs to be initialized.
•
Memory block write data: The interface block receives the data serially from the UTDI port of the
UJTAG macro and loads it in parallel into the write data ports of the memory blocks.
•
Memory block write clock: Drives the WCLK of the memory block and synchronizes the write
data, write address, and chip select signals.
Figure 6-8 shows the user interface between UJTAG and the memory blocks.
RAM1
WD
WADDR
WCLK
UJTAG
TRST
TDO
TDI
TRST
TDO
TDI
TMS
TMS
TCK
TCK
User Interface
IR[7:0]
Reset
UIREG[7:0]
URSTB
UDRUPD
DR_UPDATE
UDRSH
UDRCAP
UDRCK
UTDI
UTDO
DR_SHIFT
DR_CAPTURE
DR_CLK
DIN
DOUT
WDATA
WADDR
WCLK
WEN1
WEN
RAM2
WD
WADDR
WCLK
WEN2
WEN
WEN3
RAM3
WD
WADDR
WCLK
WEN
Figure 6-8 • Interfacing TAP Ports and SRAM Blocks
An important component of the interface between the UJTAG macro and the RAM blocks is a serialin/parallel-out shift register. The width of the shift register should equal the data width of the RAM blocks.
The RAM data arrives serially from the UTDI output of the UJTAG macro. The data must be shifted into a
shift register clocked by the JTAG clock (provided at the UDRCK output of the UJTAG macro).
Then, after the shift register is fully loaded, the data must be transferred to the write data port of the RAM
block. To synchronize the loading of the write data with the write address and write clock, the output of
the shift register can be pipelined before driving the RAM block.
The write address can be generated in different ways. It can be imported through the TAP using a
different instruction opcode and another shift register, or generated internally using a simple counter.
Using a counter to generate the address bits and sweep through the address range of the RAM blocks is
148
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
recommended, since it reduces the complexity of the user interface block and the board-level JTAG
driver.
Moreover, using an internal counter for address generation speeds up the initialization procedure, since
the user only needs to import the data through the JTAG port.
The designer may use different methods to select among the multiple RAM blocks. Using counters along
with demultiplexers is one approach to set the write enable signals. Basically, the number of RAM blocks
needing initialization determines the most efficient approach. For example, if all the blocks are initialized
with the same data, one enable signal is enough to activate the write procedure for all of them at the
same time. Another alternative is to use different opcodes to initialize each memory block. For a small
number of RAM blocks, using counters is an optimal choice. For example, a ring counter can be used to
select from multiple RAM blocks. The clock driver of this counter needs to be controlled by the address
generation process.
Once the addressing of one block is finished, a clock pulse is sent to the (ring) counter to select the next
memory block.
Figure 6-9 illustrates a simple block diagram of an interface block between UJTAG and RAM blocks.
Serial-to-Port Shift Register
UTDI
UDRSH
UDRCK
SIN
POUT
Data Reg.
D
Q
n
n
Enable
CLK
SOUT
CLK
UTDO
UDRUPDI
UIREG
WDATA
WCLK
In
Compare Result
with
Defined Opcode
En
Reset
CLK
URSTB
Chip Select
WEN1
Ring
Counter
WEN2
WENi
m
En
Reset
CLK
Addr Counter
Q
m
WADDR
Binary
Counter
Figure 6-9 • Block Diagram of a Sample User Interface
In the circuit shown in Figure 6-9, the shift register is enabled by the UDRSH output of the UJTAG macro.
The counters and chip select outputs are controlled by the value of the TAP Instruction Register. The
comparison block compares the UIREG value with the "start initialization" opcode value (defined by the
user). If the result is true, the counters start to generate addresses and activate the WEN inputs of
appropriate RAM blocks.
The UDRUPD output of the UJTAG macro, also shown in Figure 6-9, is used for generating the write
clock (WCLK) and synchronizing the data register and address counter with WCLK. UDRUPD is HIGH
when the TAP Controller is in the Data Register Update state, which is an indication of completing the
loading of one data word. Once the TAP Controller goes into the Data Register Update state, the
UDRUPD output of the UJTAG macro goes HIGH. Therefore, the pipeline register and the address
counter place the proper data and address on the outputs of the interface block. Meanwhile, WCLK is
defined as the inverted UDRUPD. This will provide enough time (equal to the UDRUPD HIGH time) for
the data and address to be placed at the proper ports of the RAM block before the rising edge of WCLK.
The inverter is not required if the RAM blocks are clocked at the falling edge of the write clock. An
example of this is described in the "Example of RAM Initialization" section on page 150.
Revision 5
149
SRAM and FIFO Memories in Microsemi's Low Power Flash Devices
Example of RAM Initialization
This section of the document presents a sample design in which a 4×4 RAM block is being initialized
through the JTAG port. A test feature has been implemented in the design to read back the contents of
the RAM after initialization to verify the procedure.
The interface block of this example performs two major functions: initialization of the RAM block and
running a test procedure to read back the contents. The clock output of the interface is either the write
clock (for initialization) or the read clock (for reading back the contents). The Verilog code for the
interface block is included in the "Sample Verilog Code" section on page 151.
For simulation purposes, users can declare the input ports of the UJTAG macro for easier assignment in
the testbench. However, the UJTAG input ports should not be declared on the top level during synthesis.
If the input ports of the UJTAG are declared during synthesis, the synthesis tool will instantiate input
buffers on these ports. The input buffers on the ports will cause Compile to fail in Designer.
Figure 6-10 shows the simulation results for the initialization step of the example design.
The CLK_OUT signal, which is the clock output of the interface block, is the inverted DR_UPDATE output
of the UJTAG macro. It is clear that it gives sufficient time (while the TAP Controller is in the Data
Register Update state) for the write address and data to become stable before loading them into the RAM
block.
Figure 6-11 presents the test procedure of the example. The data read back from the memory block
matches the written data, thus verifying the design functionality.
Figure 6-10 • Simulation of Initialization Step
Figure 6-11 • Simulation of the Test Procedure of the Example
150
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
The ROM emulation application is based on RAM block initialization. If the user's main design has
access only to the read ports of the RAM block (RADDR, RD, RCLK, and REN), and the contents of the
RAM are already initialized through the TAP, then the memory blocks will emulate ROM functionality for
the core design. In this case, the write ports of the RAM blocks are accessed only by the user interface
block, and the interface is activated only by the TAP Instruction Register contents.
Users should note that the contents of the RAM blocks are lost in the absence of applied power.
However, the 1 kbit of flash memory, FlashROM, in low power flash devices can be used to retain data
after power is removed from the device. Refer to the "SRAM and FIFO Memories in Microsemi's Low
Power Flash Devices" section on page 131 for more information.
Sample Verilog Code
Interface Block
`define Initialize_start 8'h22 //INITIALIZATION START COMMAND VALUE
`define Initialize_stop 8'h23 //INITIALIZATION START COMMAND VALUE
module interface(IR, rst_n, data_shift, clk_in, data_update, din_ser, dout_ser, test,
test_out,test_clk,clk_out,wr_en,rd_en,write_word,read_word,rd_addr, wr_addr);
input [7:0] IR;
input [3:0] read_word; //RAM DATA READ BACK
input rst_n, data_shift, clk_in, data_update, din_ser; //INITIALIZATION SIGNALS
input test, test_clk; //TEST PROCEDURE CLOCK AND COMMAND INPUT
output [3:0] test_out; //READ DATA
output [3:0] write_word; //WRITE DATA
output [1:0] rd_addr; //READ ADDRESS
output [1:0] wr_addr; //WRITE ADDRESS
output dout_ser; //TDO DRIVER
output clk_out, wr_en, rd_en;
wire
wire
wire
wire
wire
[3:0] write_word;
[1:0] rd_addr;
[1:0] wr_addr;
[3:0] Q_out;
enable, test_active;
reg clk_out;
//SELECT CLOCK FOR INITIALIZATION OR READBACK TEST
always @(enable or test_clk or data_update)
begin
case ({test_active})
1 : clk_out = test_clk ;
0 : clk_out = !data_update;
default : clk_out = 1'b1;
endcase
end
assign
assign
assign
assign
assign
assign
test_active = test && (IR == 8'h23);
enable = (IR == 8'h22);
wr_en = !enable;
rd_en = !test_active;
test_out = read_word;
dout_ser = Q_out[3];
//4-bit SIN/POUT SHIFT REGISTER
shift_reg data_shift_reg (.Shiften(data_shift), .Shiftin(din_ser), .Clock(clk_in),
.Q(Q_out));
//4-bit PIPELINE REGISTER
D_pipeline pipeline_reg (.Data(Q_out), .Clock(data_update), .Q(write_word));
Revision 5
151
SRAM and FIFO Memories in Microsemi's Low Power Flash Devices
//
addr_counter counter_1 (.Clock(data_update), .Q(wr_addr), .Aset(rst_n),
.Enable(enable));
addr_counter counter_2 (.Clock(test_clk), .Q(rd_addr), .Aset(rst_n),
.Enable( test_active));
endmodule
Interface Block / UJTAG Wrapper
This example is a sample wrapper, which connects the interface block to the UJTAG and the memory
blocks.
// WRAPPER
module top_init (TDI, TRSTB, TMS, TCK, TDO, test, test_clk, test_ out);
input TDI, TRSTB, TMS, TCK;
output TDO;
input test, test_clk;
output [3:0] test_out;
wire
wire
wire
wire
wire
[7:0] IR;
reset, DR_shift, DR_cap, init_clk, DR_update, data_in, data_out;
clk_out, wen, ren;
[3:0] word_in, word_out;
[1:0] write_addr, read_addr;
UJTAG UJTAG_U1 (.UIREG0(IR[0]), .UIREG1(IR[1]), .UIREG2(IR[2]), .UIREG3(IR[3]),
.UIREG4(IR[4]), .UIREG5(IR[5]), .UIREG6(IR[6]), .UIREG7(IR[7]), .URSTB(reset),
.UDRSH(DR_shift), .UDRCAP(DR_cap), .UDRCK(init_clk), .UDRUPD(DR_update),
.UT-DI(data_in), .TDI(TDI), .TMS(TMS), .TCK(TCK), .TRSTB(TRSTB), .TDO(TDO),
.UT-DO(data_out));
mem_block RAM_block (.DO(word_out), .RCLOCK(clk_out), .WCLOCK(clk_out), .DI(word_in),
.WRB(wen), .RDB(ren), .WAD-DR(write_addr), .RADDR(read_addr));
interface init_block (.IR(IR), .rst_n(reset), .data_shift(DR_shift), .clk_in(init_clk),
.data_update(DR_update), .din_ser(data_in), .dout_ser(data_out), .test(test),
.test_out(test_out), .test_clk(test_clk), .clk_out(clk_out), .wr_en(wen),
.rd_en(ren), .write_word(word_in), .read_word(word_out), .rd_addr(read_addr),
.wr_addr(write_addr));
endmodule
Address Counter
module addr_counter (Clock, Q, Aset, Enable);
input Clock;
output [1:0] Q;
input Aset;
input Enable;
reg [1:0] Qaux;
always @(posedge Clock or negedge Aset)
begin
if (!Aset) Qaux PAD, Y => Y);
DDR_REG_0_inst : DDR_REG
port map(D => Y, CLK => CLK, CLR => CLR, QR => QR, QF => QF);
end DEF_ARCH;
Revision 5
211
DDR for Microsemi’s Low Power Flash Devices
DDR Output Register
DDR_OUT
DR
DF
DataR
DataF
OUTBUF_SSTL3_I
Q
D
PAD
CLK
CLR
CLR
Figure 9-6 • DDR Output Register (SSTL3 Class I)
Verilog
module DDR_OutBuf_SSTL3_I(DataR,DataF,CLR,CLK,PAD);
input
output
DataR, DataF, CLR, CLK;
PAD;
wire Q, VCC;
VCC VCC_1_net(.Y(VCC));
DDR_OUT DDR_OUT_0_inst(.DR(DataR),.DF(DataF),.CLK(CLK),.CLR(CLR),.Q(Q));
OUTBUF_SSTL3_I OUTBUF_SSTL3_I_0_inst(.D(Q),.PAD(PAD));
endmodule
VHDL
library ieee;
use ieee.std_logic_1164.all;
library proasic3; use proasic3.all;
entity DDR_OutBuf_SSTL3_I is
port(DataR, DataF, CLR, CLK : in std_logic;
end DDR_OutBuf_SSTL3_I;
architecture DEF_ARCH of
PAD : out std_logic) ;
DDR_OutBuf_SSTL3_I is
component DDR_OUT
port(DR, DF, CLK, CLR : in std_logic := 'U'; Q : out std_logic) ;
end component;
component OUTBUF_SSTL3_I
port(D : in std_logic := 'U'; PAD : out std_logic) ;
end component;
component VCC
port( Y : out std_logic);
end component;
signal Q, VCC_1_net : std_logic ;
begin
VCC_2_net : VCC port map(Y => VCC_1_net);
DDR_OUT_0_inst : DDR_OUT
port map(DR => DataR, DF => DataF, CLK => CLK, CLR => CLR, Q => Q);
OUTBUF_SSTL3_I_0_inst : OUTBUF_SSTL3_I
port map(D => Q, PAD => PAD);
end DEF_ARCH;
212
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
DDR Tristate Output Register
INV
A
Y
TrienAux
Trien
DDR_OUT
DataR
DataF
DR
DF
Q
D
PAD
TRIBUFF_F_8U
CLK
CLR
CLR
Figure 9-7 • DDR Tristate Output Register, LOW Enable, 8 mA, Pull-Up (LVTTL)
Verilog
module DDR_TriStateBuf_LVTTL_8mA_HighSlew_LowEnb_PullUp(DataR, DataF, CLR, CLK, Trien,
PAD);
input
output
DataR, DataF, CLR, CLK, Trien;
PAD;
wire TrienAux, Q;
INV Inv_Tri(.A(Trien),.Y(TrienAux));
DDR_OUT DDR_OUT_0_inst(.DR(DataR),.DF(DataF),.CLK(CLK),.CLR(CLR),.Q(Q));
TRIBUFF_F_8U TRIBUFF_F_8U_0_inst(.D(Q),.E(TrienAux),.PAD(PAD));
endmodule
VHDL
library ieee;
use ieee.std_logic_1164.all;
library proasic3; use proasic3.all;
entity DDR_TriStateBuf_LVTTL_8mA_HighSlew_LowEnb_PullUp is
port(DataR, DataF, CLR, CLK, Trien : in std_logic; PAD : out std_logic) ;
end DDR_TriStateBuf_LVTTL_8mA_HighSlew_LowEnb_PullUp;
architecture DEF_ARCH of DDR_TriStateBuf_LVTTL_8mA_HighSlew_LowEnb_PullUp is
component INV
port(A : in std_logic := 'U'; Y : out std_logic) ;
end component;
component DDR_OUT
port(DR, DF, CLK, CLR : in std_logic := 'U'; Q : out std_logic) ;
end component;
component TRIBUFF_F_8U
port(D, E : in std_logic := 'U'; PAD : out std_logic) ;
end component;
signal TrienAux, Q : std_logic ;
begin
Inv_Tri : INV
port map(A => Trien, Y => TrienAux);
Revision 5
213
DDR for Microsemi’s Low Power Flash Devices
DDR_OUT_0_inst : DDR_OUT
port map(DR => DataR, DF => DataF, CLK => CLK, CLR => CLR, Q => Q);
TRIBUFF_F_8U_0_inst : TRIBUFF_F_8U
port map(D => Q, E => TrienAux, PAD => PAD);
end DEF_ARCH;
DDR Bidirectional Buffer
Trien
INV
A
Y
E
DDR_OUT
DataR
DataF
DR
DF
Q
CLK
CLR
D
PAD
BIBUF_HSTL_I
CLR
QR
QF
DDR_REG
QR
D
Y
QF
CLR
Figure 9-8 • DDR Bidirectional Buffer, LOW Output Enable (HSTL Class II)
Verilog
module DDR_BiDir_HSTL_I_LowEnb(DataR,DataF,CLR,CLK,Trien,QR,QF,PAD);
input
output
inout
DataR, DataF, CLR, CLK, Trien;
QR, QF;
PAD;
wire TrienAux, D, Q;
INV Inv_Tri(.A(Trien), .Y(TrienAux));
DDR_OUT DDR_OUT_0_inst(.DR(DataR),.DF(DataF),.CLK(CLK),.CLR(CLR),.Q(Q));
DDR_REG DDR_REG_0_inst(.D(D),.CLK(CLK),.CLR(CLR),.QR(QR),.QF(QF));
BIBUF_HSTL_I BIBUF_HSTL_I_0_inst(.PAD(PAD),.D(Q),.E(TrienAux),.Y(D));
endmodule
214
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
VHDL
library ieee;
use ieee.std_logic_1164.all;
library proasic3; use proasic3.all;
entity DDR_BiDir_HSTL_I_LowEnb is
port(DataR, DataF, CLR, CLK, Trien : in std_logic; QR, QF : out std_logic;
PAD : inout std_logic) ;
end DDR_BiDir_HSTL_I_LowEnb;
architecture DEF_ARCH of
DDR_BiDir_HSTL_I_LowEnb is
component INV
port(A : in std_logic := 'U'; Y : out std_logic) ;
end component;
component DDR_OUT
port(DR, DF, CLK, CLR : in std_logic := 'U'; Q : out std_logic) ;
end component;
component DDR_REG
port(D, CLK, CLR : in std_logic := 'U'; QR, QF : out std_logic) ;
end component;
component BIBUF_HSTL_I
port(PAD : inout std_logic := 'U'; D, E : in std_logic := 'U'; Y : out std_logic) ;
end component;
signal TrienAux, D, Q : std_logic ;
begin
Inv_Tri : INV
port map(A => Trien, Y => TrienAux);
DDR_OUT_0_inst : DDR_OUT
port map(DR => DataR, DF => DataF, CLK => CLK, CLR => CLR, Q => Q);
DDR_REG_0_inst : DDR_REG
port map(D => D, CLK => CLK, CLR => CLR, QR => QR, QF => QF);
BIBUF_HSTL_I_0_inst : BIBUF_HSTL_I
port map(PAD => PAD, D => Q, E => TrienAux, Y => D);
end DEF_ARCH;
Revision 5
215
DDR for Microsemi’s Low Power Flash Devices
Design Example
Figure 9-9 shows a simple example of a design using both DDR input and DDR output registers. The
user can copy the HDL code in Libero SoC software and go through the design flow. Figure 9-10 and
Figure 9-11 on page 217 show the netlist and ChipPlanner views of the ddr_test design. Diagrams may
vary slightly for different families.
PAD
INBUF_SSTL2_I DDR_REG
Y
PAD
D
QR
CLK
QF
DataR
DDR_OUT
DR
DF
DataF
CLR
CLR
CLR
Figure 9-9 • Design Example
Figure 9-10 • DDR Test Design as Seen by NetlistViewer for IGLOO/e Devices
216
R e vi s i o n 5
Q
OUTBUF_SSTL3_I
D
PAD
ProASIC3 nano FPGA Fabric User’s Guide
Figure 9-11 • DDR Input/Output Cells as Seen by ChipPlanner for IGLOO/e Devices
Verilog
module Inbuf_ddr(PAD,CLR,CLK,QR,QF);
input PAD, CLR, CLK;
output QR, QF;
wire Y;
DDR_REG DDR_REG_0_inst(.D(Y), .CLK(CLK), .CLR(CLR), .QR(QR), .QF(QF));
INBUF INBUF_0_inst(.PAD(PAD), .Y(Y));
endmodule
module Outbuf_ddr(DataR,DataF,CLR,CLK,PAD);
input DataR, DataF, CLR, CLK;
output PAD;
wire Q, VCC;
VCC VCC_1_net(.Y(VCC));
DDR_OUT DDR_OUT_0_inst(.DR(DataR), .DF(DataF), .CLK(CLK), .CLR(CLR), .Q(Q));
OUTBUF OUTBUF_0_inst(.D(Q), .PAD(PAD));
endmodule
Revision 5
217
DDR for Microsemi’s Low Power Flash Devices
module ddr_test(DIN, CLK, CLR, DOUT);
input DIN, CLK, CLR;
output DOUT;
Inbuf_ddr Inbuf_ddr (.PAD(DIN), .CLR(clr), .CLK(clk), .QR(qr), .QF(qf));
Outbuf_ddr Outbuf_ddr (.DataR(qr),.DataF(qf), .CLR(clr), .CLK(clk),.PAD(DOUT));
INBUF INBUF_CLR (.PAD(CLR), .Y(clr));
INBUF INBUF_CLK (.PAD(CLK), .Y(clk));
endmodule
Simulation Consideration
Microsemi DDR simulation models use inertial delay modeling by default (versus transport delay
modeling). As such, pulses that are shorter than the actual gate delays should be avoided, as they will
not be seen by the simulator and may be an issue in post-routed simulations. The user must be aware of
the default delay modeling and must set the correct delay model in the simulator as needed.
Conclusion
Fusion, IGLOO, and ProASIC3 devices support a wide range of DDR applications with different I/O
standards and include built-in DDR macros. The powerful capabilities provided by SmartGen and its GUI
can simplify the process of including DDR macros in designs and minimize design errors. Additional
considerations should be taken into account by the designer in design floorplanning and placement of I/O
flip-flops to minimize datapath skew and to help improve system timing margins. Other system-related
issues to consider include PLL and clock partitioning.
218
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
July 2010
v1.4
(December 2008)
Changes
Page
This chapter is no longer published separately with its own part number and version
but is now part of several FPGA fabric user’s guides.
N/A
Notes were added where appropriate to point out that IGLOO nano and ProASIC3
nano devices do not support differential inputs (SAR 21449).
N/A
IGLOO nano and ProASIC3 nano devices were added to Table 9-1 • Flash-Based
FPGAs.
206
The "I/O Cell Architecture" section was updated with information applicable to nano
devices.
207
The output buffer (OUTBUF_SSTL3_I) input was changed to D, instead of Q, in
205,
Figure 9-1 • DDR Support in Low Power Flash Devices, Figure 9-3 • DDR Output
209,
Register (SSTL3 Class I), Figure 9-6 • DDR Output Register (SSTL3 Class I), 212, 213
Figure 9-7 • DDR Tristate Output Register, LOW Enable, 8 mA, Pull-Up (LVTTL), and
the output from the DDR_OUT macro was connected to the input of the TRIBUFF
macro in Figure 9-7 • DDR Tristate Output Register, LOW Enable, 8 mA, Pull-Up
(LVTTL).
v1.3
(October 2008)
v1.2
(June 2008)
The "Double Data Rate (DDR) Architecture" section was updated to include mention
of the AFS600 and AFS1500 devices.
205
The "DDR Support in Flash-Based Devices" section was revised to include new
families and make the information more concise.
206
The following changes were made to the family descriptions in Table 9-1 • FlashBased FPGAs:
206
•
•
v1.1
(March 2008)
ProASIC3L was updated to include 1.5 V.
The number of PLLs for ProASIC3E was changed from five to six.
The "IGLOO Terminology" section and "ProASIC3 Terminology" section are new.
Revision 5
206
219
10 – Programming Flash Devices
Introduction
This document provides an overview of the various programming options available for the Microsemi
flash families. The electronic version of this document includes active links to all programming resources,
which are available at http://www.microsemi.com/soc/products/hardware/default.aspx. For Microsemi
antifuse devices, refer to the Programming Antifuse Devices document.
Summary of Programming Support
FlashPro4 and FlashPro3 are high-performance in-system programming (ISP) tools targeted at the latest
generation of low power flash devices offered by the SmartFusion,® Fusion, IGLOO,® and ProASIC®3
families, including ARM-enabled devices. FlashPro4 and FlashPro3 offer extremely high performance
through the use of USB 2.0, are high-speed compliant for full use of the 480 Mbps bandwidth, and can
program ProASIC3 devices in under 30 seconds. Powered exclusively via USB, FlashPro4 and
FlashPro3 provide a VPUMP voltage of 3.3 V for programming these devices.
FlashPro4 replaced FlashPro3 in 2010. FlashPro4 supports SmartFusion, Fusion, ProASIC3,and IGLOO
devices as well as future generation flash devices. FlashPro4 also adds 1.2 V programming for IGLOO
nano V2 devices. FlashPro4 is compatible with FlashPro3; however it adds a programming mode
(PROG_MODE) signal to the previously unused pin 4 of the JTAG connector. The PROG_MODE goes
high during programming and can be used to turn on a 1.5 V external supply for those devices that
require 1.5 V for programming. If both FlashPro3 and FlashPro4 programmers are used for programming
the same boards, pin 4 of the JTAG connector must not be connected to anything on the board because
FlashPro4 uses pin 4 for PROG_MODE.
FlashPro3 or
FlashPro4
FlashPro
Software
JTAG
ProASIC3/E
Programming File:
PDB, STP, or FDB
Figure 10-1 • FlashPro Programming Setup
Revision 5
221
Programming Flash Devices
Programming Support in Flash Devices
The flash FPGAs listed in Table 10-1 support flash in-system programming and the functions described
in this document.
Table 10-1 • Flash-Based FPGAs
Family*
Series
IGLOO
ProASIC3
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO nano
The industry’s lowest-power, smallest-size solution, supporting 1.2 V to 1.5 V
core voltage with Flash*Freeze technology
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
ProASIC3
Low power, high-performance 1.5 V FPGAs
ProASIC3E
Higher density ProASIC3 FPGAs with six PLLs and additional I/O standards
ProASIC3 nano
Lowest-cost solution with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V core voltage with Flash*Freeze
technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Automotive ProASIC3
ProASIC3 FPGAs qualified for automotive applications
SmartFusion SmartFusion
Mixed-signal FPGA integrating FPGA fabric, programmable microcontroller
subsystem (MSS), including programmable analog and ARM® Cortex™-M3
hard processor and flash memory in a monolithic device
Fusion
Fusion
Mixed signal FPGA integrating ProASIC3 FPGA fabric, programmable
analog block, support for ARM® Cortex™-M1 soft processors, and flash
memory into a monolithic device
ProASIC
ProASIC
First generation ProASIC devices
ProASICPLUS
Second generation ProASIC devices
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO devices as listed
in Table 10-1. Where the information applies to only one product line or limited devices, these exclusions
will be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 10-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
222
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
General Flash Programming Information
Programming Basics
When choosing a programming solution, there are a number of options available. This section provides a
brief overview of those options. The next sections provide more detail on those options as they apply to
Microsemi FPGAs.
Reprogrammable or One-Time-Programmable (OTP)
Depending on the technology chosen, devices may be reprogrammable or one-time-programmable. As
the name implies, a reprogrammable device can be programmed many times. Generally, the contents of
such a device will be completely overwritten when it is reprogrammed. All Microsemi flash devices are
reprogrammable.
An OTP device is programmable one time only. Once programmed, no more changes can be made to
the contents. Microsemi flash devices provide the option of disabling the reprogrammability for security
purposes. This combines the convenience of reprogrammability during design verification with the
security of an OTP technology for highly sensitive designs.
Device Programmer or In-System Programming
There are two fundamental ways to program an FPGA: using a device programmer or, if the technology
permits, using in-system programming. A device programmer is a piece of equipment in a lab or on the
production floor that is used for programming FPGA devices. The devices are placed into a socket
mounted in a programming adapter module, and the appropriate electrical interface is applied. The
programmed device can then be placed on the board. A typical programmer, used during development,
programs a single device at a time and is referred to as a single-site engineering programmer.
With ISP, the device is already mounted onto the system printed circuit board when programming occurs.
Typically, ISD programming is performed via a JTAG interface on the FPGA. The JTAG pins can be
controlled either by an on-board resource, such as a microprocessor, or by an off-board programmer
through a header connection. Once mounted, it can be programmed repeatedly and erased. If the
application requires it, the system can be designed to reprogram itself using a microprocessor, without
the use of any external programmer.
If multiple devices need to be programmed with the same program, various multi-site programming
hardware is available in order to program many devices in parallel. Microsemi In House Programming is
also available for this purpose.
Programming Features for Microsemi Devices
Flash Devices
The flash devices supplied by Microsemi are reprogrammable by either a generic device programmer or
ISP. Microsemi supports ISP using JTAG, which is supported by the FlashPro4 and FlashPro3, FlashPro
Lite, Silicon Sculptor 3, and Silicon Sculptor II programmers.
Levels of ISP support vary depending on the device chosen:
•
All SmartFusion, Fusion, IGLOO, and ProASIC3 devices support ISP.
•
IGLOO, IGLOOe, IGLOO nano V5, and IGLOO PLUS devices can be programmed in-system
when the device is using a 1.5 V supply voltage to the FPGA core.
•
IGLOO nano V2 devices can be programmed at 1.2 V core voltage (when using FlashPro4 only)
or 1.5 V. IGLOO nano V5 devices are programmed with a VCC core voltage of 1.5 V.
Revision 5
223
Programming Flash Devices
Types of Programming for Flash Devices
The number of devices to be programmed will influence the optimal programming methodology. Those
available are listed below:
•
•
•
In-system programming
–
Using a programmer
–
Using a microprocessor or microcontroller
Device programmers
–
Single-site programmers
–
Multi-site programmers, batch programmers, or gang programmers
–
Automated production (robotic) programmers
Volume programming services
–
Microsemi in-house programming
–
Programming centers
In-System Programming
Device Type Supported: Flash
ISP refers to programming the FPGA after it has been mounted on the system printed circuit board. The
FPGA may be preprogrammed and later reprogrammed using ISP.
The advantage of using ISP is the ability to update the FPGA design many times without any changes to
the board. This eliminates the requirement of using a socket for the FPGA, saving cost and improving
reliability. It also reduces programming hardware expenses, as the ISP methodology is die-/packageindependent.
There are two methods of in-system programming: external and internal.
•
Programmer ISP—Refer to the "In-System Programming (ISP) of Microsemi’s Low Power Flash
Devices Using FlashPro4/3/3X" section on page 261 for more information.
Using an external programmer and a cable, the device can be programmed through a header on
the system board. In Microsemi SoC Products Group documentation, this is referred to as
external ISP. Microsemi provides FlashPro4, FlashPro3, FlashPro Lite, or Silicon Sculptor 3 to
perform external ISP. Note that Silicon Sculptor II and Silicon Sculptor 3 can only provide ISP for
ProASIC and ProASICPLUS® families, not for SmartFusion, Fusion, IGLOO, or ProASIC3. Silicon
Sculptor II and Silicon Sculptor 3 can be used for programming ProASIC and ProASICPLUS
devices by using an adapter module (part number SMPA-ISP-ACTEL-3).
•
–
Advantages: Allows local control of programming and data files for maximum security. The
programming algorithms and hardware are available from Microsemi. The only hardware
required on the board is a programming header.
–
Limitations: A negligible board space requirement for the programming header and JTAG
signal routing
Microprocessor ISP—Refer to the "Microprocessor Programming of Microsemi’s Low Power
Flash Devices" chapter of an appropriate FPGA fabric user’s guide for more information.
Using a microprocessor and an external or internal memory, you can store the program in
memory and use the microprocessor to perform the programming. In Microsemi documentation,
this is referred to as internal ISP. Both the code for the programming algorithm and the FPGA
programming file must be stored in memory on the board. Programming voltages must also be
generated on the board.
224
–
Advantages: The programming code is stored in the system memory. An external programmer
is not required during programming.
–
Limitations: This is the approach that requires the most design work, since some way of
getting and/or storing the data is needed; a system interface to the device must be designed;
and the low-level API to the programming firmware must be written and linked into the code
provided by Microsemi. While there are benefits to this methodology, serious thought and
planning should go into the decision.
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Device Programmers
Single Device Programmer
Single device programmers are used to program a device before it is mounted on the system board.
The advantage of using device programmers is that no programming hardware is required on the system
board. Therefore, no additional components or board space are required.
Adapter modules are purchased with single device programmers to support the FPGA packages used.
The FPGA is placed in the adapter module and the programming software is run from a PC. Microsemi
supplies the programming software for all of the Microsemi programmers. The software allows for the
selection of the correct die/package and programming files. It will then program and verify the device.
•
Single-site programmers
A single-site programmer programs one device at a time. Microsemi offers Silicon Sculptor 3, built
by BP Microsystems, as a single-site programmer. Silicon Sculptor 3 and associated software are
available only from Microsemi.
•
–
Advantages: Lower cost than multi-site programmers. No additional overhead for
programming on the system board. Allows local control of programming and data files for
maximum security. Allows on-demand programming on-site.
–
Limitations: Only programs one device at a time.
Multi-site programmers
Often referred to as batch or gang programmers, multi-site programmers can program multiple devices at
the same time using the same programming file. This is often used for large volume programming and by
programming houses. The sites often have independent processors and memory enabling the sites to
operate concurrently, meaning each site may start programming the same file independently. This
enables the operator to change one device while the other sites continue programming, which increases
throughput. Multiple adapter modules for the same package are required when using a multi-site
programmer. Silicon Sculptor I, II, and 3 programmers can be cascaded to program multiple devices in a
chain. Multi-site programmers, such as the BP2610 and BP2710, can also be purchased from BP
Microsystems. When using BP Microsystems multi-site programmers, users must use programming
adapter modules available only from Microsemi. Visit the Microsemi SoC Products Group website to view
the part numbers of the desired adapter module:
http://www.microsemi.com/soc/products/hardware/program_debug/ss/modules.aspx.
Also when using BP Microsystems programmers, customers must use Microsemi
programming software to ensure the best programming result will occur.
•
–
Advantages: Provides the capability of programming multiple devices at the same time. No
additional overhead for programming on the system board. Allows local control of
programming and data files for maximum security.
–
Limitations: More expensive than a single-site programmer
Automated production (robotic) programmers
Automated production programmers are based on multi-site programmers. They consist of a large input
tray holding multiple parts and a robotic arm to select and place parts into appropriate programming
sockets automatically. When the programming of the parts is complete, the parts are removed and
placed in a finished tray. The automated programmers are often used in volume programming houses to
program parts for which the programming time is small. BP Microsystems part number BP4710, BP4610,
BP3710 MK2, and BP3610 are available for this purpose. Auto programmers cannot be used to program
RTAX-S devices.
Where an auto-programmer is used, the appropriate open-top adapter module from BP Microsystems
must be used.
Revision 5
225
Programming Flash Devices
Volume Programming Services
Device Type Supported: Flash and Antifuse
Once the design is stable for applications with large production volumes, preprogrammed devices can be
purchased. Table 10-2 describes the volume programming services.
Table 10-2 • Volume Programming Services
Programmer
In-House Programming
Distributor Programming Centers
Independent Programming Centers
Vendor
Availability
Microsemi
Contact Microsemi Sales
Memec Unique
Contact Distribution
Various
Contact Vendor
Advantages: As programming is outsourced, this solution is easier to implement than creating a
substantial in-house programming capability. As programming houses specialize in large-volume
programming, this is often the most cost-effective solution.
Limitations: There are some logistical issues with the use of a programming service provider, such as the
transfer of programming files and the approval of First Articles. By definition, the programming file must
be released to a third-party programming house. Nondisclosure agreements (NDAs) can be signed to
help ensure data protection; however, for extremely security-conscious designs, this may not be an
option.
•
Microsemi In-House Programming
When purchasing Microsemi devices in volume, IHP can be requested as part of the purchase. If
this option is chosen, there is a small cost adder for each device programmed. Each device is
marked with a special mark to distinguish it from blank parts. Programming files for the design will
be sent to Microsemi. Sample parts with the design programmed, First Articles, will be returned
for customer approval. Once approval of First Articles has been received, Microsemi will proceed
with programming the remainder of the order. To request Microsemi IHP, contact your local
Microsemi representative.
•
Distributor Programming Centers
If purchases are made through a distributor, many distributors will provide programming for their
customers. Consult with your preferred distributor about this option.
226
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Programming Solutions
Details for the available programmers can be found in the programmer user's guides listed in the
"Related Documents" section on page 231.
All the programmers except FlashPro4, FlashPro3, FlashPro Lite, and FlashPro require adapter
modules, which are designed to support device packages. All modules are listed on the Microsemi SoC
Products Group website at
http://www.microsemi.com/soc/products/hardware/program_debug/ss/modules.aspx. They are not listed
in this document, since this list is updated frequently with new package options and any upgrades
required to improve programming yield or support new families.
Table 10-3 • Programming Solutions
Programmer
FlashPro4
FlashPro3
FlashPro
Lite2
Vendor
Microsemi
Microsemi
Microsemi
ISP
Single
Device
Only
Only
Only
Multi-Device
Availability
Yes
Yes
1
Available
Yes
Yes1
Available
Yes
Yes1
Available
Discontinued
Microsemi
Only
Yes
Yes1
Silicon Sculptor 3
Microsemi
Yes3
Yes
Cascade option
(up to two)
Available
Silicon Sculptor II
Microsemi
Yes3
Yes
Cascade option
(up to two)
Available
Silicon Sculptor
Microsemi
Yes
Yes
Cascade option
(up to four)
Discontinued
Sculptor 6X
Microsemi
No
Yes
Yes
Discontinued
BP
Microsystems
No
Yes
Yes
Contact BP
Microsystems at
www.bpmicro.com
FlashPro
BP MicroProgrammers
Notes:
1. Multiple devices can be connected in the same JTAG chain for programming.
2. If FlashPro Lite is used for programming, the programmer derives all of its power from the target pc
board's VDD supply. The FlashPro Lite's VPP and VPN power supplies use the target pc board's
VDD as a power source. The target pc board must supply power to both the VDDP and VDD power
pins of the ProASICPLUS device in addition to supplying VDD to the FlashPro Lite. The target pc
board needs to provide at least 500 mA of current to the FlashPro Lite VDD connection for
programming.
3. Silicon Sculptor II and Silicon Sculptor 3 can only provide ISP for ProASIC and ProASICPLUS
families, not for Fusion, IGLOO, or ProASIC3 devices.
Revision 5
227
Programming Flash Devices
Programmer Ordering Codes
The products shown in Table 10-4 can be ordered through Microsemi sales and will be shipped directly
from Microsemi. Products can also be ordered from Microsemi distributors, but will still be shipped
directly from Microsemi. Table 10-4 includes ordering codes for the full kit, as well as codes for
replacement items and any related hardware. Some additional products can be purchased from external
suppliers for use with the programmers. Ordering codes for adapter modules used with Silicon Sculptor
are available at http://www.microsemi.com/soc/products/hardware/program_debug/ss/modules.aspx.
Table 10-4 • Programming Ordering Codes
Description
Vendor
Ordering Code
Comment
FlashPro4 ISP
programmer
Microsemi
FLASHPRO 4
Uses a 2×5, RA male header connector
FlashPro Lite ISP
programmer
Microsemi
FLASHPRO LITE
Supports small programming header or
large header through header converter
(not included)
Silicon Sculptor 3
Microsemi
SILICON-SCULPTOR 3
USB 2.0 high-speed production
programmer
Silicon Sculptor II
Microsemi
SILICON-SCULPTOR II
Requires add-on adapter modules to
support devices
Silicon Sculptor ISP
module
Microsemi
SMPA-ISP-ACTEL-3-KIT Ships with both large and small header
support
ISP cable for small
header
Microsemi
ISP-CABLE-S
Supplied with SMPA-ISP-ACTEL-3-KIT
ISP cable for large
header
Microsemi
PA-ISP-CABLE
Supplied with SMPA-ISP-ACTEL-3-KIT
Programmer Device Support
Refer to www.microsemi.com/soc for the current information on programmer and device support.
Certified Programming Solutions
The Microsemi-certified programmers for flash devices are FlashPro4, FlashPro3, FlashPro Lite,
FlashPro, Silicon Sculptor II, Silicon Sculptor 3, and any programmer that is built by BP Microsystems. All
other programmers are considered noncertified programmers.
•
FlashPro4, FlashPro3, FlashPro Lite, FlashPro
The Microsemi family of FlashPro device programmers provides in-system programming in an
easy-to-use, compact system that supports all flash families. Whether programming a board
containing a single device or multiple devices connected in a chain, the Microsemi line of
FlashPro programmers enables fast programming and reprogramming. Programming with the
FlashPro series of programmers saves board space and money as it eliminates the need for
sockets on the board. There are no built-in algorithms, so there is no delay between product
release and programming support. The FlashPro programmer is no longer available.
•
Silicon Sculptor 3, Silicon Sculptor II
Silicon Sculptor 3 and Silicon Sculptor II are robust, compact, single-device programmers with
standalone software for the PC. They are designed to enable concurrent programming of multiple
units from the same PC with speeds equivalent to or faster than previous Microsemi
programmers.
•
Noncertified Programmers
Microsemi does not test programming solutions from other vendors, and DOES NOT guarantee
programming yield. Also, Microsemi will not perform any failure analysis on devices programmed
on non-certified programmers. Please refer to the Programming and Functional Failure
Guidelines document for more information.
228
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
•
Programming Centers
Microsemi programming hardware policy also applies to programming centers. Microsemi
expects all programming centers to use certified programmers to program Microsemi devices. If a
programming center uses noncertified programmers to program Microsemi devices, the
"Noncertified Programmers" policy applies.
Important Programming Guidelines
Preprogramming Setup
Before programming, several steps are required to ensure an optimal programming yield.
Use Proper Handling and Electrostatic Discharge (ESD) Precautions
Microsemi FPGAs are sensitive electronic devices that are susceptible to damage from ESD and other
types of mishandling. For more information about ESD, refer to the Quality and Reliability Guide,
beginning with page 41.
Use the Latest Version of the Designer Software to Generate Your
Programming File (recommended)
The files used to program Microsemi flash devices (*.bit, *.stp, *.pdb) contain important information about
the switches that will be programmed in the FPGA. Find the latest version and corresponding release
notes at http://www.microsemi.com/soc/download/software/designer/. Also, programming files must
always be zipped during file transfer to avoid the possibility of file corruption.
Use the Latest Version of the Programming Software
The programming software is frequently updated to accommodate yield enhancements in FPGA
manufacturing. These updates ensure maximum programming yield and minimum programming times.
Before programming, always check the version of software being used to ensure it is the most recent.
Depending on the programming software, refer to one of the following:
•
FlashPro: http://www.microsemi.com/soc/download/program_debug/flashpro/
•
Silicon Sculptor: http://www.microsemi.com/soc/download/program_debug/ss/
Use the Most Recent Adapter Module with Silicon Sculptor
Occasionally, Microsemi makes modifications to the adapter modules to improve programming yields
and programming times. To identify the latest version of each module before programming, visit
http://www.microsemi.com/soc/products/hardware/program_debug/ss/modules.aspx.
Perform Routine Hardware Self-Diagnostic Test
•
Adapter modules must be regularly cleaned. Adapter modules need to be inserted carefully into
the programmer to make sure the DIN connectors (pins at the back side) are not damaged.
•
FlashPro
The self-test is only applicable when programming with FlashPro and FlashPro3 programmers. It
is not supported with FlashPro4 or FlashPro Lite. To run the self-diagnostic test, follow the
instructions given in the "Performing a Self-Test" section of
http://www.microsemi.com/soc/documents/FlashPro_UG.pdf.
•
Silicon Sculptor
The self-diagnostic test verifies correct operation of the pin drivers, power supply, CPU, memory,
and adapter module. This test should be performed with an adapter module installed and before
every programming session. At minimum, the test must be executed every week. To perform selfdiagnostic testing using the Silicon Sculptor software, perform the following steps, depending on
the operating system:
–
DOS: From anywhere in the software, type ALT + D.
–
Windows: Click Device > choose Actel Diagnostic > select the Test tab > click OK.
Silicon Sculptor programmers must be verified annually for calibration. Refer to the Silicon
Sculptor Verification of Calibration Work Instruction document on the website.
Revision 5
229
Programming Flash Devices
Signal Integrity While Using ISP
For ISP of flash devices, customers are expected to follow the board-level guidelines provided on the
Microsemi SoC Products Group website. These guidelines are discussed in the datasheets and
application notes (refer to the “Related Documents” section of the datasheet for application note links).
Customers are also expected to troubleshoot board-level signal integrity issues by measuring voltages
and taking oscilloscope plots.
Programming Failure Allowances
Microsemi has strict policies regarding programming failure allowances. Please refer to Programming
and Functional Failure Guidelines on the Microsemi SoC Products Group website for details.
Contacting the Customer Support Group
Highly skilled engineers staff the Customer Applications Center from 7:00 A.M. to 6:00 P.M., Pacific time,
Monday through Friday. You can contact the center by one of the following methods:
Electronic Mail
You can communicate your technical questions to our email address and receive answers back by email,
fax, or phone. Also, if you have design problems, you can email your design files to receive assistance.
Microsemi monitors the email account throughout the day. When sending your request to us, please be
sure to include your full name, company name, and contact information for efficient processing of your
request. The technical support email address is soc_tech@microsemi.com.
Telephone
Our Technical Support Hotline answers all calls. The center retrieves information, such as your name,
company name, telephone number, and question. Once this is done, a case number is assigned. Then
the center forwards the information to a queue where the first available applications engineer receives
the data and returns your call. The phone hours are from 7:00 A.M. to 6:00 P.M., Pacific time, Monday
through Friday.
The Customer Applications Center number is (800) 262-1060.
European customers can call +44 (0) 1256 305 600.
230
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Related Documents
Below is a list of related documents, their location on the Microsemi SoC Products Group website, and a
brief summary of each document.
Application Notes
Programming Antifuse Devices
http://www.microsemi.com/soc/documents/AntifuseProgram_AN.pdf
Implementation of Security in Actel's ProASIC and ProASICPLUS Flash-Based FPGAs
http://www.microsemi.com/soc/documents/Flash_Security_AN.pdf
User’s Guides
FlashPro Programmers
FlashPro4,1 FlashPro3, FlashPro Lite, and FlashPro2
http://www.microsemi.com/soc/products/hardware/program_debug/flashpro/default.aspx
FlashPro User's Guide
http://www.microsemi.com/soc/documents/FlashPro_UG.pdf
The FlashPro User’s Guide includes hardware and software setup, self-test instructions, use instructions,
and a troubleshooting / error message guide.
Silicon Sculptor 3 and Silicon Sculptor II
http://www.microsemi.com/soc/products/hardware/program_debug/ss/default.aspx
Other Documents
http://www.microsemi.com/soc/products/solutions/security/default.aspx#flashlock
The security resource center describes security in Microsemi Flash FPGAs.
Quality and Reliability Guide
http://www.microsemi.com/soc/documents/RelGuide.pdf
Programming and Functional Failure Guidelines
http://www.microsemi.com/soc/documents/FA_Policies_Guidelines_5-06-00002.pdf
1.
2.
FlashPro4 replaced FlashPro3 in Q1 2010.
FlashPro is no longer available.
Revision 5
231
Programming Flash Devices
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
July 2010
Changes
Page
FlashPro4 is a replacement for FlashPro3 and has been added to this chapter.
FlashPro is no longer available.
N/A
The chapter was updated to include SmartFusion devices.
N/A
The following were deleted:
N/A
"Live at Power-Up (LAPU) or Boot PROM" section
"Design Security" section
Table 14-2 • Programming Features for Actel Devices and much of the text in the
"Programming Features for Microsemi Devices" section
"Programming Flash FPGAs" section
"Return Material Authorization (RMA) Policies" section
232
The "Device Programmers" section was revised.
225
The Independent Programming Centers information was removed from the "Volume
Programming Services" section.
226
Table 10-3 • Programming Solutions was revised to add FlashPro4 and note that
FlashPro is discontinued. A note was added for FlashPro Lite regarding power
supply requirements.
227
Most items were removed from Table 10-4 • Programming Ordering Codes,
including FlashPro3 and FlashPro.
228
The "Programmer Device Support" section was deleted and replaced with a
reference to the Microsemi SoC Products Group website for the latest information.
228
The "Certified Programming Solutions" section was revised to add FlashPro4 and
remove Silicon Sculptor I and Silicon Sculptor 6X. Reference to Programming and
Functional Failure Guidelines was added.
228
The file type *.pdb was added to the "Use the Latest Version of the Designer
Software to Generate Your Programming File (recommended)" section.
229
Instructions on cleaning and careful insertion were added to the "Perform Routine
Hardware Self-Diagnostic Test" section. Information was added regarding testing
Silicon Sculptor programmers with an adapter module installed before every
programming session verifying their calibration annually.
229
The "Signal Integrity While Using ISP" section is new.
230
The "Programming Failure Allowances" section was revised.
230
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Date
v1.3
(December 2008)
v1.2
(October 2008)
Changes
The "Programming Support in Flash Devices" section was updated to include
IGLOO nano and ProASIC3 nano devices.
222
The "Flash Devices" section was updated to include information for IGLOO nano
devices. The following sentence was added: IGLOO PLUS devices can also be
operated at any voltage between 1.2 V and 1.5 V; the Designer software allows
50 mV increments in the voltage.
223
Table 10-4 · Programming Ordering Codes was updated to replace FP3-26PINADAPTER with FP3-10PIN-ADAPTER-KIT.
228
Table 14-6 · Programmer Device Support was updated to add IGLOO nano and
ProASIC3 nano devices. AGL400 was added to the IGLOO portion of the table.
317
The "Programming Support in Flash Devices" section was revised to include new
families and make the information more concise.
222
Figure 10-1 · FlashPro Programming Setup and the "Programming Support in Flash
Devices" section are new.
v1.1
(March 2008)
Page
221, 222
Table 14-6 · Programmer Device Support was updated to include A3PE600L with
the other ProASIC3L devices, and the RT ProASIC3 family was added.
317
The "Flash Devices" section was updated to include the IGLOO PLUS family. The
text, "Voltage switching is required in-system to switch from a 1.2 V core to 1.5 V
core for programming," was revised to state, "Although the device can operate at
1.2 V core voltage, the device can only be reprogrammed when the core voltage is
1.5 V. Voltage switching is required in-system to switch from a 1.2 V supply (VCC,
VCCI, and VJTAG) to 1.5 V for programming."
223
The ProASIC3L family was added to Table 14-6 · Programmer Device Support as a
separate set of rows rather than combined with ProASIC3 and ProASIC3E devices.
The IGLOO PLUS family was included, and AGL015 and A3P015 were added.
317
Revision 5
233
11 – Security in Low Power Flash Devices
Security in Programmable Logic
The need for security on FPGA programmable logic devices (PLDs) has never been greater than today.
If the contents of the FPGA can be read by an external source, the intellectual property (IP) of the system
is vulnerable to unauthorized copying. Fusion, IGLOO, and ProASIC3 devices contain state-of-the-art
circuitry to make the flash-based devices secure during and after programming. Low power flash devices
have a built-in 128-bit Advanced Encryption Standard (AES) decryption core (except for 30 k gate
devices and smaller). The decryption core facilitates secure in-system programming (ISP) of the FPGA
core array fabric, the FlashROM, and the Flash Memory Blocks (FBs) in Fusion devices. The FlashROM,
Flash Blocks, and FPGA core fabric can be programmed independently of each other, allowing the
FlashROM or Flash Blocks to be updated without the need for change to the FPGA core fabric.
Microsemi has incorporated the AES decryption core into the low power flash devices and has also
included the Microsemi flash-based lock technology, FlashLock.® Together, they provide leading-edge
security in a programmable logic device. Configuration data loaded into a device can be decrypted prior
to being written to the FPGA core using the AES 128-bit block cipher standard. The AES encryption key
is stored in on-chip, nonvolatile flash memory.
This document outlines the security features offered in low power flash devices, some applications and
uses, as well as the different software settings for each application.
Figure 11-1 • Overview on Security
Revision 5
235
Security in Low Power Flash Devices
Security Support in Flash-Based Devices
The flash FPGAs listed in Table 11-1 support the security feature and the functions described in this
document.
Table 11-1 • Flash-Based FPGAs
Family*
Series
IGLOO
ProASIC3
Fusion
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO nano
The industry’s lowest-power, smallest-size solution
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
ProASIC3
Low power, high-performance 1.5 V FPGAs
ProASIC3E
Higher density ProASIC3 FPGAs with six PLLs and additional I/O standards
ProASIC3 nano
Lowest-cost solution with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V with Flash*Freeze technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Automotive ProASIC3
ProASIC3 FPGAs qualified for automotive applications
Fusion
Mixed signal FPGA integrating ProASIC3 FPGA fabric, programmable
analog block, support for ARM Cortex™-M1 soft processors, and flash
memory into a monolithic device
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO devices as listed
in Table 11-1. Where the information applies to only one product line or limited devices, these exclusions
will be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 11-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
236
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Security Architecture
Fusion, IGLOO, and ProASIC3 devices have been designed with the most comprehensive programming
logic design security in the industry. In the architecture of these devices, security has been designed into
the very fabric. The flash cells are located beneath seven metal layers, and the use of many device
design and layout techniques makes invasive attacks difficult. Since device layers cannot be removed
without disturbing the charge on the programmed (or erased) flash gates, devices cannot be easily
deconstructed to decode the design. Low power flash devices are unique in being reprogrammable and
having inherent resistance to both invasive and noninvasive attacks on valuable IP. Secure, remote ISP
is now possible with AES encryption capability for the programming file during electronic transfer.
Figure 11-2 shows a view of the AES decryption core inside an IGLOO device; Figure 11-3 on page 238
shows the AES decryption core inside a Fusion device. The AES core is used to decrypt the encrypted
programming file when programming.
Bank 0
Bank 1
Bank 3
CCC
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
I/Os
Bank 1
Bank 3
VersaTile
ISP AES
Decryption*
User Nonvolatile
FlashRom
Flash*Freeze
Technology
Charge
Pumps
RAM Block
4,608-Bit Dual-Port
SRAM or FIFO Block
Bank 2
Note: *ISP AES Decryption is not supported by 30 k gate devices and smaller. For details of other architecture features
by device, refer to the appropriate family datasheet.
Figure 11-2 • Block Representation of the AES Decryption Core in IGLOO and ProASIC3 Devices
Revision 5
237
Security in Low Power Flash Devices
Bank 0
Bank 1
CCC
SRAM Block
4,608-Bit Dual-Port SRAM
or FIFO Block
OSC
I/Os
CCC/PLL
Bank 2
Bank 4
VersaTile
ISP AES
Decryption
User Nonvolatile
FlashROM
Flash Memory Blocks
Analog
Quad
Analog
Quad
Analog
Quad
Analog
Quad
ADC
Analog
Quad
Charge Pumps
SRAM Block
4,608-Bit Dual-Port SRAM
or FIFO Block
Flash Memory Blocks
Analog
Quad
Analog
Quad
Analog
Quad
Analog
Quad
Analog
Quad
CCC
Bank 3
Figure 11-3 • Block Representation of the AES Decryption Core in a Fusion AFS600 FPGA
Security Features
IGLOO and ProASIC3 devices have two entities inside: FlashROM and the FPGA core fabric. Fusion
devices contain three entities: FlashROM, FBs, and the FPGA core fabric. The parts can be programmed
or updated independently with a STAPL programming file. The programming files can be AES-encrypted
or plaintext. This allows maximum flexibility in providing security to the entire device. Refer to the
"Programming Flash Devices" section on page 221 for information on the FlashROM structure.
Unlike SRAM-based FPGA devices, which require a separate boot PROM to store programming data,
low power flash devices are nonvolatile, and the secured configuration data is stored in on-chip flash
cells that are part of the FPGA fabric. Once programmed, this data is an inherent part of the FPGA array
and does not need to be loaded at system power-up. SRAM-based FPGAs load the configuration
bitstream upon power-up; therefore, the configuration is exposed and can be read easily.
The built-in FPGA core, FBs, and FlashROM support programming files encrypted with the 128-bit AES
(FIPS-192) block ciphers. The AES key is stored in dedicated, on-chip flash memory and can be
programmed before the device is shipped to other parties (allowing secure remote field updates).
Security in ARM-Enabled Low Power Flash Devices
There are slight differences between the regular flash devices and the ARM®-enabled flash devices,
which have the M1 and M7 prefix.
The AES key is used by Microsemi and preprogrammed into the device to protect the ARM IP. As a
result, the design is encrypted along with the ARM IP, according to the details below.
238
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Cortex-M1 Device Security
Cortex-M1–enabled devices are shipped with the following security features:
•
FPGA array enabled for AES-encrypted programming and verification
•
FlashROM enabled for AES-encrypted Write and Verify
•
Fusion Embedded Flash Memory enabled for AES-encrypted Write
AES Encryption of Programming Files
Low power flash devices employ AES as part of the security mechanism that prevents invasive and
noninvasive attacks. The mechanism entails encrypting the programming file with AES encryption and
then passing the programming file through the AES decryption core, which is embedded in the device.
The file is decrypted there, and the device is successfully programmed. The AES master key is stored in
on-chip nonvolatile memory (flash). The AES master key can be preloaded into parts in a secure
programming environment (such as the Microsemi In-House Programming center), and then "blank"
parts can be shipped to an untrusted programming or manufacturing center for final personalization with
an AES-encrypted bitstream. Late-stage product changes or personalization can be implemented easily
and securely by simply sending a STAPL file with AES-encrypted data. Secure remote field updates over
public networks (such as the Internet) are possible by sending and programming a STAPL file with AESencrypted data.
The AES key protects the programming data for file transfer into the device with 128-bit AES encryption.
If AES encryption is used, the AES key is stored or preprogrammed into the device. To program, you
must use an AES-encrypted file, and the encryption used on the file must match the encryption key
already in the device.
The AES key is protected by a FlashLock security Pass Key that is also implemented in each device. The
AES key is always protected by the FlashLock Key, and the AES-encrypted file does NOT contain the
FlashLock Key. This FlashLock Pass Key technology is exclusive to the Microsemi flash-based device
families. FlashLock Pass Key technology can also be implemented without the AES encryption option,
providing a choice of different security levels.
In essence, security features can be categorized into the following three options:
•
AES encryption with FlashLock Pass Key protection
•
FlashLock protection only (no AES encryption)
•
No protection
Each of the above options is explained in more detail in the following sections with application examples
and software implementation options.
Advanced Encryption Standard
The 128-bit AES standard (FIPS-192) block cipher is the NIST (National Institute of Standards and
Technology) replacement for DES (Data Encryption Standard FIPS46-2). AES has been designed to
protect sensitive government information well into the 21st century. It replaces the aging DES, which
NIST adopted in 1977 as a Federal Information Processing Standard used by federal agencies to protect
sensitive, unclassified information. The 128-bit AES standard has 3.4 × 1038 possible 128-bit key
variants, and it has been estimated that it would take 1,000 trillion years to crack 128-bit AES cipher text
using exhaustive techniques. Keys are stored (securely) in low power flash devices in nonvolatile flash
memory. All programming files sent to the device can be authenticated by the part prior to programming
to ensure that bad programming data is not loaded into the part that may possibly damage it. All
programming verification is performed on-chip, ensuring that the contents of low power flash devices
remain secure.
Microsemi has implemented the 128-bit AES (Rijndael) algorithm in low power flash devices. With this
key size, there are approximately 3.4 × 1038 possible 128-bit keys. DES has a 56-bit key size, which
provides approximately 7.2 × 1016 possible keys. In their AES fact sheet, the National Institute of
Standards and Technology uses the following hypothetical example to illustrate the theoretical security
provided by AES. If one were to assume that a computing system existed that could recover a DES key
in a second, it would take that same machine approximately 149 trillion years to crack a 128-bit AES key.
NIST continues to make their point by stating the universe is believed to be less than 20 billion years
old.1
Revision 5
239
Security in Low Power Flash Devices
The AES key is securely stored on-chip in dedicated low power flash device flash memory and cannot be
read out. In the first step, the AES key is generated and programmed into the device (for example, at a
secure or trusted programming site). The Microsemi Designer software tool provides AES key generation
capability. After the key has been programmed into the device, the device will only correctly decrypt
programming files that have been encrypted with the same key. If the individual programming file content
is incorrect, a Message Authentication Control (MAC) mechanism inside the device will fail in
authenticating the programming file. In other words, when an encrypted programming file is being loaded
into a device that has a different programmed AES key, the MAC will prevent this incorrect data from
being loaded, preventing possible device damage. See Figure 11-3 on page 238 and Figure 11-4 on
page 240 for graphical representations of this process.
It is important to note that the user decides what level of protection will be implemented for the device.
When AES protection is desired, the FlashLock Pass Key must be set. The AES key is a content
protection mechanism, whereas the FlashLock Pass Key is a device protection mechanism. When the
AES key is programmed into the device, the device still needs the Pass Key to protect the FPGA and
FlashROM contents and the security settings, including the AES key. Using the FlashLock Pass Key
prevents modification of the design contents by means of simply programming the device with a different
AES key.
AES Decryption and MAC Authentication
Low power flash devices have a built-in 128-bit AES decryption core, which decrypts the encrypted
programming file and performs a MAC check that authenticates the file prior to programming.
MAC authenticates the entire programming data stream. After AES decryption, the MAC checks the data
to make sure it is valid programming data for the device. This can be done while the device is still
operating. If the MAC validates the file, the device will be erased and programmed. If the MAC fails to
validate, then the device will continue to operate uninterrupted.
This will ensure the following:
•
Correct decryption of the encrypted programming file
•
Prevention of erroneous or corrupted data being programmed during the programming file
transfer
•
Correct bitstream passed to the device for decryption
IGLOO and ProASIC3
MAC
Validation
Designer
Software
Programming
File Generation
with AES
Encryption
Decrypted
Bitstream
AES
Key
AES
Decryption Core
FlashROM
FPGA
Core
Transmit Medium /
Public Network
Encrypted Bitstream
Figure 11-4 • Example Application Scenario Using AES in IGLOO and ProASIC3 Devices
1.
240
National Institute of Standards and Technology, “ADVANCED ENCRYPTION STANDARD (AES) Questions and Answers,”
28 January 2002 (10 January 2005). See http://csrc.nist.gov/archive/aes/index1.html for more information.
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Fusion
MAC
Validation
Designer
Software
Programming
File Generation
with AES
Encryption
Decrypted
Bitstream
AES
Key
AES
Decryption Core
FlashROM
FPGA
Core
FBs
Transmit Medium /
Public Network
Encrypted Bitstream
Figure 11-5 • Example Application Scenario Using AES in Fusion Devices
FlashLock
Additional Options for IGLOO and ProASIC3 Devices
The user also has the option of prohibiting Write operations to the FPGA array but allowing Verify
operations on the FPGA array and/or Read operations on the FlashROM without the use of the
FlashLock Pass Key. This option provides the user the freedom of verifying the FPGA array and/or
reading the FlashROM contents after the device is programmed, without having to provide the FlashLock
Pass Key. The user can incorporate AES encryption on the programming files to better enhance the level
of security used.
Permanent Security Setting Options
In applications where a permanent lock is not desired, yet the security settings should not be modifiable,
IGLOO and ProASIC3 devices can accommodate this requirement.
This application is particularly useful in cases where a device is located at a remote location and must be
reprogrammed with a design or data update. Refer to the "Application 3: Nontrusted Environment—Field
Updates/Upgrades" section on page 244 for further discussion and examples of how this can be
achieved.
The user must be careful when considering the Permanent FlashLock or Permanent Security Settings
option. Once the design is programmed with the permanent settings, it is not possible to reconfigure the
security settings already employed on the device. Therefore, exercise careful consideration before
programming permanent settings.
Permanent FlashLock
The purpose of the permanent lock feature is to provide the benefits of the highest level of security to
IGLOO and ProASIC3 devices. If selected, the permanent FlashLock feature will create a permanent
barrier, preventing any access to the contents of the device. This is achieved by permanently disabling
Write and Verify access to the array, and Write and Read access to the FlashROM. After permanently
locking the device, it has been effectively rendered one-time-programmable. This feature is useful if the
intended applications do not require design or system updates to the device.
Revision 5
241
Security in Low Power Flash Devices
Security in Action
This section illustrates some applications of the security advantages of Microsemi’s devices (Figure 11-6).
.
Plaintext
Source File
AES
Encryption
Cipher Text
Source File
Application 3
Application 2
Application 1
Public
Domain
AES Decryption Core
FlashROM
FPGA Core
Flash Device
Note: Flash blocks are only used in Fusion devices
Figure 11-6 • Security Options
242
R e vi s i o n 5
Flash Blocks
ProASIC3 nano FPGA Fabric User’s Guide
Application 1: Trusted Environment
As illustrated in Figure 11-7, this application allows the programming of devices at design locations
where research and development take place. Therefore, encryption is not necessary and is optional to
the user. This is often a secure way to protect the design, since the design program files are not sent
elsewhere. In situations where production programming is not available at the design location,
programming centers (such as Microsemi In-House Programming) provide a way of programming
designs at an alternative, secure, and trusted location. In this scenario, the user generates a STAPL
programming file from the Designer software in plaintext format, containing information on the entire
design or the portion of the design to be programmed. The user can choose to employ the FlashLock
Pass Key feature with the design. Once the design is programmed to unprogrammed devices, the design
is protected by this FlashLock Pass Key. If no future programming is needed, the user can consider
permanently securing the IGLOO and ProASIC3 device, as discussed in the "Permanent FlashLock"
section on page 241.
Application 2: Nontrusted Environment—Unsecured Location
Often, programming of devices is not performed in the same location as actual design implementation, to
reduce manufacturing cost. Overseas programming centers and contract manufacturers are examples of
this scenario.
To achieve security in this case, the AES key and the FlashLock Pass Key can be initially programmed
in-house (trusted environment). This is done by generating a programming file with only the security
settings and no design contents. The design FPGA core, FlashROM, and (for Fusion) FB contents are
generated in a separate programming file. This programming file must be set with the same AES key that
was used to program to the device previously so the device will correctly decrypt this encrypted
programming file. As a result, the encrypted design content programming file can be safely sent off-site
to nontrusted programming locations for design programming. Figure 11-7 shows a more detailed flow
for this application.
OEM
Trusted Environment
Generates and Programs Security Settings Only
(programming of the security keys)
Generates Design
Contents Encrypted
with AES
Flash Device
Security Settings*
Security Settings
FPGA/FlashROM/FBs
FPGA/FlashROM/FBs
Flash Device
AES and/or
Pass Key
Protected
Programming File
Sends File(s)
to Manufacturer
Ships Devices
to Manufacturer
Returns Programmed
Devices to Vendor
Ships Programmed
Devices to End Customer
Security Settings
Programs Design
Contents to Devices
FPGA/FlashROM/FBs
Contents
Flash Device
OEM
Customers
Nontrusted Manufacturing Environment
Notes:
1. Programmed portion indicated with dark gray.
2. Programming of FBs applies to Fusion only.
Figure 11-7 • Application 2: Device Programming in a Nontrusted Environment
Revision 5
243
Security in Low Power Flash Devices
Application 3: Nontrusted Environment—Field Updates/Upgrades
Programming or reprogramming of devices may occur at remote locations. Reconfiguration of devices in
consumer products/equipment through public networks is one example. Typically, the remote system is
already programmed with particular design contents. When design update (FPGA array contents update)
and/or data upgrade (FlashROM and/or FB contents upgrade) is necessary, an updated programming file
with AES encryption can be generated, sent across public networks, and transmitted to the remote
system. Reprogramming can then be done using this AES-encrypted programming file, providing easy
and secure field upgrades. Low power flash devices support this secure ISP using AES. The detailed
flow for this application is shown in Figure 11-8. Refer to the "Microprocessor Programming of
Microsemi’s Low Power Flash Devices" chapter of an appropriate FPGA fabric user’s guide for more
information.
To prepare devices for this scenario, the user can initially generate a programming file with the available
security setting options. This programming file is programmed into the devices before shipment. During
the programming file generation step, the user has the option of making the security settings permanent
or not. In situations where no changes to the security settings are necessary, the user can select this
feature in the software to generate the programming file with permanent security settings. Microsemi
recommends that the programming file use encryption with an AES key, especially when ISP is done via
public domain.
For example, if the designer wants to use an AES key for the FPGA array and the FlashROM,
Permanent needs to be chosen for this setting. At first, the user chooses the options to use an AES key
for the FPGA array and the FlashROM, and then chooses Permanently lock the security settings. A
unique AES key is chosen. Once this programming file is generated and programmed to the devices, the
AES key is permanently stored in the on-chip memory, where it is secured safely. The devices are sent to
distant locations for the intended application. When an update is needed, a new programming file must
be generated. The programming file must use the same AES key for encryption; otherwise, the
authentication will fail and the file will not be programmed in the device.
Trusted Environment
OEM
Generates Updated Design Contents
Encrypted with AES
AES Encrypted
Programming File
Transmits to
Remote System
Update/Upgrade
Flash Device
Original Design
Contents AES
Encrypted and
FlashLock Pass Key
Protected
Remote Environment / System
Figure 11-8 • Application 3: Nontrusted Environment—Field Updates/Upgrades
244
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
FlashROM Security Use Models
Each of the subsequent sections describes in detail the available selections in Microsemi Designer as an
aid to understanding security applications and generating appropriate programming files for those
applications. Before proceeding, it is helpful to review Figure 11-7 on page 243, which gives a general
overview of the programming file generation flow within the Designer software as well as what occurs
during the device programming stage. Specific settings are discussed in the following sections.
In Figure 11-7 on page 243, the flow consists of two sub-flows. Sub-flow 1 describes programming
security settings to the device only, and sub-flow 2 describes programming the design contents only.
In Application 1, described in the "Application 1: Trusted Environment" section on page 243, the user
does not need to generate separate files but can generate one programming file containing both security
settings and design contents. Then programming of the security settings and design contents is done in
one step. Both sub-flow 1 and sub-flow 2 are used.
In Application 2, described in the "Application 2: Nontrusted Environment—Unsecured Location" section
on page 243, the trusted site should follow sub-flows 1 and 2 separately to generate two separate
programming files. The programming file from sub-flow 1 will be used at the trusted site to program the
device(s) first. The programming file from sub-flow 2 will be sent off-site for production programming.
In Application 3, described in the "Application 3: Nontrusted Environment—Field Updates/Upgrades"
section on page 244, typically only sub-flow 2 will be used, because only updates to the design content
portion are needed and no security settings need to be changed.
In the event that update of the security settings is necessary, see the "Reprogramming Devices" section
on page 255 for details. For more information on programming low power flash devices, refer to the "InSystem Programming (ISP) of Microsemi’s Low Power Flash Devices Using FlashPro4/3/3X" section on
page 261.
Revision 5
245
Security in Low Power Flash Devices
User
Designer Software
1
Program
Security
Settings
Programming Software
User Assigns Desired Security Settings
To FPGA/FlashROM/FB/All:
– AES Key and FlashLock Pass Key
– FlashLock Pass Key Only
Software Generates Programming File
with Desired Security Settings:
– Encrypted with AES and Protected
with FlashLock Pass Key
– Protected with FlashLock Pass Key Only
Device
Previously
Programmed?
No
Software
Programs
Selected
Security Settings
into Device
Yes
Yes
Software Performs
Comparison of
FlashLock Pass Key
between
Programming File
and Device
Does
FlashLock
Pass Key
Match?
No
Returns Error
2
Program
Design
Contents
Programming
Previously
Secured
Device(s)?
No
Software Generates
Programming File
with Desired
Design Contents
(FPGA Array,
FlashROM, FB,
or All)
Yes
AES Key Used
Previously?
No
User Must
Reassign Exact
FlashLock Pass Key
Previously
Programmed
into the Device
Design Content
Programmed
into Device
Yes
Software Performs
Comparison of
FlashLock Pass Key
between
Programming File
and Device
No
Yes
User Must
Reassign Exact
AES Key
Previously
Programmed
into the Device
Does
FlashLock
Pass Key
Match?
Returns Error
Software Generates
Programming File
with FlashLock
Pass Key and
Design Contents
No
Encrypted Design
Content Passes
through MAC for
Authentication
Correct?
Yes
Software Generates
Programming File
with Encrypted
Design Contents
Design Content
Decrypted and
Programmed
into Device
Note: If programming the Security Header only, just perform sub-flow 1.
If programming design content only, just perform sub-flow 2.
Figure 11-9 • Security Programming Flows
246
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Generating Programming Files
Generation of the Programming File in a Trusted Environment—
Application 1
As discussed in the "Application 1: Trusted Environment" section on page 243, in a trusted environment,
the user can choose to program the device with plaintext bitstream content. It is possible to use plaintext
for programming even when the FlashLock Pass Key option has been selected. In this application, it is
not necessary to employ AES encryption protection. For AES encryption settings, refer to the next
sections.
The generated programming file will include the security setting (if selected) and the plaintext
programming file content for the FPGA array, FlashROM, and/or FBs. These options are indicated in
Table 11-2 and Table 11-3.
Table 11-2 • IGLOO and ProASIC3 Plaintext Security Options, No AES
FlashROM Only
FPGA Core Only
Both FlashROM
and FPGA
No AES / no FlashLock
✓
✓
✓
FlashLock only
✓
✓
✓
AES and FlashLock
–
–
–
Security Protection
Table 11-3 • Fusion Plaintext Security Options
Security Protection
FlashROM Only
FPGA Core Only
FB Core Only
All
No AES / no FlashLock
✓
✓
✓
✓
FlashLock
✓
✓
✓
✓
AES and FlashLock
–
–
–
–
Note: For all instructions, the programming of Flash Blocks refers to Fusion only.
For this scenario, generate the programming file as follows:
1. Select the Silicon features to be programmed (Security Settings, FPGA Array, FlashROM,
Flash Memory Blocks), as shown in Figure 11-10 on page 248 and Figure 11-11 on page 248.
Click Next.
If Security Settings is selected (i.e., the FlashLock security Pass Key feature), an additional
dialog will be displayed to prompt you to select the security level setting. If no security setting is
selected, you will be directed to Step 3.
Revision 5
247
Security in Low Power Flash Devices
Figure 11-10 • All Silicon Features Selected for IGLOO and ProASIC3 Devices
Figure 11-11 • All Silicon Features Selected for Fusion
248
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
2. Choose the appropriate security level setting and enter a FlashLock Pass Key. The default is the
Medium security level (Figure 11-12). Click Next.
If you want to select different options for the FPGA and/or FlashROM, this can be set by clicking
Custom Level. Refer to the "Advanced Options" section on page 256 for different custom
security level options and descriptions of each.
Figure 11-12 • Medium Security Level Selected for Low Power Flash Devices
Revision 5
249
Security in Low Power Flash Devices
3. Choose the desired settings for the FlashROM configurations to be programmed (Figure 11-13).
Click Finish to generate the STAPL programming file for the design.
Figure 11-13 • FlashROM Configuration Settings for Low Power Flash Devices
Generation of Security Header Programming File Only—
Application 2
As mentioned in the "Application 2: Nontrusted Environment—Unsecured Location" section on page 243,
the designer may employ FlashLock Pass Key protection or FlashLock Pass Key with AES encryption on
the device before sending it to a nontrusted or unsecured location for device programming. To achieve
this, the user needs to generate a programming file containing only the security settings desired (Security
Header programming file).
Note: If AES encryption is configured, FlashLock Pass Key protection must also be configured.
The available security options are indicated in Table 11-4 and Table 11-5 on page 251.
Table 11-4 • FlashLock Security Options for IGLOO and ProASIC3
FlashROM Only
FPGA Core Only
Both FlashROM
and FPGA
No AES / no FlashLock
–
–
–
FlashLock only
✓
✓
✓
AES and FlashLock
✓
✓
✓
Security Option
250
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Table 11-5 • FlashLock Security Options for Fusion
Security Option
FlashROM Only
FPGA Core Only
FB Core Only
All
No AES / no FlashLock
–
–
–
–
FlashLock
✓
✓
✓
✓
AES and FlashLock
✓
✓
✓
✓
For this scenario, generate the programming file as follows:
1. Select only the Security settings option, as indicated in Figure 11-14 and Figure 11-15 on
page 252. Click Next.
Figure 11-14 • Programming IGLOO and ProASIC3 Security Settings Only
Revision 5
251
Security in Low Power Flash Devices
Figure 11-15 • Programming Fusion Security Settings Only
2. Choose the desired security level setting and enter the key(s).
–
The High security level employs FlashLock Pass Key with AES Key protection.
–
The Medium security level employs FlashLock Pass Key protection only.
Figure 11-16 • High Security Level to Implement FlashLock Pass Key and AES Key Protection
252
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Table 11-6 and Table 11-7 show all available options. If you want to implement custom levels,
refer to the "Advanced Options" section on page 256 for information on each option and how to
set it.
3. When done, click Finish to generate the Security Header programming file.
Table 11-6 • All IGLOO and ProASIC3 Header File Security Options
FlashROM Only
FPGA Core Only
Both FlashROM
and FPGA
No AES / no FlashLock
✓
✓
✓
FlashLock only
✓
✓
✓
AES and FlashLock
✓
✓
✓
Security Option
Note: ✓ = options that may be used
Table 11-7 • All Fusion Header File Security Options
Security Option
FlashROM Only
FPGA Core Only
FB Core Only
All
No AES / No FlashLock
✓
✓
✓
✓
FlashLock
✓
✓
✓
✓
AES and FlashLock
✓
✓
✓
✓
Generation of Programming Files with AES Encryption—
Application 3
This section discusses how to generate design content programming files needed specifically at
unsecured or remote locations to program devices with a Security Header (FlashLock Pass Key and AES
key) already programmed ("Application 2: Nontrusted Environment—Unsecured Location" section on
page 243 and "Application 3: Nontrusted Environment—Field Updates/Upgrades" section on page 244).
In this case, the encrypted programming file must correspond to the AES key already programmed into
the device. If AES encryption was previously selected to encrypt the FlashROM, FBs, and FPGA array,
AES encryption must be set when generating the programming file for them. AES encryption can be
applied to the FlashROM only, the FBs only, the FPGA array only, or all. The user must ensure both the
FlashLock Pass Key and the AES key match those already programmed to the device(s), and all security
settings must match what was previously programmed. Otherwise, the encryption and/or device
unlocking will not be recognized when attempting to program the device with the programming file.
The generated programming file will be AES-encrypted.
In this scenario, generate the programming file as follows:
1. Deselect Security settings and select the portion of the device to be programmed (Figure 11-17
on page 254). Select Programming previously secured device(s). Click Next.
Revision 5
253
Security in Low Power Flash Devices
Note: The settings in this figure are used to show the generation of an AES-encrypted programming file for the FPGA
array, FlashROM, and FB contents. One or all locations may be selected for encryption.
Figure 11-17 • Settings to Program a Device Secured with FlashLock and using AES Encryption
Choose the High security level to reprogram devices using both the FlashLock Pass Key and AES key
protection (Figure 11-18 on page 255). Enter the AES key and click Next.
A device that has already been secured with FlashLock and has an AES key loaded must recognize the
AES key to program the device and generate a valid bitstream in authentication. The FlashLock Key is
only required to unlock the device and change the security settings.
This is what makes it possible to program in an untrusted environment. The AES key is protected inside
the device by the FlashLock Key, so you can only program if you have the correct AES key. In fact, the
AES key is not in the programming file either. It is the key used to encrypt the data in the file. The same
key previously programmed with the FlashLock Key matches to decrypt the file.
An AES-encrypted file programmed to a device without FlashLock would not be secure, since without
FlashLock to protect the AES key, someone could simply reprogram the AES key first, then program with
any AES key desired or no AES key at all. This option is therefore not available in the software.
254
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Figure 11-18 • Security Level Set High to Reprogram Device with AES Key
Programming with this file is intended for an unsecured environment. The AES key encrypts the
programming file with the same AES key already used in the device and utilizes it to program the device.
Reprogramming Devices
Previously programmed devices can be reprogrammed using the steps in the "Generation of the
Programming File in a Trusted Environment—Application 1" section on page 247 and "Generation of
Security Header Programming File Only—Application 2" section on page 250. In the case where a
FlashLock Pass Key has been programmed previously, the user must generate the new programming file
with a FlashLock Pass Key that matches the one previously programmed into the device. The software
will check the FlashLock Pass Key in the programming file against the FlashLock Pass Key in the device.
The keys must match before the device can be unlocked to perform further programming with the new
programming file.
Figure 11-10 on page 248 and Figure 11-11 on page 248 show the option Programming previously
secured device(s), which the user should select before proceeding. Upon going to the next step, the
user will be notified that the same FlashLock Pass Key needs to be entered, as shown in Figure 11-19 on
page 256.
Revision 5
255
Security in Low Power Flash Devices
Figure 11-19 • FlashLock Pass Key, Previously Programmed Devices
It is important to note that when the security settings need to be updated, the user also needs to select
the Security settings check box in Step 1, as shown in Figure 11-10 on page 248 and Figure 11-11 on
page 248, to modify the security settings. The user must consider the following:
•
If only a new AES key is necessary, the user must re-enter the same Pass Key previously
programmed into the device in Designer and then generate a programming file with the same
Pass Key and a different AES key. This ensures the programming file can be used to access and
program the device and the new AES key.
•
If a new Pass Key is necessary, the user can generate a new programming file with a new Pass
Key (with the same or a new AES key if desired). However, for programming, the user must first
load the original programming file with the Pass Key that was previously used to unlock the
device. Then the new programming file can be used to program the new security settings.
Advanced Options
As mentioned, there may be applications where more complicated security settings are required. The
“Custom Security Levels” section in the FlashPro User's Guide describes different advanced options
available to aid the user in obtaining the best available security settings.
256
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Programming File Header Definition
In each STAPL programming file generated, there will be information about how the AES key and
FlashLock Pass Key are configured. Table 11-8 shows the header definitions in STAPL programming
files for different security levels.
Table 11-8 • STAPL Programming File Header Definitions by Security Level
Security Level
STAPL File Header Definition
No security (no FlashLock Pass Key or AES key)
NOTE "SECURITY" "Disable";
FlashLock Pass Key with no AES key
NOTE "SECURITY" "KEYED ";
FlashLock Pass Key with AES key
NOTE "SECURITY" "KEYED ENCRYPT ";
Permanent Security Settings option enabled
NOTE "SECURITY" "PERMLOCK ENCRYPT ";
AES-encrypted FPGA array (for programming updates)
NOTE "SECURITY" "ENCRYPT CORE ";
AES-encrypted FlashROM (for programming updates)
NOTE "SECURITY" "ENCRYPT FROM ";
AES-encrypted FPGA
programming updates)
array
and
FlashROM
(for NOTE "SECURITY" "ENCRYPT FROM CORE ";
Example File Headers
STAPL Files Generated with FlashLock Key and AES Key Containing Key Information
•
FlashLock Key / AES key indicated in STAPL file header definition
•
Intended ONLY for secured/trusted environment programming applications
=============================================
NOTE "CREATOR" "Designer Version: 6.1.1.108";
NOTE "DEVICE" "A3PE600";
NOTE "PACKAGE" "208 PQFP";
NOTE "DATE" "2005/04/08";
NOTE "STAPL_VERSION" "JESD71";
NOTE "IDCODE" "$123261CF";
NOTE "DESIGN" "counter32";
NOTE "CHECKSUM" "$EDB9";
NOTE "SAVE_DATA" "FRomStream";
NOTE "SECURITY" "KEYED ENCRYPT ";
NOTE "ALG_VERSION" "1";
NOTE "MAX_FREQ" "20000000";
NOTE "SILSIG" "$00000000";
NOTE "PASS_KEY" "$00123456789012345678901234567890";
NOTE "AES_KEY" "$ABCDEFABCDEFABCDEFABCDEFABCDEFAB";
==============================================
Revision 5
257
Security in Low Power Flash Devices
STAPL File with AES Encryption
•
Does not contain AES key / FlashLock Key information
•
Intended for transmission through web or service to unsecured locations for programming
=============================================
NOTE "CREATOR" "Designer Version: 6.1.1.108";
NOTE "DEVICE" "A3PE600";
NOTE "PACKAGE" "208 PQFP";
NOTE "DATE" "2005/04/08";
NOTE "STAPL_VERSION" "JESD71";
NOTE "IDCODE" "$123261CF";
NOTE "DESIGN" "counter32";
NOTE "CHECKSUM" "$EF57";
NOTE "SAVE_DATA" "FRomStream";
NOTE "SECURITY" "ENCRYPT FROM CORE ";
NOTE "ALG_VERSION" "1";
NOTE "MAX_FREQ" "20000000";
NOTE "SILSIG" "$00000000";
Conclusion
The new and enhanced security features offered in Fusion, IGLOO, and ProASIC3 devices provide stateof-the-art security to designs programmed into these flash-based devices. Microsemi low power flash
devices employ the encryption standard used by NIST and the U.S. government—AES using the 128-bit
Rijndael algorithm.
The combination of an on-chip AES decryption engine and FlashLock technology provides the highest
level of security against invasive attacks and design theft, implementing the most robust and secure ISP
solution. These security features protect IP within the FPGA and protect the system from cloning,
wholesale “black box” copying of a design, invasive attacks, and explicit IP or data theft.
Glossary
Term
Explanation
Security Header
programming file
Programming file used to program the FlashLock Pass Key and/or AES key into the device to
secure the FPGA, FlashROM, and/or FBs.
AES (encryption) key
128-bit key defined by the user when the AES encryption option is set in the Microsemi
Designer software when generating the programming file.
FlashLock Pass Key
128-bit key defined by the user when the FlashLock option is set in the Microsemi Designer
software when generating the programming file.
The FlashLock Key protects the security settings programmed to the device. Once a device
is programmed with FlashLock, whatever settings were chosen at that time are secure.
FlashLock
The combined security features that protect the device content from attacks. These features
are the following:
•
Flash technology that does not require an external bitstream to program the device
•
FlashLock Pass Key that secures device content by locking the security settings and
preventing access to the device as defined by the user
•
AES key that allows secure, encrypted device reprogrammability
References
National Institute of Standards and Technology. “ADVANCED ENCRYPTION STANDARD (AES)
Questions and Answers.” 28 January 2002 (10 January 2005).
See http://csrc.nist.gov/archive/aes/index1.html for more information.
258
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Related Documents
User’s Guides
FlashPro User's Guide
http://www.microsemi.com/soc/documents/flashpro_ug.pdf
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
Changes
Page
July 2010
This chapter is no longer published separately with its own part number and version
but is now part of several FPGA fabric user’s guides.
N/A
v1.5
(August 2009)
The "CoreMP7 Device Security" section was removed from "Security in ARMEnabled Low Power Flash Devices", since M7-enabled devices are no longer
supported.
238
v1.4
(December 2008)
IGLOO nano and ProASIC3 nano devices were added to Table 11-1 • Flash-Based
FPGAs.
236
v1.3
(October 2008)
The "Security Support in Flash-Based Devices" section was revised to include new
families and make the information more concise.
236
v1.2
(June 2008)
The following changes were made to the family descriptions in Table 11-1 • FlashBased FPGAs:
236
v1.1
(March 2008)
•
ProASIC3L was updated to include 1.5 V.
•
The number of PLLs for ProASIC3E was changed from five to six.
The chapter was updated to include the IGLOO PLUS family and information
regarding 15 k gate devices.
N/A
The "IGLOO Terminology" section and "ProASIC3 Terminology" section are new.
236
Revision 5
259
12 – In-System Programming (ISP) of Microsemi’s
Low Power Flash Devices Using FlashPro4/3/3X
Introduction
Microsemi’s low power flash devices are all in-system programmable. This document describes the
general requirements for programming a device and specific requirements for the FlashPro4/3/3X
programmers1.
IGLOO, ProASIC3, SmartFusion, and Fusion devices offer a low power, single-chip, live-at-power-up
solution with the ASIC advantages of security and low unit cost through nonvolatile flash technology.
Each device contains 1 kbit of on-chip, user-accessible, nonvolatile FlashROM. The FlashROM can be
used in diverse system applications such as Internet Protocol (IP) addressing, user system preference
storage, device serialization, or subscription-based business models. IGLOO, ProASIC3, SmartFusion,
and Fusion devices offer the best in-system programming (ISP) solution, FlashLock® security features,
and AES-decryption-based ISP.
ISP Architecture
Low power flash devices support ISP via JTAG and require a single VPUMP voltage of 3.3 V during
programming. In addition, programming via a microcontroller in a target system is also supported.
Refer to the "Microprocessor Programming of Microsemi’s Low Power Flash Devices" chapter of an
appropriate FPGA fabric user’s guide.
Family-specific support:
•
ProASIC3, ProASIC3E, SmartFusion, and Fusion devices support ISP.
•
ProASIC3L devices operate using a 1.2 V core voltage; however, programming can be done only
at 1.5 V. Voltage switching is required in-system to switch from a 1.2 V core to 1.5 V core for
programming.
•
IGLOO and IGLOOe V5 devices can be programmed in-system when the device is using a 1.5 V
supply voltage to the FPGA core.
•
IGLOO nano V2 devices can be programmed at 1.2 V core voltage (when using FlashPro4 only)
or 1.5 V. IGLOO nano V5 devices are programmed with a VCC core voltage of 1.5 V. Voltage
switching is required in-system to switch from a 1.2 V supply (VCC,VCCI, and VJTAG) to 1.5 V
for programming. The exception is that V2 devices can be programmed at 1.2 V VCC with
FlashPro4.
IGLOO devices cannot be programmed in-system when the device is in Flash*Freeze mode. The device
should exit Flash*Freeze mode and be in normal operation for programming to start. Programming
operations in IGLOO devices can be achieved when the device is in normal operating mode and a 1.5 V
core voltage is used.
JTAG 1532
IGLOO, ProASIC3, SmartFusion, and Fusion devices support the JTAG-based IEEE 1532 standard for
ISP. To start JTAG operations, the IGLOO device must exit Flash*Freeze mode and be in normal
operation before starting to send JTAG commands to the device. As part of this support, when a device is
in an unprogrammed state, all user I/O pins are disabled. This is achieved by keeping the global IO_EN
1.
FlashPro4 replaced FlashPro3/3X in 2010 and is backward compatible with FlashPro3/3X as long as there is no connection
to pin 4 on the JTAG header on the board. On FlashPro3/3X, there is no connection to pin 4 on the JTAG header; however,
pin 4 is used for programming mode (Prog_Mode) on FlashPro4. When converting from FlashPro3/3X to FlashPro4, users
should make sure that JTAG connectors on system boards do not have any connection to pin 4. FlashPro3X supports
discrete TCK toggling that is needed to support non-JTAG compliant devices in the chain. This feature is included in
FlashPro4.
Revision 5
261
In-System Programming (ISP) of Microsemi’s Low Power Flash Devices Using FlashPro4/3/3X
signal deactivated, which also has the effect of disabling the input buffers. The SAMPLE/PRELOAD
instruction captures the status of pads in parallel and shifts them out as new data is shifted in for loading
into the Boundary Scan Register (BSR). When the device is in an unprogrammed state, the OE and
output BSR will be undefined; however, the input BSR will be defined as long as it is connected and
being used. For JTAG timing information on setup, hold, and fall times, refer to the FlashPro User’s
Guide.
ISP Support in Flash-Based Devices
The flash FPGAs listed in Table 12-1 support the ISP feature and the functions described in this
document.
Table 12-1 • Flash-Based FPGAs Supporting ISP
Family*
Series
IGLOO
ProASIC3
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO nano
The industry’s lowest-power, smallest-size solution
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
ProASIC3
Low power, high-performance 1.5 V FPGAs
ProASIC3E
Higher density ProASIC3 FPGAs with six PLLs and additional I/O standards
ProASIC3 nano
Lowest-cost solution with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V with Flash*Freeze technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Automotive ProASIC3
ProASIC3 FPGAs qualified for automotive applications
SmartFusion SmartFusion
Mixed signal FPGA integrating ProASIC3 FPGA fabric, programmable
microcontroller subsystem (MSS) which includes programmable analog and
an ARM® Cortex™-M3 hard processor and flash memory in a monolithic
device
Fusion
Fusion
Mixed signal FPGA integrating ProASIC3 FPGA fabric, programmable
analog block, support for ARM® Cortex™-M1 soft processors, and flash
memory into a monolithic device
ProASIC
ProASIC
First generation ProASIC devices
ProASICPLUS
Second generation ProASIC devices
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO devices as listed
in Table 12-1. Where the information applies to only one product line or limited devices, these exclusions
will be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 12-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
262
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Programming Voltage (VPUMP) and VJTAG
Low-power flash devices support on-chip charge pumps, and therefore require only a single 3.3 V
programming voltage for the VPUMP pin during programming. When the device is not being
programmed, the VPUMP pin can be left floating or can be tied (pulled up) to any voltage between 0 V
and 3.6 V2. During programming, the target board or the FlashPro4/3/3X programmer can provide
VPUMP. FlashPro4/3/3X is capable of supplying VPUMP to a single device. If more than one device is to
be programmed using FlashPro4/3/3X on a given board, FlashPro4/3/3X should not be relied on to
supply the VPUMP voltage. A FlashPro4/3/3X programmer is not capable of providing reliable VJTAG
voltage. The board must supply VJTAG voltage to the device and the VJTAG pin of the programmer
header must be connected to the device VJTAG pin. Microsemi recommends that VPUMP3 and VJTAG
power supplies be kept separate with independent filtering capacitors rather than supplying them from a
common rail. Refer to the "Board-Level Considerations" section on page 271 for capacitor requirements.
Low power flash device I/Os support a bank-based, voltage-supply architecture that simultaneously
supports multiple I/O voltage standards (Table 12-2). By isolating the JTAG power supply in a separate
bank from the user I/Os, low power flash devices provide greater flexibility with supply selection and
simplify power supply and printed circuit board (PCB) design. The JTAG pins can be run at any voltage
from 1.5 V to 3.3 V (nominal). Microsemi recommends that TCK be tied to GND through a 200 ohm to 1
Kohm resistor. This prevents a possible totempole current on the input buffer stage. For TDI, TMS, and
TRST pins, the devices provide an internal nominal 10 Kohm pull-up resistor. During programming, all
I/O pins, except for JTAG interface pins, are tristated and weakly pulled up to VCCI. This isolates the part
and prevents the signals from floating. The JTAG interface pins are driven by the FlashPro4/3/3X during
programming, including the TRST pin, which is driven HIGH.
Table 12-2 • Power Supplies
Programming Mode
Current during
Programming
VCC
1.2 V / 1.5 V
< 70 mA
VCCI
1.2 V / 1.5 V / 1.8 V / 2.5 V / 3.3 V
(bank-selectable)
I/Os are weakly pulled up.
VJTAG
1.2 V / 1.5 V / 1.8 V / 2.5 V / 3.3 V
< 20 mA
VPUMP
3.15 V to 3.45 V
< 80 mA
Power Supply
Note: All supply voltages should be at 1.5 V or higher, regardless of the setting during normal
operation, except for IGLOO nano, where 1.2 V VCC and VJTAG programming is allowed.
Nonvolatile Memory (NVM) Programming Voltage
SmartFusion and Fusion devices need stable VCCNVM/VCCENVM3 (1.5 V power supply to the
embedded nonvolatile memory blocks) and VCCOSC/VCCROSC4 (3.3 V power supply to the integrated
RC oscillator). The tolerance of VCCNVM/VCCENVM is ± 5% and VCCOSC/VCCROSC is ± 5%.
Unstable supply voltage on these pins can cause an NVM programming failure due to NVM page
corruption. The NVM page can also be corrupted if the NVM reset pin has noise. This signal must be tied
off properly.
Microsemi recommends installing the following capacitors5 on the VCCNVM/VCCENVM and
VCCOSC/VCCROSC pins:
2.
3.
4.
5.
•
Add one bypass capacitor of 10 µF for each power supply plane followed by an array of
decoupling capacitors of 0.1 µF.
•
Add one 0.1 µF capacitor near each pin.
During sleep mode in IGLOO devices connect VPUMP to GND.
VPUMP has to be quiet for successful programming. Therefore VPUMP must be separate and required capacitors must be
installed close to the FPGA VPUMP pin.
VCCROSC is for SmartFusion.
The capacitors cannot guarantee reliable operation of the device if the board layout is not done properly.
Revision 5
263
In-System Programming (ISP) of Microsemi’s Low Power Flash Devices Using FlashPro4/3/3X
IEEE 1532 (JTAG) Interface
The supported industry-standard IEEE 1532 programming interface builds on the IEEE 1149.1 (JTAG)
standard. IEEE 1532 defines the standardized process and methodology for ISP. Both silicon and
software issues are addressed in IEEE 1532 to create a simplified ISP environment. Any IEEE 1532
compliant programmer can be used to program low power flash devices. Device serialization is not
supported when using the IEEE1532 standard. Refer to the standard for detailed information about IEEE
1532.
Security
Unlike SRAM-based FPGAs that require loading at power-up from an external source such as a
microcontroller or boot PROM, Microsemi nonvolatile devices are live at power-up, and there is no
bitstream required to load the device when power is applied. The unique flash-based architecture
prevents reverse engineering of the programmed code on the device, because the programmed data is
stored in nonvolatile memory cells. Each nonvolatile memory cell is made up of small capacitors and any
physical deconstruction of the device will disrupt stored electrical charges.
Each low power flash device has a built-in 128-bit Advanced Encryption Standard (AES) decryption core,
except for the 30 k gate devices and smaller. Any FPGA core or FlashROM content loaded into the
device can optionally be sent as encrypted bitstream and decrypted as it is loaded. This is particularly
suitable for applications where device updates must be transmitted over an unsecured network such as
the Internet. The embedded AES decryption core can prevent sensitive data from being intercepted
(Figure 12-1 on page 265). A single 128-bit AES Key (32 hex characters) is used to encrypt FPGA core
programming data and/or FlashROM programming data in the Microsemi tools. The low power flash
devices also decrypt with a single 128-bit AES Key. In addition, low power flash devices support a
Message Authentication Code (MAC) for authentication of the encrypted bitstream on-chip. This allows
the encrypted bitstream to be authenticated and prevents erroneous data from being programmed into
the device. The FPGA core, FlashROM, and Flash Memory Blocks (FBs), in Fusion only, can be updated
independently using a programming file that is AES-encrypted (cipher text) or uses plain text.
264
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Security in ARM-Enabled Low Power Flash Devices
There are slight differences between the regular flash device and the ARM-enabled flash devices, which
have the M1 prefix.
The AES key is used by Microsemi and preprogrammed into the device to protect the ARM IP. As a
result, the design will be encrypted along with the ARM IP, according to the details below.
Cortex-M1 and Cortex-M3 Device Security
Cortex-M1–enabled and Cortex-M3 devices are shipped with the following security features:
•
FPGA array enabled for AES-encrypted programming and verification
•
FlashROM enabled for AES-encrypted write and verify
•
Embedded Flash Memory enabled for AES encrypted write
Flash Device
Designer
Software
MAC
Validation
User Encryption AES Key
Decrypted
Bitstream
Programming
File Generation
with AES
Encryption
AES
Decryption
FPGA Core,
FlashROM,
FBs
Transmit Medium /
Public Network
Encrypted Bistream
Figure 12-1 • AES-128 Security Features
Revision 5
265
In-System Programming (ISP) of Microsemi’s Low Power Flash Devices Using FlashPro4/3/3X
Figure 12-2 shows different applications for ISP programming.
1. In a trusted programming environment, you can program the device using the unencrypted
(plaintext) programming file.
2. You can program the AES Key in a trusted programming environment and finish the final
programming in an untrusted environment using the AES-encrypted (cipher text) programming
file.
3. For the remote ISP updating/reprogramming, the AES Key stored in the device enables the
encrypted programming bitstream to be transmitted through the untrusted network connection.
Microsemi low power flash devices also provide the unique Microsemi FlashLock feature, which protects
the Pass Key and AES Key. Unless the original FlashLock Pass Key is used to unlock the device,
security settings cannot be modified. Microsemi does not support read-back of FPGA core-programmed
data; however, the FlashROM contents can selectively be read back (or disabled) via the JTAG port
based on the security settings established by the Microsemi Designer software. Refer to the "Security in
Low Power Flash Devices" section on page 235 for more information.
Source
Plain Text
AES
Encryption
Source
Encrypted Bitstream
FlashROM
Option 3
Option 2
Option 1
TCP/IP
AES
Decryption
FPGA
Core
IGLOO or ProASIC3 Device
Figure 12-2 • Different ISP Use Models
266
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
FlashROM and Programming Files
Each low power flash device has 1 kbit of on-chip, nonvolatile flash memory that can be accessed from
the FPGA core. This nonvolatile FlashROM is arranged in eight pages of 128 bits (Figure 12-3). Each
page can be programmed independently, with or without the 128-bit AES encryption. The FlashROM can
only be programmed via the IEEE 1532 JTAG port and cannot be programmed from the FPGA core. In
addition, during programming of the FlashROM, the FPGA core is powered down automatically by the
on-chip programming control logic.
Page Number
15
14
13
12
Byte Number in Page
11 10
9
8
7
6
5
4
3
2
1
0
7
6
5
4
3
2
1
0
Figure 12-3 • FlashROM Architecture
When using FlashROM combined with AES, many subscription-based applications or device
serialization applications are possible. The FROM configurator found in the Libero SoC Catalog supports
easy management of the FlashROM contents, even over large numbers of devices. The FROM
configurator can support FlashROM contents that contain the following:
•
Static values
•
Random numbers
•
Values read from a file
•
Independent updates of each page
In addition, auto-incrementing of fields is possible. In applications where the FlashROM content is
different for each device, you have the option to generate a single STAPL file for all the devices or
individual serialization files for each device. For more information on how to generate the FlashROM
content for device serialization, refer to the "FlashROM in Microsemi’s Low Power Flash Devices" section
on page 117.
Libero SoC includes a unique tool to support the generation and management of FlashROM and FPGA
programming files. This tool is called FlashPoint.
Depending on the applications, designers can use the FlashPoint software to generate a STAPL file with
different contents. In each case, optional AES encryption and/or different security settings can be set.
In Designer, when you click the Programming File icon, FlashPoint launches, and you can generate
STAPL file(s) with four different cases (Figure 12-4 on page 268). When the serialization feature is used
during the configuration of FlashROM, you can generate a single STAPL file that will program all the
devices or an individual STAPL file for each device.
The following cases present the FPGA core and FlashROM programming file combinations that can be
used for different applications. In each case, you can set the optional security settings (FlashLock Pass
Key and/or AES Key) depending on the application.
1. A single STAPL file or multiple STAPL files with multiple FlashROM contents and the FPGA core
content. A single STAPL file will be generated if the device serialization feature is not used. You
can program the whole FlashROM or selectively program individual pages.
2. A single STAPL file for the FPGA core content
Revision 5
267
In-System Programming (ISP) of Microsemi’s Low Power Flash Devices Using FlashPro4/3/3X
3. A single STAPL file or multiple STAPL files with multiple FlashROM contents. A single STAPL file
will be generated if the device serialization feature is not used. You can program the whole
FlashROM or selectively program individual pages.
4. A single STAPL file to configure the security settings for the device, such as the AES Key and/or
Pass Key.
Libero SoC
Catalog
Designer Software Suite
Netlist
Programming
File
(FlashPoint)
1
2
FlashROM
Configuration
File (*.ufc)
3
Security
Settings
Security
Settings
Security
Settings
Single/Multiple
FlashROM
Content(s)
FPGA Core
Content
Single/Multiple
FlashROM
Content(s)
4
Security
Settings
FPGA Core
Content
Figure 12-4 • Flexible Programming File Generation for Different Applications
Programming Solution
For device programming, any IEEE 1532–compliant programmer can be used; however, the
FlashPro4/3/3X programmer must be used to control the low power flash device's rich security features
and FlashROM programming options. The FlashPro4/3/3X programmer is a low-cost portable
programmer for the Microsemi flash families. It can also be used with a powered USB hub for parallel
programming. General specifications for the FlashPro4/3/3X programmer are as follows:
268
•
Programming clock – TCK is used with a maximum frequency of 20 MHz, and the default
frequency is 4 MHz.
•
Programming file – STAPL
•
Daisy chain – Supported. You can use the ChainBuilder software to build the programming file for
the chain.
•
Parallel programming – Supported. Multiple FlashPro4/3/3X programmers can be connected
together using a powered USB hub or through the multiple USB ports on the PC.
•
Power supply – The target board must provide VCC, VCCI, VPUMP, and VJTAG during
programming. However, if there is only one device on the target board, the FlashPro4/3/3X
programmer can generate the required VPUMP voltage from the USB port.
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
ISP Programming Header Information
The FlashPro4/3/3X programming cable connector can be connected with a 10-pin, 0.1"-pitch
programming header. The recommended programming headers are manufactured by AMP (103310-1)
and 3M (2510-6002UB). If you have limited board space, you can use a compact programming header
manufactured by Samtec (FTSH-105-01-L-D-K). Using this compact programming header, you are
required to order an additional header adapter manufactured by Microsemi SoC Products Group (FP310PIN-ADAPTER-KIT).
Existing ProASICPLUS family customers who are using the Samtec Small Programming Header
(FTSH-113-01-L-D-K) and are planning to migrate to IGLOO or ProASIC3 devices can also use
FP3-10PIN-ADAPTER-KIT.
Table 12-3 • Programming Header Ordering Codes
Manufacturer
Part Number
Description
103310-1
10-pin, 0.1"-pitch cable header (right-angle PCB mount
angle)
2510-6002UB
10-pin, 0.1"-pitch cable header (straight PCB mount
angle)
Samtec
FTSH-113-01-L-D-K
Small programming header supported by FlashPro and
Silicon Sculptor
Samtec
FTSH-105-01-L-D-K
Compact programming header
Samtec
FFSD-05-D-06.00-01-N
AMP
3M
Microsemi
10-pin cable with 50 mil pitch sockets; included in FP310PIN-ADAPTER-KIT.
FP3-10PIN-ADAPTER-KIT Transition adapter kit to allow FP3 to be connected to a
micro 10-pin header (50 mil pitch). Includes a 6 inch
Samtec FFSD-05-D-06.00-01-N cable in the kit. The
transition adapter board was previously offered as
FP3-26PIN-ADAPTER and includes a 26-pin adapter for
design transitions from ProASICPLUS based boards to
ProASIC3 based boards.
TCK
TDO
TMS
VPUMP
TDI
1 2
3 4
5 6
7 8
9 10
GND
NC (FlashPro3/3X); Prog_Mode* (FlashPro4)
VJTAG
TRST
GND
Note: *Prog_Mode on FlashPro4 is an output signal that goes High during device programming and
returns to Low when programming is complete. This signal can be used to drive a system to provide
a 1.5 V programming signal to IGLOO nano, ProASIC3L, and RT ProASIC3 devices that can run
with 1.2 V core voltage but require 1.5 V for programming. IGLOO nano V2 devices can be
programmed at 1.2 V core voltage (when using FlashPro4 only), but IGLOO nano V5 devices are
programmed with a VCC core voltage of 1.5 V.
Figure 12-5 • Programming Header (top view)
Revision 5
269
In-System Programming (ISP) of Microsemi’s Low Power Flash Devices Using FlashPro4/3/3X
Table 12-4 • Programming Header Pin Numbers and Description
Pin
Signal
Source
1
TCK
Programmer
2
GND1
–
Signal Reference
3
TDO
Target Board
Test Data Output
4
NC
–
5
TMS
Programmer
Test Mode Select
6
VJTAG
Target Board
JTAG Supply Voltage
2
Description
JTAG Clock
No Connect (FlashPro3/3X); Prog_Mode (FlashPro4).
See note associated with Figure 12-5 on page 269
regarding Prog_Mode on FlashPro4.
7
VPUMP
Programmer/Target Board
8
nTRST
Programmer
JTAG Test Reset (Hi-Z with 10 kΩ pull-down, HIGH,
LOW, or toggling)
9
TDI
Programmer
Test Data Input
10
1
GND
–
Programming Supply Voltage
Signal Reference
Notes:
1. Both GND pins must be connected.
2. FlashPro4/3/3X can provide VPUMP if there is only one device on the target board.
270
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Board-Level Considerations
A bypass capacitor is required from VPUMP to GND for all low power flash devices during programming.
This bypass capacitor protects the devices from voltage spikes that may occur on the VPUMP supplies
during the erase and programming cycles. Refer to the "Pin Descriptions and Packaging" chapter of the
appropriate device datasheet for specific recommendations. For proper programming, 0.01 µF and 0.33
µF capacitors (both rated at 16 V) are to be connected in parallel across VPUMP and GND, and
positioned as close to the FPGA pins as possible. The bypass capacitor must be placed within 2.5 cm of
the device pins.
VJTAG from the target board
VCCI from the target board
VCC from the target board
VCC
VCCI
VJTAG
Low Power
Flash Device
Polarizing Notch
GND
TCK
TDO
TMS
VPUMP
1 TCK
3 TDO
5 TMS
7 VPUMP
TDI
9 TDI
2 GND
4 NC*
6 VJTAG
8 TRST
10 GND
TRST
C1 C2
R
R
Note: *NC (FlashPro3/3X); Prog_Mode (FlashPro4). Prog_Mode on FlashPro4 is an output signal that goes High during
device programming and returns to Low when programming is complete. This signal can be used to drive a
system to provide a 1.5 V programming signal to IGLOO nano, ProASIC3L, and RT ProASIC3 devices that can
run with 1.2 V core voltage but require 1.5 V for programming. IGLOO nano V2 devices can be programmed at
1.2 V core voltage (when using FlashPro4 only), but IGLOO nano V5 devices are programmed with a VCC core
voltage of 1.5 V.
Figure 12-6 • Board Layout and Programming Header Top View
Troubleshooting Signal Integrity
Symptoms of a Signal Integrity Problem
A signal integrity problem can manifest itself in many ways. The problem may show up as extra or
dropped bits during serial communication, changing the meaning of the communication. There is a
normal variation of threshold voltage and frequency response between parts even from the same lot.
Because of this, the effects of signal integrity may not always affect different devices on the same board
in the same way. Sometimes, replacing a device appears to make signal integrity problems go away, but
this is just masking the problem. Different parts on identical boards will exhibit the same problem sooner
or later. It is important to fix signal integrity problems early. Unless the signal integrity problems are
severe enough to completely block all communication between the device and the programmer, they
may show up as subtle problems. Some of the FlashPro4/3/3X exit codes that are caused by signal
integrity problems are listed below. Signal integrity problems are not the only possible cause of these
Revision 5
271
In-System Programming (ISP) of Microsemi’s Low Power Flash Devices Using FlashPro4/3/3X
errors, but this list is intended to show where problems can occur. FlashPro4/3/3X allows TCK to be
lowered from 6 MHz down to 1 MHz to allow you to address some signal integrity problems that may
occur with impedance mismatching at higher frequencies. Customers are expected to troubleshoot
board-level signal integrity issues by measuring voltages and taking scope plots.
Scan Chain Failure
Normally, the FlashPro4/3/3X Scan Chain command expects to see 0x1 on the TDO pin. If the command
reports reading 0x0 or 0x3, it is seeing the TDO pin stuck at 0 or 1. The only time the TDO pin comes out
of tristate is when the JTAG TAP state machine is in the Shift-IR or Shift-DR state. If noise or reflections
on the TCK or TMS lines have disrupted the correct state transitions, the device's TAP state controller
might not be in one of these two states when the programmer tries to read the device. When this
happens, the output is floating when it is read and does not match the expected data value. This can also
be caused by a broken TDO net. Only a small amount of data is read from the device during the Scan
Chain command, so marginal problems may not always show up during this command. Occasionally a
faulty programmer can cause intermittent scan chain failures.
Exit 11
This error occurs during the verify stage of programming a device. After programming the design into the
device, the device is verified to ensure it is programmed correctly. The verification is done by shifting the
programming data into the device. An internal comparison is performed within the device to verify that all
switches are programmed correctly. Noise induced by poor signal integrity can disrupt the writes and
reads or the verification process and produce a verification error. While technically a verification error, the
root cause is often related to signal integrity.
Refer to the FlashPro User's Guide for other error messages and solutions. For the most up-to-date
known issues and solutions, refer to http://www.microsemi.com/soc/support.
Conclusion
IGLOO, ProASIC3, SmartFusion, and Fusion devices offer a low-cost, single-chip solution that is live at
power-up through nonvolatile flash technology. The FlashLock Pass Key and 128-bit AES Key security
features enable secure ISP in an untrusted environment. On-chip FlashROM enables a host of new
applications, including device serialization, subscription-based applications, and IP addressing.
Additionally, as the FlashROM is nonvolatile, all of these services can be provided without battery
backup.
Related Documents
User’s Guides
FlashPro User's Guide
http://www.microsemi.com/soc/documents/flashpro_ug.pdf
272
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
August 2012
June 2011
Changes
Page
This chapter will now be published standalone as an application note in addition to
being part of the IGLOO/ProASIC3/Fusion FPGA fabric user’s guides (SAR 38769).
N/A
The "ISP Programming Header Information" section was revised to update the
description of FP3-10PIN-ADAPTER-KIT in Table 12-3 • Programming Header
Ordering Codes, clarifying that it is the adapter kit used for ProASICPLUS based
boards, and also for ProASIC3 based boards where a compact programming
header is being used (SAR 36779).
269
The VPUMP programming mode voltage was corrected in Table 12-2 • Power
Supplies. The correct value is 3.15 V to 3.45 V (SAR 30668).
263
The notes associated with Figure 12-5 • Programming Header (top view) and 269, 271
Figure 12-6 • Board Layout and Programming Header Top View were revised to
make clear the fact that IGLOO nano V2 devices can be programmed at 1.2 V (SAR
30787).
July 2010
Figure 12-6 • Board Layout and Programming Header Top View was revised to
include resistors tying TCK and TRST to GND. Microsemi recommends tying off
TCK and TRST to GND if JTAG is not used (SAR 22921). RT ProASIC3 was added
to the list of device families.
271
In the "ISP Programming Header Information" section, the kit for adapting
ProASICPLUS devices was changed from FP3-10PIN-ADAPTER-KIT to FP3-26PINADAPTER-KIT (SAR 20878).
269
This chapter is no longer published separately with its own part number and version
but is now part of several FPGA fabric user’s guides.
N/A
References to FlashPro4 and FlashPro3X were added to this chapter, giving
distinctions between them. References to SmartGen were deleted and replaced
with Libero IDE Catalog.
N/A
The "ISP Architecture" section was revised to indicate that V2 devices can be
programmed at 1.2 V VCC with FlashPro4.
261
SmartFusion was added to Table 12-1 • Flash-Based FPGAs Supporting ISP.
262
The "Programming Voltage (VPUMP) and VJTAG" section was revised and 1.2 V
was added to Table 12-2 • Power Supplies.
263
The "Nonvolatile Memory (NVM) Programming Voltage" section is new.
263
Cortex-M3 was added to the "Cortex-M1 and Cortex-M3 Device Security" section.
265
In the "ISP Programming Header Information" section, the additional header
adapter ordering number was changed from FP3-26PIN-ADAPTER to FP3-10PINADAPTER-KIT, which contains 26-pin migration capability.
269
The description of NC was updated in Figure 12-5 • Programming Header (top 269, 270
view), Table 12-4 • Programming Header Pin Numbers and Description and
Figure 12-6 • Board Layout and Programming Header Top View.
The "Symptoms of a Signal Integrity Problem" section was revised to add that
customers are expected to troubleshoot board-level signal integrity issues by
measuring voltages and taking scope plots. "FlashPro4/3/3X allows TCK to be
lowered from 6 MHz down to 1 MHz to allow you to address some signal integrity
problems" formerly read, "from 24 MHz down to 1 MHz." "The Scan Chain
command expects to see 0x2" was changed to 0x1.
Revision 5
271
273
In-System Programming (ISP) of Microsemi’s Low Power Flash Devices Using FlashPro4/3/3X
Date
Changes
Page
July 2010
(continued)
The "Chain Integrity Test Error Analyze Chain Failure" section was renamed to the
"Scan Chain Failure" section, and the Analyze Chain command was changed to
Scan Chain. It was noted that occasionally a faulty programmer can cause scan
chain failures.
272
v1.5
(August 2009)
The "CoreMP7 Device Security" section was removed from "Security in ARMEnabled Low Power Flash Devices", since M7-enabled devices are no longer
supported.
265
v1.4
(December 2008)
The "ISP Architecture" section was revised to include information about core
voltage for IGLOO V2 and ProASIC3L devices, as well as 50 mV increments
allowable in Designer software.
261
IGLOO nano and ProASIC3 nano devices were added to Table 12-1 • Flash-Based
FPGAs Supporting ISP.
262
A second capacitor was added to Figure 12-6 • Board Layout and Programming
Header Top View.
271
v1.3
(October 2008)
The "ISP Support in Flash-Based Devices" section was revised to include new
families and make the information more concise.
262
v1.2
(June 2008)
The following changes were made to the family descriptions in Table 12-1 • FlashBased FPGAs Supporting ISP:
262
•
ProASIC3L was updated to include 1.5 V.
•
The number of PLLs for ProASIC3E was changed from five to six.
The "ISP Architecture" section was updated to included the IGLOO PLUS family in
the discussion of family-specific support. The text, "When 1.2 V is used, the device
can be reprogrammed in-system at 1.5 V only," was revised to state, "Although the
device can operate at 1.2 V core voltage, the device can only be reprogrammed
when all supplies (VCC, VCCI, and VJTAG) are at 1.5 V."
261
The "ISP Support in Flash-Based Devices" section and Table 12-1 • Flash-Based
FPGAs Supporting ISP were updated to include the IGLOO PLUS family. The
"IGLOO Terminology" section and "ProASIC3 Terminology" section are new.
262
The "Security" section was updated to mention that 15 k gate devices do not have a
built-in 128-bit decryption core.
264
Table 12-2 • Power Supplies was revised to remove the Normal Operation column
and add a table note stating, "All supply voltages should be at 1.5 V or higher,
regardless of the setting during normal operation."
263
The "ISP Programming Header Information" section was revised to change
FP3-26PIN-ADAPTER to FP3-10PIN-ADAPTER-KIT. Table 12-3 • Programming
Header Ordering Codes was updated with the same change, as well as adding the
part number FFSD-05-D-06.00-01-N, a 10-pin cable with 50-mil-pitch sockets.
269
The "Board-Level Considerations" section was updated to describe connecting two
capacitors in parallel across VPUMP and GND for proper programming.
271
v1.0
(January 2008)
Information was added to the "Programming Voltage (VPUMP) and VJTAG" section
about the JTAG interface pin.
263
51900055-2/7.06
ACTgen was changed to SmartGen.
N/A
In Figure 12-6 • Board Layout and Programming Header Top View, the order of the
text was changed to:
271
v1.1
(March 2008)
VJTAG from the target board
VCCI from the target board
VCC from the target board
274
R e vi s i o n 5
13 – Core Voltage Switching Circuit for IGLOO and
ProASIC3L In-System Programming
Introduction
The IGLOO® and ProASIC®3L families offer devices that can be powered by either 1.5 V or, in the case
of V2 devices, a core supply voltage anywhere in the range of 1.2 V to 1.5 V, in 50 mV increments.
Since IGLOO and ProASIC3L devices are flash-based, they can be programmed and reprogrammed
multiple times in-system using Microsemi FlashPro3. FlashPro3 uses the JTAG standard interface (IEEE
1149.1) and STAPL file (defined in JESD 71 to support programming of programmable devices using
IEEE 1149.1) for in-system configuration/programming (IEEE 1532) of a device. Programming can also
be executed by other methods, such as an embedded microcontroller that follows the same standards
above.
All IGLOO and ProASIC3L devices must be programmed with the VCC core voltage at 1.5 V. Therefore,
applications using IGLOO or ProASIC3L devices powered by a 1.2 V supply must switch the core supply
to 1.5 V for in-system programming.
The purpose of this document is to describe an easy-to-use and cost-effective solution for switching the
core supply voltage from 1.2 V to 1.5 V during in-system programming for IGLOO and ProASIC3L
devices.
Revision 5
275
Core Voltage Switching Circuit for IGLOO and ProASIC3L In-System Programming
Microsemi’s Flash Families Support Voltage Switching Circuit
The flash FPGAs listed in Table 13-1 support the voltage switching circuit feature and the functions
described in this document.
Table 13-1 • Flash-Based FPGAs Supporting Voltage Switching Circuit
Family*
Series
IGLOO
ProASIC3
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO nano
The industry’s lowest-power, smallest-size solution
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V with Flash*Freeze technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO devices as listed
in Table 13-1. Where the information applies to only one product line or limited devices, these exclusions
will be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 13-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
276
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Circuit Description
All IGLOO devices as well as the ProASIC3L product family are available in two versions: V5 devices,
which are powered by a 1.5 V supply and V2 devices, which are powered by a supply anywhere in the
range of 1.2 V to 1.5 V in 50 mV increments. Applications that use IGLOO or ProASIC3L devices
powered by a 1.2 V core supply must have a mechanism that switches the core voltage from 1.2 V (or
other voltage below 1.5 V) to 1.5 V during in-system programming (ISP). There are several possible
techniques to meet this requirement. Microsemi recommends utilizing a linear voltage regulator, a
resistor voltage divider, and an N-Channel Digital FET to set the appropriate VCC voltage, as shown in
Figure 13-1.
Where 1.2 V is mentioned in the following text, the meaning applies to any voltage below the 1.5 V
range. Resistor values in the figures have been calculated for 1.2 V, so refer to power regulator
datasheets if a different core voltage is required.
The main component of Microsemi's recommended circuit is the LTC3025 linear voltage regulator from
LinearTech. The output voltage of the LTC3025 on the OUT pin is set by the ratio of two external
resistors, R37 and R38, in a voltage divider. The linear voltage regulator adjusts the voltage on the OUT
pin to maintain the ADJ pin voltage at 0.4 V (referenced to ground). By using an R38 value of 40.2 kΩ
and an R37 value of 80.6 kΩ, the output voltage on the OUT pin is 1.2 V. To achieve 1.5 V on the OUT
pin, R44 can be used in parallel with R38. The OUT pin can now be used as a switchable source for the
VCC supply. Refer to the LTC3025 Linear Voltage Regulator datasheet for more information.
In Figure 13-1, the N-Channel Digital FET is used to enable and disable R44. This FET is controlled by
the JTAG TRST signal driven by the FlashPro3 programmer. During programming of the device, the
TRST signal is driven HIGH by the FlashPro3, and turns the N-Channel Digital FET ON. When the FET is
ON, R44 becomes enabled as a parallel resistance to R38, which forces the regulator to set OUT to
1.5 V.
When the FlashPro3 is connected and not in programming mode or when it is not connected, the pulldown resistor, R10, will pull the TRST signal LOW. When this signal is LOW, the N-Channel Digital FET
is "open" and R44 is not part of the resistance seen by the LTC3025. The new resistance momentarily
changes the voltage value on the ADJ pin, which in turn causes the output of the LTC3025 to
compensate by setting OUT to 1.2 V. Now the device will run in regular active mode at the regular 1.2 V
core voltage.
Figure 13-1 • Circuit Diagram
Revision 5
277
Core Voltage Switching Circuit for IGLOO and ProASIC3L In-System Programming
Circuit Verification
The power switching circuit recommended above is implemented on Microsemi's Icicle board
(Figure 13-2). On the Icicle board, VJTAGENB is used to control the N-Channel Digital FET; however,
this circuit was modified to use TRST instead of VJTAGENB in this application. There are three important
aspects of this circuit that were verified:
1. The rise on VCC from 1.2 V to 1.5 V when TRST is HIGH
2. VCC rises to 1.5 V before programming begins.
3. VCC switches from 1.5 V to 1.2 V when TRST is LOW.
Verification Steps
1. The rise on VCC from 1.2 V to 1.5 V when TRST is HIGH.
VCC Signal
TRST Signal
Figure 13-2 • Core Voltage on the IGLOO AGL125-QNG132 Device
In the oscilloscope plots (Figure 13-2), the TRST from FlashPro3 and the VCC core voltage of the
IGLOO device are labeled. This plot shows the rise characteristic of the TRST signal from FlashPro3.
Once the TRST signal is asserted HIGH, the LTC3025 shown in Figure 13-1 on page 277 senses the
increase in voltage and changes the output from 1.2 V to 1.5 V. It takes the circuit approximately 100 µs
to respond to TRST and change the voltage to 1.5 V on the VCC core.
278
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
2. VCC rises to 1.5 V before programming begins.
TMS Signal Green
Floating Signal TDI/TMS
VCC Core Voltage
TRST Signal (purple)
TDI Signal (yellow)
Figure 13-3 • Programming Algorithm
The oscilloscope plot in Figure 13-3 shows a wider time interval for the programming algorithm and
includes the TDI and TMS signals from the FlashPro3. These signals carry the programming information
that is programmed into the device and should only start toggling after the VCC core voltage reaches 1.5
V. Again, TRST from FlashPro3 and the VCC core voltage of the IGLOO device are labeled. As shown in
Figure 13-3, TDI and TMS are floating initially, and the core voltage is 1.2 V. When a programming
command on the FlashPro3 is executed, TRST is driven HIGH and TDI is momentarily driven to ground.
In response to the HIGH TRST signal, the circuit responds and pulls the core voltage to 1.5 V. After
100 ms, TRST is briefly driven LOW by the FlashPro software. This is expected behavior that ensures
the device JTAG state machine is in Reset prior to programming. TRST remains HIGH for the duration of
the programming. It can be seen in Figure 13-3 that the VCC core voltage signal remains at 1.5 V for
approximately 50 ms before information starts passing through on TDI and TMS. This confirms that the
voltage switching circuit drives the VCC core supply voltage to 1.5 V prior to programming.
Revision 5
279
Core Voltage Switching Circuit for IGLOO and ProASIC3L In-System Programming
3. VCC switches from 1.5 V to 1.2 V when TRST is LOW.
TRST Signal
VCC Core Signal
Figure 13-4 • TRST Toggled LOW
In Figure 13-4, the TRST signal and the VCC core voltage signal are labeled. As TRST is pulled to
ground, the core voltage is observed to switch from 1.5 V to 1.2 V. The observed fall time is
approximately 2 ms.
DirectC
The above analysis is based on FlashPro3, but there are other solutions to ISP, such as DirectC. DirectC
is a microprocessor program that can be run in-system to program Microsemi flash devices. For
FlashPro3, TRST is the most convenient control signal to use for the recommended circuit. However, for
DirectC, users may use any signal to control the FET. For example, the DirectC code can be edited so
that a separate non-JTAG signal can be asserted from the microcontroller that signals the board that it is
about to start programming the device. After asserting the N-Channel Digital FET control signal, the
programming algorithm must allow sufficient time for the supply to rise to 1.5 V before initiating DirectC
programming. As seen in Figure 13-3 on page 279, 50 ms is adequate time. Depending on the size of
the PCB and the capacitance on the VCC supply, results may vary from system to system. Microsemi
recommends using a conservative value for the wait time to make sure that the VCC core voltage is at
the right level.
Conclusion
For applications using IGLOO and ProASIC3L low power FPGAs and taking advantage of the low core
voltage power supplies with less than 1.5 V operation, there must be a way for the core voltage to switch
from 1.2 V (or other voltage) to 1.5 V, which is required during in-system programming. The circuit
explained in this document illustrates one simple, cost-effective way of handling this requirement. A
JTAG signal from the FlashPro3 programmer allows the circuit to sense when programming is in
progress, enabling it to switch to the correct core voltage.
280
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
Changes
Page
July 2010
This chapter is no longer published separately with its own part number and version
but is now part of several FPGA fabric user’s guides.
N/A
v1.1
(October 2008)
The "Introduction" was revised to include information about the core supply voltage
range of operation in V2 devices.
275
IGLOO nano device support was added to Table 13-1 • Flash-Based FPGAs
Supporting Voltage Switching Circuit.
276
The "Circuit Description" section was updated to include IGLOO PLUS core
operation from 1.2 V to 1.5 V in 50 mV increments.
277
The "Microsemi’s Flash Families Support Voltage Switching Circuit" section was
revised to include new families and make the information more concise.
276
v1.0
(August 2008)
Revision 5
281
14 – Microprocessor Programming of Microsemi’s
Low Power Flash Devices
Introduction
The Fusion, IGLOO, and ProASIC3 families of flash FPGAs support in-system programming (ISP) with
the use of a microprocessor. Flash-based FPGAs store their configuration information in the actual cells
within the FPGA fabric. SRAM-based devices need an external configuration memory, and hybrid
nonvolatile devices store the configuration in a flash memory inside the same package as the SRAM
FPGA. Since the programming of a true flash FPGA is simpler, requiring only one stage, it makes sense
that programming with a microprocessor in-system should be simpler than with other SRAM FPGAs.
This reduces bill-of-materials costs and printed circuit board (PCB) area, and increases system reliability.
Nonvolatile flash technology also gives the low power flash devices the advantage of a secure, low
power, live-at-power-up, and single-chip solution. Low power flash devices are reprogrammable and offer
time-to-market benefits at an ASIC-level unit cost. These features enable engineers to create highdensity systems using existing ASIC or FPGA design flows and tools.
This document is an introduction to microprocessor programming only. To explain the difference between
the options available, user's guides for DirectC and STAPL provide more detail on implementing each
style.
Microprocessor
Internal/External
Memory Running
DirectC
Internal RAM
On-Board
Memory
Device
.dat file
I/O Functions
JTAG Bus
Flash
Device
Figure 14-1 • ISP Using Microprocessor
Revision 5
283
Microprocessor Programming of Microsemi’s Low Power Flash Devices
Microprocessor Programming Support in Flash Devices
The flash-based FPGAs listed in Table 14-1 support programming with a microprocessor and the
functions described in this document.
Table 14-1 • Flash-Based FPGAs
Family*
Series
IGLOO
ProASIC3
Fusion
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO nano
The industry’s lowest-power, smallest-size solution
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
ProASIC3
Low power, high-performance 1.5 V FPGAs
ProASIC3E
Higher density ProASIC3 FPGAs with six PLLs and additional I/O standards
ProASIC3 nano
Lowest-cost solution with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V with Flash*Freeze technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Automotive ProASIC3
ProASIC3 FPGAs qualified for automotive applications
Fusion
Mixed signal FPGA integrating ProASIC3 FPGA fabric, programmable
analog block, support for ARM® Cortex™-M1 soft processors, and flash
memory into a monolithic device
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO devices as listed
in Table 14-1. Where the information applies to only one device or limited devices, these exclusions will
be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 14-1. Where the information applies to only one device or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
284
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Programming Algorithm
JTAG Interface
The low power flash families are fully compliant with the IEEE 1149.1 (JTAG) standard. They support all
the mandatory boundary scan instructions (EXTEST, SAMPLE/PRELOAD, and BYPASS) as well as six
optional public instructions (USERCODE, IDCODE, HIGHZ, and CLAMP).
IEEE 1532
The low power flash families are also fully compliant with the IEEE 1532 programming standard. The
IEEE 1532 standard adds programming instructions and associated data registers to devices that comply
with the IEEE 1149.1 standard (JTAG). These instructions and registers extend the capabilities of the
IEEE 1149.1 standard such that the Test Access Port (TAP) can be used for configuration activities. The
IEEE 1532 standard greatly simplifies the programming algorithm, reducing the amount of time needed
to implement microprocessor ISP.
Implementation Overview
To implement device programming with a microprocessor, the user should first download the C-based
STAPL player or DirectC code from the Microsemi SoC Products Group website. Refer to the website for
future updates regarding the STAPL player and DirectC code.
http://www.microsemi.com/soc/download/program_debug/stapl/default.aspx
http://www.microsemi.com/soc/download/program_debug/directc/default.aspx
Using the easy-to-follow user's guide, create the low-level application programming interface (API) to
provide the necessary basic functions. These API functions act as the interface between the
programming software and the actual hardware (Figure 14-2).
Programming
Algorithm and Data
STAPL File
STAPL Player or DirectC
Programming
Software
I/O and Memory
Functions
API
Figure 14-2 • Device Programming Code Relationship
The API is then linked with the STAPL player or DirectC and compiled using the microprocessor's
compiler. Once the entire code is compiled, the user must download the resulting binary into the MCU
system's program memory (such as ROM, EEPROM, or flash). The system is now ready for
programming.
To program a design into the FPGA, the user creates a bitstream or STAPL file using the Microsemi
Designer software, downloads it into the MCU system's volatile memory, and activates the stored
programming binary file (Figure 14-3 on page 286). Once the programming is completed, the bitstream
or STAPL file can be removed from the system, as the configuration profile is stored in the flash FPGA
fabric and does not need to be reloaded at every system power-on.
Revision 5
285
Microprocessor Programming of Microsemi’s Low Power Flash Devices
Programming
Software
Source Code
Programming
File
Microprocessor Compiler
BIN File
Download to System
Program Device
Figure 14-3 • MCU FPGA Programming Model
FlashROM
Microsemi low power flash devices have 1 kbit of user-accessible, nonvolatile, FlashROM on-chip. This
nonvolatile FlashROM can be programmed along with the core or on its own using the standard IEEE
1532 JTAG programming interface.
The FlashROM is architected as eight pages of 128 bits. Each page can be individually programmed
(erased and written). Additionally, on-chip AES security decryption can be used selectively to load data
securely into the FlashROM (e.g., over public or private networks, such as the Internet). Refer to the
"FlashROM in Microsemi’s Low Power Flash Devices" section on page 117.
286
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
STAPL vs. DirectC
Programming the low power flash devices is performed using DirectC or the STAPL player. Both tools
use the STAPL file as an input. DirectC is a compiled language, whereas STAPL is an interpreted
language. Microprocessors will be able to load the FPGA using DirectC much more quickly than STAPL.
This speed advantage becomes more apparent when lower clock speeds of 8- or 16-bit microprocessors
are used. DirectC also requires less memory than STAPL, since the programming algorithm is directly
implemented. STAPL does have one advantage over DirectC—the ability to upgrade. When a new
programming algorithm is required, the STAPL user simply needs to regenerate a STAPL file using the
latest version of the Designer software and download it to the system. The DirectC user must download
the latest version of DirectC from Microsemi, compile everything, and download the result into the system
(Figure 14-4).
STAPL Flow
DirectC Flow
Generate the
New STAPL File
DirectC Source Code
Download to System
Microprocessor
Compiler
Program Device
BIN File
Input STAPL File
Download to System
Program Device
Figure 14-4 • STAPL vs. DirectC
Revision 5
287
Microprocessor Programming of Microsemi’s Low Power Flash Devices
Remote Upgrade via TCP/IP
Transmission Control Protocol (TCP) provides a reliable bitstream transfer service between two
endpoints on a network. TCP depends on Internet Protocol (IP) to move packets around the network on
its behalf. TCP protects against data loss, data corruption, packet reordering, and data duplication by
adding checksums and sequence numbers to transmitted data and, on the receiving side, sending back
packets and acknowledging the receipt of data.
The system containing the low power flash device can be assigned an IP address when deployed in the
field. When the device requires an update (core or FlashROM), the programming instructions along with
the new programming data (AES-encrypted cipher text) can be sent over the Internet to the target system
via the TCP/IP protocol. Once the MCU receives the instruction and data, it can proceed with the FPGA
update. Low power flash devices support Message Authentication Code (MAC), which can be used to
validate data for the target device. More details are given in the "Message Authentication Code (MAC)
Validation/Authentication" section.
Hardware Requirement
To facilitate the programming of the low power flash families, the system must have a microprocessor
(with access to the device JTAG pins) to process the programming algorithm, memory to store the
programming algorithm, programming data, and the necessary programming voltage. Refer to the
relevant datasheet for programming voltages.
Security
Encrypted Programming
As an additional security measure, the devices are equipped with AES decryption. AES works in two
steps. The first step is to program a key into the devices in a secure or trusted programming center (such
as Microsemi SoC Products Group In-House Programming (IHP) center). The second step is to encrypt
any programming files with the same encryption key. The encrypted programming file will only work with
the devices that have the same key. The AES used in the low power flash families is the 128-bit AES
decryption engine (Rijndael algorithm).
Message Authentication Code (MAC) Validation/Authentication
As part of the AES decryption flow, the devices are equipped with a MAC validation/authentication
system. MAC is an authentication tag, also called a checksum, derived by applying an on-chip
authentication scheme to a STAPL file as it is loaded into the FPGA. MACs are computed and verified
with the same key so they can only be verified by the intended recipient. When the MCU system receives
the AES-encrypted programming data (cipher text), it can validate the data by loading it into the FPGA
and performing a MAC verification prior to loading the data, via a second programming pass, into the
FPGA core cells. This prevents erroneous or corrupt data from getting into the FPGA.
Low power flash devices with AES and MAC are superior to devices with only DES or 3DES encryption.
Because the MAC verifies the correctness of the data, the FPGA is protected from erroneous loading of
invalid programming data that could damage a device (Figure 14-5 on page 289).
The AES with MAC enables field updates over public networks without fear of having the design stolen.
An encrypted programming file can only work on devices with the correct key, rendering any stolen files
288
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
useless to the thief. To learn more about the low power flash devices’ security features, refer to the
"Security in Low Power Flash Devices" section on page 235.
ProASIC3
MAC
Validation
Designer Software
AES KEY
AES
Encryption
Decrypted Stream
AES
Decryption
Programming
Control
TCP/IP
Public Network
Encrypted Stream
Encrypted Stream
Figure 14-5 • ProASIC3 Device Encryption Flow
Conclusion
The Fusion, IGLOO, and ProASIC3 FPGAs are ideal for applications that require field upgrades. The
single-chip devices save board space by eliminating the need for EEPROM. The built-in AES with MAC
enables transmission of programming data over any network without fear of design theft. Fusion, IGLOO,
and ProASIC3 FPGAs are IEEE 1532–compliant and support STAPL, making the target programming
software easy to implement.
Revision 5
289
Microprocessor Programming of Microsemi’s Low Power Flash Devices
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
Changes
Page
September 2012
The "Security" section was modified to clarify that Microsemi does not support
read-back of FPGA core-programmed data (SAR 41235).
288
July 2010
This chapter is no longer published separately with its own part number and
version but is now part of several FPGA fabric user’s guides.
N/A
v1.4
(December 2008)
IGLOO nano and ProASIC3 nano devices were added to Table 14-1 • FlashBased FPGAs.
284
v1.3
(October 2008)
The "Microprocessor Programming Support in Flash Devices" section was
revised to include new families and make the information more concise.
284
v1.2
(June 2008)
The following changes were made to the family descriptions in Table 14-1 •
Flash-Based FPGAs:
284
v1.1
(March 2008)
290
•
ProASIC3L was updated to include 1.5 V.
•
The number of PLLs for ProASIC3E was changed from five to six.
The "Microprocessor Programming Support in Flash Devices" section was
updated to include information on the IGLOO PLUS family. The "IGLOO
Terminology" section and "ProASIC3 Terminology" section are new.
R e vi s i o n 5
284
15 – Boundary Scan in Low Power Flash Devices
Boundary Scan
Low power flash devices are compatible with IEEE Standard 1149.1, which defines a hardware
architecture and the set of mechanisms for boundary scan testing. JTAG operations are used during
boundary scan testing.
The basic boundary scan logic circuit is composed of the TAP controller, test data registers, and
instruction register (Figure 15-2 on page 294).
Low power flash devices support three types of test data registers: bypass, device identification, and
boundary scan. The bypass register is selected when no other register needs to be accessed in a device.
This speeds up test data transfer to other devices in a test data path. The 32-bit device identification
register is a shift register with four fields (LSB, ID number, part number, and version). The boundary scan
register observes and controls the state of each I/O pin. Each I/O cell has three boundary scan register
cells, each with serial-in, serial-out, parallel-in, and parallel-out pins.
TAP Controller State Machine
The TAP controller is a 4-bit state machine (16 states) that operates as shown in Figure 15-1.
The 1s and 0s represent the values that must be present on TMS at a rising edge of TCK for the given
state transition to occur. IR and DR indicate that the instruction register or the data register is operating in
that state.
The TAP controller receives two control inputs (TMS and TCK) and generates control and clock signals
for the rest of the test logic architecture. On power-up, the TAP controller enters the Test-Logic-Reset
state. To guarantee a reset of the controller from any of the possible states, TMS must remain HIGH for
five TCK cycles. The TRST pin can also be used to asynchronously place the TAP controller in the TestLogic-Reset state.
1
TEST_LOGIC_RESET
0
0
RUN_TEST_IDLE
1
SELECT_DR
1
SELECT_IR
0
0
CAPTURE_DR
1
1
CAPTURE_IR
0
0
0
SHIFT_DR
1
EXIT1_DR
EXIT1_IR
0
0
0
PAUSE_DR
1
0
PAUSE_IR
1
1
EXIT2_DR
0
0
EXIT2_IR
1
1
UPDATE_DR
1
0
SHIFT_IR
1
1
1
0
UPDATE_IR
1
0
Figure 15-1 • TAP Controller State Machine
Revision 5
291
Boundary Scan in Low Power Flash Devices
Microsemi’s Flash Devices Support the JTAG Feature
The flash-based FPGAs listed in Table 15-1 support the JTAG feature and the functions described in this
document.
Table 15-1 • Flash-Based FPGAs
Family*
Series
IGLOO
ProASIC3
Fusion
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO nano
The industry’s lowest-power, smallest-size solution
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
ProASIC3
Low power, high-performance 1.5 V FPGAs
ProASIC3E
Higher density ProASIC3 FPGAs with six PLLs and additional I/O standards
ProASIC3 nano
Lowest-cost solution with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V with Flash*Freeze technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Automotive ProASIC3
ProASIC3 FPGAs qualified for automotive applications
Fusion
Mixed signal FPGA integrating ProASIC®3 FPGA fabric, programmable
analog block, support for ARM® Cortex™-M1 soft processors, and flash
memory into a monolithic device
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO devices as listed
in Table 15-1. Where the information applies to only one product line or limited devices, these exclusions
will be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 15-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
292
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Boundary Scan Support in Low Power Devices
The information in this document applies to all Fusion, IGLOO, and ProASIC3 devices. For IGLOO,
IGLOO PLUS, and ProASIC3L devices, the Flash*Freeze pin must be deasserted for successful
boundary scan operations. Devices cannot enter JTAG mode directly from Flash*Freeze mode.
Boundary Scan Opcodes
Low power flash devices support all mandatory IEEE 1149.1 instructions (EXTEST, SAMPLE/PRELOAD,
and BYPASS) and the optional IDCODE instruction (Table 15-2).
Table 15-2 • Boundary Scan Opcodes
Hex Opcode
EXTEST
00
HIGHZ
07
USERCODE
0E
SAMPLE/PRELOAD
01
IDCODE
0F
CLAMP
05
BYPASS
FF
Boundary Scan Chain
The serial pins are used to serially connect all the boundary scan register cells in a device into a
boundary scan register chain (Figure 15-2 on page 294), which starts at the TDI pin and ends at the TDO
pin. The parallel ports are connected to the internal core logic I/O tile and the input, output, and control
ports of an I/O buffer to capture and load data into the register to control or observe the logic state of
each I/O.
Each test section is accessed through the TAP, which has five associated pins: TCK (test clock input),
TDI, TDO (test data input and output), TMS (test mode selector), and TRST (test reset input). TMS, TDI,
and TRST are equipped with pull-up resistors to ensure proper operation when no input data is supplied
to them. These pins are dedicated for boundary scan test usage. Refer to the "JTAG Pins" section in the
"Pin Descriptions and Packaging" chapter of the appropriate device datasheet for pull-up/-down
recommendations for TCK and TRST pins. Pull-down recommendations are also given in Table 15-3 on
page 294
Revision 5
293
Boundary Scan in Low Power Flash Devices
I/O
I/O
I/O
I/O
I/O
TDI
Test Data
Registers
Instruction
Register
TAP
Controller
Device
Logic
TDO
I/O
TRST
I/O
TMS
I/O
TCK
I/O
Bypass Register
I/O
I/O
I/O
I/O
I/O
Figure 15-2 • Boundary Scan Chain
Board-Level Recommendations
Table 15-3 gives pull-down recommendations for the TRST and TCK pins.
Table 15-3 • TRST and TCK Pull-Down Recommendations
VJTAG
Tie-Off Resistance*
VJTAG at 3.3 V
200 Ω to 1 kΩ
VJTAG at 2.5 V
200 Ω to 1 kΩ
VJTAG at 1.8 V
500 Ω to 1 kΩ
VJTAG at 1.5 V
500 Ω to 1 kΩ
VJTAG at 1.2 V
TBD
Note: Equivalent parallel resistance if more than one device is on JTAG chain (Figure 15-3)
294
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
1.5 V
JTAG
Header
VJTAG
TRST
GND
TCK
2 kΩ
TDO
TDI
Microsemi
FPGA 1
TDO
TDI
Microsemi
FPGA 2
1.5 kΩ
2 kΩ
1.5 kΩ
2 kΩ
TDO
TDI
Microsemi
FPGA3
TDO
TDI
Microsemi
FPGA 4
1.5 kΩ
2 kΩ
1.5 kΩ
Note: TCK is correctly wired with an equivalent tie-off resistance of 500 Ω, which satisfies the table for
VJTAG of 1.5 V. The resistor values for TRST are not appropriate in this case, as the tie-off
resistance of 375 Ω is below the recommended minimum for VJTAG = 1.5 V, but would be
appropriate for a VJTAG setting of 2.5 V or 3.3 V.
Figure 15-3 • Parallel Resistance on JTAG Chain of Devices
Advanced Boundary Scan Register Settings
You will not be able to control the order in which I/Os are released from boundary scan control. Testing
has produced cases where, depending on I/O placement and FPGA routing, a 5 ns glitch has been seen
on exiting programming mode. The following setting is recommended to prevent such I/O glitches:
1. In the FlashPro software, configure the advanced BSR settings for Specify I/O Settings During
Programming.
2. Set the input BSR cell to Low for the input I/O.
Revision 5
295
Boundary Scan in Low Power Flash Devices
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
Changes
Page
In the "Boundary Scan Chain" section, the reference made to the datasheet for
pull-up/-down recommendations was changed to mention TCK and TRST pins
rather than TDO and TCK pins. TDO is an output, so no pull resistor is needed
(SAR 35937).
293
The "Advanced Boundary Scan Register Settings" section is new (SAR 38432).
295
This chapter is no longer published separately with its own part number and version
but is now part of several FPGA fabric user’s guides.
N/A
Table 15-3 • TRST and TCK Pull-Down Recommendations was revised to add
VJTAG at 1.2 V.
294
v1.4
(December 2008)
IGLOO nano and ProASIC3 nano devices were added to Table 15-1 • Flash-Based
FPGAs.
292
v1.3
(October 2008)
The "Boundary Scan Support in Low Power Devices" section was revised to include
new families and make the information more concise.
293
v1.2
(June 2008)
The following changes were made to the family descriptions in Table 15-1 • FlashBased FPGAs:
292
August 2012
July 2010
v1.1
(March 2008)
296
•
ProASIC3L was updated to include 1.5 V.
•
The number of PLLs for ProASIC3E was changed from five to six.
The chapter was updated to include the IGLOO PLUS family and information
regarding 15 k gate devices.
N/A
The "IGLOO Terminology" section and "ProASIC3 Terminology" section are new.
292
R e vi s i o n 5
16 – UJTAG Applications in Microsemi’s Low
Power Flash Devices
Introduction
In Fusion, IGLOO, and ProASIC3 devices, there is bidirectional access from the JTAG port to the core
VersaTiles during normal operation of the device (Figure 16-1). User JTAG (UJTAG) is the ability for the
design to use the JTAG ports for access to the device for updates, etc. While regular JTAG is used, the
UJTAG tiles, located at the southeast area of the die, are directly connected to the JTAG Test Access
Port (TAP) Controller in normal operating mode. As a result, all the functional blocks of the device, such
as Clock Conditioning Circuits (CCCs) with PLLs, SRAM blocks, embedded FlashROM, flash memory
blocks, and I/O tiles, can be reached via the JTAG ports. The UJTAG functionality is available by
instantiating the UJTAG macro directly in the source code of a design. Access to the FPGA core
VersaTiles from the JTAG ports enables users to implement different applications using the TAP
Controller (JTAG port). This document introduces the UJTAG tile functionality and discusses a few
application examples. However, the possible applications are not limited to what is presented in this
document. UJTAG can serve different purposes in many designs as an elementary or auxiliary part of the
design. For detailed usage information, refer to the "Boundary Scan in Low Power Flash Devices"
section on page 291.
UJTAG
Address Generation and
Data Serlialization
UIREG[7:0]
TDO
TDI
TMS
TCK
TRST
Enable
FROM
RESET
URSTB
UDRUPD
Control
UDRCK
CLK
UDRCAP
SDI
UDRSH
Addr[6:0]
Addr [6:0]
Data[7:0]
Data[7:0]
SDO
UTDI
UTDO
Figure 16-1 • Block Diagram of Using UJTAG to Read FlashROM Contents
Revision 5
297
UJTAG Applications in Microsemi’s Low Power Flash Devices
UJTAG Support in Flash-Based Devices
The flash-based FPGAs listed in Table 16-1 support the UJTAG feature and the functions described in
this document.
Table 16-1 • Flash-Based FPGAs
Family*
Series
IGLOO
ProASIC3
Fusion
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO nano
The industry’s lowest-power, smallest-size solution
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
ProASIC3
Low power, high-performance 1.5 V FPGAs
ProASIC3E
Higher density ProASIC3 FPGAs with six PLLs and additional I/O standards
ProASIC3 nano
Lowest-cost solution with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V with Flash*Freeze technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Automotive ProASIC3
ProASIC3 FPGAs qualified for automotive applications
Fusion
Mixed signal FPGA integrating ProASIC3 FPGA fabric, programmable
analog block, support for ARM® Cortex™-M1 soft processors, and flash
memory into a monolithic device
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO devices as listed
in Table 16-1. Where the information applies to only one product line or limited devices, these exclusions
will be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 16-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
298
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
UJTAG Macro
The UJTAG tiles can be instantiated in a design using the UJTAG macro from the Fusion, IGLOO, or
ProASIC3 macro library. Note that "UJTAG" is a reserved name and cannot be used for any other userdefined blocks. A block symbol of the UJTAG tile macro is presented in Figure 16-2. In this figure, the
ports on the left side of the block are connected to the JTAG TAP Controller, and the right-side ports are
accessible by the FPGA core VersaTiles. The TDI, TMS, TDO, TCK, and TRST ports of UJTAG are only
provided for design simulation purposes and should be treated as external signals in the design netlist.
However, these ports must NOT be connected to any I/O buffer in the netlist. Figure 16-3 on page 300
illustrates the correct connection of the UJTAG macro to the user design netlist. Microsemi Designer
software will automatically connect these ports to the TAP during place-and-route. Table 16-2 gives the
port descriptions for the rest of the UJTAG ports:
Table 16-2 • UJTAG Port Descriptions
Port
Description
UIREG [7:0] This 8-bit bus carries the contents of the JTAG Instruction Register of each device. Instruction Register
values 16 to 127 are not reserved and can be employed as user-defined instructions.
URSTB
URSTB is an active-low signal and will be asserted when the TAP Controller is in Test-Logic-Reset
mode. URSTB is asserted at power-up, and a power-on reset signal resets the TAP Controller. URSTB
will stay asserted until an external TAP access changes the TAP Controller state.
UTDI
This port is directly connected to the TAP's TDI signal.
UTDO
This port is the user TDO output. Inputs to the UTDO port are sent to the TAP TDO output MUX when
the IR address is in user range.
UDRSH
Active-high signal enabled in the ShiftDR TAP state
UDRCAP
Active-high signal enabled in the CaptureDR TAP state
UDRCK
This port is directly connected to the TAP's TCK signal.
UDRUPD
Active-high signal enabled in the UpdateDR TAP state
TDO
TDI
TMS
UIREG0
UIREG1
UIREG2
UIREG3
UIREG4
UIREG5
UIREG6
UIREG7
URSTB
UDRUPD
TCK
TRST
UDRCK
UDRCAP
UDRSH
UTDI
UTDO
Figure 16-2 • UJTAG Tile Block Symbol
Revision 5
299
UJTAG Applications in Microsemi’s Low Power Flash Devices
a) CORRECT Instantiation
UIREG[7:0]
TDO
URSTB
TDI
INPUTS
UDRUPD
UDRCK
TMS
UDRCAP
TCK
FPGA
VersaTiles
UDRSH
TRST
UTDI
UTDO
OUTPUTS
b) INCORRECT Instantiation
TDO
UIREG[7:0]
URSTB
TDI
INPUTS
UDRUPD
TMS
TCK
UDRCK
UDRCAP
FPGA
VersaTiles
UDRSH
TRST
UTDI
UTDO
OUTPUTS
Note: Do not connect JTAG pins (TDO, TDI, TMS, TCK, or TRST) to I/Os in the design.
Figure 16-3 • Connectivity Method of UJTAG Macro
UJTAG Operation
There are a few basic functions of the UJTAG macro that users must understand before designing with it.
The most important fundamental concept of the UJTAG design is its connection with the TAP Controller
state machine.
TAP Controller State Machine
The 16 states of the TAP Controller state machine are shown in Figure 16-4 on page 301. The 1s and 0s,
shown adjacent to the state transitions, represent the TMS values that must be present at the time of a
rising TCK edge for a state transition to occur. In the states that include the letters "IR," the instruction
register operates; in the states that contain the letters "DR," the test data register operates. The TAP
Controller receives two control inputs, TMS and TCK, and generates control and clock signals for the rest
of the test logic.
On power-up (or the assertion of TRST), the TAP Controller enters the Test-Logic-Reset state. To reset
the controller from any other state, TMS must be held HIGH for at least five TCK cycles. After reset, the
TAP state changes at the rising edge of TCK, based on the value of TMS.
300
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
1
Test_Logic_Reset
0
0
Run_Test/
Idle
1
Select_
DR_Scan
0
1
1
Select_
IR_Scan
0
1
Capture_DR
Capture_IR
0
0
0
Shift_DR
1
Exit1_DR
1
Exit1_IR
0
1
0
Pause_DR
0
Pause_IR
1
0
1
Exit2_DR
0
Exit2_IR
1
1
Update_DR
1
0
Shift_IR
1
0
1
0
Update_IR
1
0
Figure 16-4 • TAP Controller State Diagram
UJTAG Port Usage
UIREG[7:0] hold the contents of the JTAG instruction register. The UIREG vector value is updated when
the TAP Controller state machine enters the Update_IR state. Instructions 16 to 127 are user-defined and
can be employed to encode multiple applications and commands within an application. Loading new
instructions into the UIREG vector requires users to send appropriate logic to TMS to put the TAP
Controller in a full IR cycle starting from the Select IR_Scan state and ending with the Update_IR state.
UTDI, UTDO, and UDRCK are directly connected to the JTAG TDI, TDO, and TCK ports, respectively.
The TDI input can be used to provide either data (TAP Controller in the Shift_DR state) or the new
contents of the instruction register (TAP Controller in the Shift_IR state).
UDRSH, UDRUPD, and UDRCAP are HIGH when the TAP Controller state machine is in the Shift_DR,
Update_DR, and Capture_DR states, respectively. Therefore, they act as flags to indicate the stages of
the data shift process. These flags are useful for applications in which blocks of data are shifted into the
design from JTAG pins. For example, an active UDRSH can indicate that UTDI contains the data
bitstream, and UDRUPD is a candidate for the end-of-data-stream flag.
As mentioned earlier, users should not connect the TDI, TDO, TCK, TMS, and TRST ports of the UJTAG
macro to any port or net of the design netlist. The Designer software will automatically handle the port
connection.
Revision 5
301
UJTAG Applications in Microsemi’s Low Power Flash Devices
Typical UJTAG Applications
Bidirectional access to the JTAG port from VersaTiles—without putting the device into test mode—
creates flexibility to implement many different applications. This section describes a few of these. All are
based on importing/exporting data through the UJTAG tiles.
Clock Conditioning Circuitry—Dynamic Reconfiguration
In low power flash devices, CCCs, which include PLLs, can be configured dynamically through either an
81-bit embedded shift register or static flash programming switches. These 81 bits control all the
characteristics of the CCC: routing MUX architectures, delay values, divider values, etc. Table 16-3 lists
the 81 configuration bits in the CCC.
Table 16-3 • Configuration Bits of Fusion, IGLOO, and ProASIC3 CCC Blocks
Bit Number(s)
Control Function
80
RESET ENABLE
79
DYNCSEL
78
DYNBSEL
77
DYNASEL
VCOSEL [2:0]
73
STATCSEL
72
STATBSEL
71
STATASEL
DLYC [4:0]
DLYB {4:0]
DLYGLC [4:0]
DLYGLB [4:0]
DLYGLA [4:0]
45
XDLYSEL
FBDLY [4:0]
FBSEL
OCMUX [2:0]
OBMUX [2:0]
OAMUX [2:0]
OCDIV [4:0]
OBDIV [4:0]
OADIV [4:0]
FBDIV [6:0]
FINDIV [6:0]
The embedded 81-bit shift register (for the dynamic configuration of the CCC) is accessible to the
VersaTiles, which, in turn, have access to the UJTAG tiles. Therefore, the CCC configuration shift
register can receive and load the new configuration data stream from JTAG.
Dynamic reconfiguration eliminates the need to reprogram the device when reconfiguration of the CCC
functional blocks is needed. The CCC configuration can be modified while the device continues to
operate. Employing the UJTAG core requires the user to design a module to provide the configuration
data and control the CCC configuration shift register. In essence, this is a user-designed TAP Controller
requiring chip resources.
Similar reconfiguration capability exists in the ProASICPLUS® family. The only difference is the number of
shift register bits controlling the CCC (27 in ProASICPLUS and 81 in IGLOO, ProASIC3, and Fusion).
302
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Fine Tuning
In some applications, design constants or parameters need to be modified after programming the original
design. The tuning process can be done using the UJTAG tile without reprogramming the device with
new values. If the parameters or constants of a design are stored in distributed registers or embedded
SRAM blocks, the new values can be shifted onto the JTAG TAP Controller pins, replacing the old
values. The UJTAG tile is used as the “bridge” for data transfer between the JTAG pins and the FPGA
VersaTiles or SRAM logic. Figure 16-5 shows a flow chart example for fine-tuning application steps using
the UJTAG tile.
In Figure 16-5, the TMS signal sets the TAP Controller state machine to the appropriate states. The flow
mainly consists of two steps: a) shifting the defined instruction and b) shifting the new data. If the target
parameter is constantly used in the design, the new data can be shifted into a temporary shift register
from UTDI. The UDRSH output of UJTAG can be used as a shift-enable signal, and UDRCK is the shift
clock to the shift register. Once the shift process is completed and the TAP Controller state is moved to
the Update_DR state, the UDRUPD output of the UJTAG can latch the new parameter value from the
temporary register into a permanent location. This avoids any interruption or malfunctioning during the
serial shift of the new value.
TAP Controller in
Test_Logic_Reset
State
Set TAP state to
SHIFT_DR
Set TAP state to
SHIFT_IR
Shift data into TDI and
record UTDI in a shift
register
Shift the user-defined
instruction of tuning
application
Set TAP state in
Update_DR
Set TAP state to
Update_IR
UIREG Equal to
the user-defined
instruction
Yes
Latch the recorded data
onto the location of stored
parameter
No
Figure 16-5 • Flow Chart Example of Fine-Tuning an Application Using UJTAG
Revision 5
303
UJTAG Applications in Microsemi’s Low Power Flash Devices
Silicon Testing and Debugging
In many applications, the design needs to be tested, debugged, and verified on real silicon or in the final
embedded application. To debug and test the functionality of designs, users may need to monitor some
internal logic (or nets) during device operation. The approach of adding design test pins to monitor the
critical internal signals has many disadvantages, such as limiting the number of user I/Os. Furthermore,
adding external I/Os for test purposes may require additional or dedicated board area for testing and
debugging.
The UJTAG tiles of low power flash devices offer a flexible and cost-effective solution for silicon test and
debug applications. In this solution, the signals under test are shifted out to the TDO pin of the TAP
Controller. The main advantage is that all the test signals are monitored from the TDO pin; no pins or
additional board-level resources are required. Figure 16-6 illustrates this technique. Multiple test nets are
brought into an internal MUX architecture. The selection of the MUX is done using the contents of the
TAP Controller instruction register, where individual instructions (values from 16 to 127) correspond to
different signals under test. The selected test signal can be synchronized with the rising or falling edge of
TCK (optional) and sent out to UTDO to drive the TDO output of JTAG.
For flash devices, TDO (the output) is configured as low slew and the highest drive strength available in
the technology and/or device. Here are some examples:
1. If the device is A3P1000 and VCCI is 3.3 V, TDO will be configured as LVTTL 3.3 V output,
24 mA, low slew.
2. If the device is AGLN020 and VCCI is 1.8 V, TDO will be configured as LVCMOS 1.8 V output,
4 mA, low slew.
3. If the device is AGLE300 and VCCI is 2.5 V, TDO will be configured as LVCMOS 2.5 V output,
24 mA, low slew.
The test and debug procedure is not limited to the example in Figure 16-5 on page 303. Users can
customize the debug and test interface to make it appropriate for their applications. For example, multiple
test signals can be registered and then sent out through UTDO, each at a different edge of TCK. In other
words, n signals are sampled with an FTCK / n sampling rate. The bandwidth of the information sent out
to TDO is always proportional to the frequency of TCK.
Internal Test Nets
Instruction
Decode
UIREG[7:0]
To Scope Channel
TDO
TDI
URSTB
UDRUPD
UDRCK
TMS
TCK
TRST
D
UDRCAP
CLK
UDRSH
UTDI
UTDO
Figure 16-6 • UJTAG Usage Example in Test and Debug Applications
304
R e vi s i o n 5
Q
ProASIC3 nano FPGA Fabric User’s Guide
SRAM Initialization
Users can also initialize embedded SRAMs of the low power flash devices. The initialization of the
embedded SRAM blocks of the design can be done using UJTAG tiles, where the initialization data is
imported using the TAP Controller. Similar functionality is available in ProASICPLUS devices using JTAG.
The guidelines for implementation and design examples are given in the RAM Initialization and ROM
Emulation in ProASICPLUS Devices application note.
SRAMs are volatile by nature; data is lost in the absence of power. Therefore, the initialization process
should be done at each power-up if necessary.
FlashROM Read-Back Using JTAG
The low power flash architecture contains a dedicated nonvolatile FlashROM block, which is formatted
into eight 128-bit pages. For more information on FlashROM, refer to the "FlashROM in Microsemi’s Low
Power Flash Devices" section on page 117. The contents of FlashROM are available to the VersaTiles
during normal operation through a read operation. As a result, the UJTAG macro can be used to provide
the FlashROM contents to the JTAG port during normal operation. Figure 16-7 illustrates a simple block
diagram of using UJTAG to read the contents of FlashROM during normal operation.
The FlashROM read address can be provided from outside the FPGA through the TDI input or can be
generated internally using the core logic. In either case, data serialization logic is required (Figure 16-7)
and should be designed using the VersaTile core logic. FlashROM contents are read asynchronously in
parallel from the flash memory and shifted out in a synchronous serial format to TDO. Shifting the serial
data out of the serialization block should be performed while the TAP is in UDRSH mode. The
coordination between TCK and the data shift procedure can be done using the TAP state machine by
monitoring UDRSH, UDRCAP, and UDRUPD.
UJTAG
Address Generation and
Data Serlialization
UIREG[7:0]
TDO
TDI
TMS
TCK
TRST
Enable
FROM
RESET
URSTB
UDRUPD
Control
UDRCK
CLK
UDRCAP
SDI
UDRSH
Addr[6:0]
Addr [6:0]
Data[7:0]
Data[7:0]
SDO
UTDI
UTDO
Figure 16-7 • Block Diagram of Using UJTAG to Read FlashROM Contents
Revision 5
305
UJTAG Applications in Microsemi’s Low Power Flash Devices
Conclusion
Microsemi low power flash FPGAs offer many unique advantages, such as security, nonvolatility,
reprogrammablity, and low power—all in a single chip. In addition, Fusion, IGLOO, and ProASIC3
devices provide access to the JTAG port from core VersaTiles while the device is in normal operating
mode. A wide range of available user-defined JTAG opcodes allows users to implement various types of
applications, exploiting this feature of these devices. The connection between the JTAG port and core
tiles is implemented through an embedded and hardwired UJTAG tile. A UJTAG tile can be instantiated in
designs using the UJTAG library cell. This document presents multiple examples of UJTAG applications,
such as dynamic reconfiguration, silicon test and debug, fine-tuning of the design, and RAM initialization.
Each of these applications offers many useful advantages.
Related Documents
Application Notes
RAM Initialization and ROM Emulation in ProASICPLUS Devices
http://www.microsemi.com/soc/documents/APA_RAM_Initd_AN.pdf
List of Changes
The following table lists critical changes that were made in each revision of the chapter.
Date
Changes
Page
December 2011
Information on the drive strength and slew rate of TDO pins was added to the
"Silicon Testing and Debugging" section (SAR 31749).
304
July 2010
This chapter is no longer published separately with its own part number and version
but is now part of several FPGA fabric user’s guides.
N/A
v1.4
(December 2008)
IGLOO nano and ProASIC3 nano devices were added to Table 16-1 • Flash-Based
FPGAs.
298
v1.3
(October 2008)
The "UJTAG Support in Flash-Based Devices" section was revised to include new
families and make the information more concise.
298
The title of Table 16-3 • Configuration Bits of Fusion, IGLOO, and ProASIC3 CCC
Blocks was revised to include Fusion.
302
The following changes were made to the family descriptions in Table 16-1 • FlashBased FPGAs:
298
v1.2
(June 2008)
v1.1
(March 2008)
306
•
ProASIC3L was updated to include 1.5 V.
•
The number of PLLs for ProASIC3E was changed from five to six.
The chapter was updated to include the IGLOO PLUS family and information
regarding 15 k gate devices.
N/A
The "IGLOO Terminology" section and "ProASIC3 Terminology" section are new.
298
R e vi s i o n 5
17 – Power-Up/-Down Behavior of Low Power
Flash Devices
Introduction
Microsemi’s low power flash devices are flash-based FPGAs manufactured on a 0.13 µm process node.
These devices offer a single-chip, reprogrammable solution and support Level 0 live at power-up (LAPU)
due to their nonvolatile architecture.
Microsemi's low power flash FPGA families are optimized for logic area, I/O features, and performance.
IGLOO® devices are optimized for power, making them the industry's lowest power programmable
solution. IGLOO PLUS FPGAs offer enhanced I/O features beyond those of the IGLOO ultra-low power
solution for I/O-intensive low power applications. IGLOO nano devices are the industry's lowest-power
cost-effective solution. ProASIC3®L FPGAs balance low power with high performance. The ProASIC3
family is Microsemi's high-performance flash FPGA solution. ProASIC3 nano devices offer the lowestcost solution with enhanced I/O capabilities.
Microsemi’s low power flash devices exhibit very low transient current on each power supply during
power-up. The peak value of the transient current depends on the device size, temperature, voltage
levels, and power-up sequence.
The following devices can have inputs driven in while the device is not powered:
•
IGLOO (AGL015 and AGL030)
•
IGLOO nano (all devices)
•
IGLOO PLUS (AGLP030, AGLP060, AGLP125)
•
IGLOOe (AGLE600, AGLE3000)
•
ProASIC3L (A3PE3000L)
•
ProASIC3 (A3P015, A3P030)
•
ProASIC3 nano (all devices)
•
ProASIC3E (A3PE600, A3PE1500, A3PE3000)
•
Military ProASIC3EL (A3PE600L, A3PE3000L, but not A3P1000)
•
RT ProASIC3 (RT3PE600L, RT3PE3000L)
The driven I/Os do not pull up power planes, and the current draw is limited to very small leakage current,
making them suitable for applications that require cold-sparing. These devices are hot-swappable,
meaning they can be inserted in a live power system.1
1.
For more details on the levels of hot-swap compatibility in Microsemi’s low power flash devices, refer to the "Hot-Swap
Support" section in the I/O Structures chapter of the FPGA fabric user’s guide for the device you are using.
Revision 5
307
Power-Up/-Down Behavior of Low Power Flash Devices
Flash Devices Support Power-Up Behavior
The flash FPGAs listed in Table 17-1 support power-up behavior and the functions described in this
document.
Table 17-1 • Flash-Based FPGAs
Family*
Series
IGLOO
ProASIC3
Description
IGLOO
Ultra-low power 1.2 V to 1.5 V FPGAs with Flash*Freeze technology
IGLOOe
Higher density IGLOO FPGAs with six PLLs and additional I/O standards
IGLOO nano
The industry’s lowest-power, smallest-size solution
IGLOO PLUS
IGLOO FPGAs with enhanced I/O capabilities
ProASIC3
Low power, high-performance 1.5 V FPGAs
ProASIC3E
Higher density ProASIC3 FPGAs with six PLLs and additional I/O standards
ProASIC3 nano
Lowest-cost solution with enhanced I/O capabilities
ProASIC3L
ProASIC3 FPGAs supporting 1.2 V to 1.5 V with Flash*Freeze technology
RT ProASIC3
Radiation-tolerant RT3PE600L and RT3PE3000L
Military ProASIC3/EL
Military temperature A3PE600L, A3P1000, and A3PE3000L
Automotive ProASIC3
ProASIC3 FPGAs qualified for automotive applications
Note: *The device names link to the appropriate datasheet, including product brief, DC and switching characteristics,
and packaging information.
IGLOO Terminology
In documentation, the terms IGLOO series and IGLOO devices refer to all of the IGLOO devices as listed
in Table 17-1. Where the information applies to only one product line or limited devices, these exclusions
will be explicitly stated.
ProASIC3 Terminology
In documentation, the terms ProASIC3 series and ProASIC3 devices refer to all of the ProASIC3 devices
as listed in Table 17-1. Where the information applies to only one product line or limited devices, these
exclusions will be explicitly stated.
To further understand the differences between the IGLOO and ProASIC3 devices, refer to the Industry’s
Lowest Power FPGAs Portfolio.
308
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
Power-Up/-Down Sequence and Transient Current
Microsemi's low power flash devices use the following main voltage pins during normal operation:2
•
VCCPLX
•
VJTAG
•
VCC: Voltage supply to the FPGA core
–
VCC is 1.5 V ± 0.075 V for IGLOO, IGLOO nano, IGLOO PLUS, and ProASIC3 devices
operating at 1.5 V.
–
VCC is 1.2 V ± 0.06 V for IGLOO, IGLOO nano, IGLOO PLUS, and ProASIC3L devices
operating at 1.2 V.
–
V5 devices will require a 1.5 V VCC supply, whereas V2 devices can utilize either a 1.2 V or
1.5 V VCC.
•
VCCIBx: Supply voltage to the bank's I/O output buffers and I/O logic. Bx is the I/O bank number.
•
VMVx: Quiet supply voltage to the input buffers of each I/O bank. x is the bank number. (Note:
IGLOO nano, IGLOO PLUS, and ProASIC3 nano devices do not have VMVx supply pins.)
The I/O bank VMV pin must be tied to the VCCI pin within the same bank. Therefore, the supplies that
need to be powered up/down during normal operation are VCC and VCCI. These power supplies can be
powered up/down in any sequence during normal operation of IGLOO, IGLOO nano, IGLOO PLUS,
ProASIC3L, ProASIC3, and ProASIC3 nano FPGAs. During power-up, I/Os in each bank will remain
tristated until the last supply (either VCCIBx or VCC) reaches its functional activation voltage. Similarly,
during power-down, I/Os of each bank are tristated once the first supply reaches its brownout
deactivation voltage.
Although Microsemi's low power flash devices have no power-up or power-down sequencing
requirements, Microsemi identifies the following power conditions that will result in higher than normal
transient current. Use this information to help maximize power savings:
Microsemi recommends tying VCCPLX to VCC and using proper filtering circuits to decouple VCC noise
from the PLL.
a. If VCCPLX is powered up before VCC, a static current of up to 5 mA (typical) per PLL may be
measured on VCCPLX.
The current vanishes as soon as VCC reaches the VCCPLX voltage level.
The same current is observed at power-down (VCC before VCCPLX).
b. If VCCPLX is powered up simultaneously or after VCC:
i.
Microsemi's low power flash devices exhibit very low transient current on VCC. For
ProASIC3 devices, the maximum transient current on VCC does not exceed the maximum
standby current specified in the device datasheet.
The source of transient current, also known as inrush current, varies depending on the FPGA technology.
Due to their volatile technology, the internal registers in SRAM FPGAs must be initialized before
configuration can start. This initialization is the source of significant inrush current in SRAM FPGAs
during power-up. Due to the nonvolatile nature of flash technology, low power flash devices do not
require any initialization at power-up, and there is very little or no crossbar current through PMOS and
NMOS devices. Therefore, the transient current at power-up is significantly less than for SRAM FPGAs.
Figure 17-1 on page 310 illustrates the types of power consumption by SRAM FPGAs compared to
Microsemi's antifuse and flash FPGAs.
2.
For more information on Microsemi FPGA voltage supplies, refer to the appropriate datasheet located at
http://www.microsemi.com/soc/techdocs/ds.
Revision 5
309
Power-Up/-Down Behavior of Low Power Flash Devices
Current
Power-On Inrush
SRAM FPGAs
SRAM
Microsemi
FPGAs
Active
Frequency
Dependent
System
Supply
Voltage
Configuration
SRAM FPGAs
Static
Time (or frequency)
Figure 17-1 • Types of Power Consumption in SRAM FPGAs and Microsemi Nonvolatile FPGAs
Transient Current on VCC
The characterization of the transient current on VCC is performed on nearly all devices within the
IGLOO, ProASIC3L, and ProASIC3 families. A sample size of five units is used from each device family
member. All the device I/Os are internally pulled down while the transient current measurements are
performed. For ProASIC3 devices, the measurements at typical conditions show that the maximum
transient current on VCC, when the power supply is powered at ramp-rates ranging from 15 V/ms to
0.15 V/ms, does not exceed the maximum standby current specified in the device datasheets. Refer to
the DC and Switching Characteristics chapters of the ProASIC3 Flash Family FPGAS datasheet and
ProASIC3E Flash Family FPGAs datasheet for more information.
Similarly, IGLOO, IGLOO nano, IGLOO PLUS, and ProASIC3L devices exhibit very low transient current
on VCC. The transient current does not exceed the typical operating current of the device while in active
mode. For example, the characterization of AGL600-FG256 V2 and V5 devices has shown that the
transient current on VCC is typically in the range of 1–5 mA.
Transient Current on VCCI
The characterization of the transient current on VCCI is performed on devices within the IGLOO, IGLOO
nano, IGLOO PLUS, ProASIC3, ProASIC3 nano, and ProASIC3L groups of devices, similarly to VCC
transient current measurements. For ProASIC3 devices, the measurements at typical conditions show
that the maximum transient current on VCCI, when the power supply is powered at ramp-rates ranging
from 33 V/ms to 0.33 V/ms, does not exceed the maximum standby current specified in the device
datasheet. Refer to the DC and Switching Characteristics chapters of the ProASIC3 Flash Family
FPGAS datasheet and ProASIC3E Flash Family FPGAs datasheet for more information.
Similarly, IGLOO, IGLOO PLUS, and ProASIC3L devices exhibit very low transient current on VCCI. The
transient current does not exceed the typical operating current of the device while in active mode. For
example, the characterization of AGL600-FG256 V2 and V5 devices has shown that the transient current
on VCCI is typically in the range of 1–2 mA.
310
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
I/O Behavior at Power-Up/-Down
This section discusses the behavior of device I/Os, used and unused, during power-up/-down of VCC and
VCCI. As mentioned earlier, VMVx and VCCIBx are tied together, and therefore, inputs and outputs are
powered up/down at the same time.
I/O State during Power-Up/-Down
This section discusses the characteristics of I/O behavior during device power-up and power-down.
Before the start of power-up, all I/Os are in tristate mode. The I/Os will remain tristated during power-up
until the last voltage supply (VCC or VCCI) is powered to its functional level (power supply functional
levels are discussed in the "Power-Up to Functional Time" section on page 312). After the last supply
reaches the functional level, the outputs will exit the tristate mode and drive the logic at the input of the
output buffer. Similarly, the input buffers will pass the external logic into the FPGA fabric once the last
supply reaches the functional level. The behavior of user I/Os is independent of the VCC and VCCI
sequence or the state of other voltage supplies of the FPGA (VPUMP and VJTAG). Figure 17-2 shows
the output buffer driving HIGH and its behavior during power-up with 10 kΩ external pull-down. In
Figure 17-2, VCC is powered first, and VCCI is powered 5 ms after VCC. Figure 17-3 on page 312
shows the state of the I/O when VCCI is powered about 5 ms before VCC. In the circuitry shown in
Figure 17-3 on page 312, the output is externally pulled down.
During power-down, device I/Os become tristated once the first power supply (VCC or VCCI) drops
below its brownout voltage level. The I/O behavior during power-down is also independent of voltage
supply sequencing.
Figure 17-2 • I/O State when VCC Is Powered before VCCI
Revision 5
311
Power-Up/-Down Behavior of Low Power Flash Devices
Figure 17-3 • I/O State when VCCI Is Powered before VCC
Power-Up to Functional Time
At power-up, device I/Os exit the tristate mode and become functional once the last voltage supply in the
power-up sequence (VCCI or VCC) reaches its functional activation level. The power-up–to–functional
time is the time it takes for the last supply to power up from zero to its functional level. Note that the
functional level of the power supply during power-up may vary slightly within the specification at different
ramp-rates. Refer to Table 17-2 for the functional level of the voltage supplies at power-up.
Typical I/O behavior during power-up–to–functional time is illustrated in Figure 17-2 on page 311 and
Figure 17-3.
Table 17-2 • Power-Up Functional Activation Levels for VCC and VCCI
VCC Functional
Activation Level (V)
VCCI Functional
Activation Level (V)
ProASIC3, ProASIC3 nano, IGLOO, IGLOO nano,
IGLOO PLUS, and ProASIC3L devices running at
VCC = 1.5 V*
0.85 V ± 0.25 V
0.9 V ± 0.3 V
IGLOO, IGLOO nano, IGLOO PLUS, and
ProASIC3L devices running at VCC = 1.2 V*
0.85 V ± 0.2 V
0.9 V ± 0.15 V
Device
Note: *V5 devices will require a 1.5 V VCC supply, whereas V2 devices can utilize either a 1.2 V or 1.5 V
VCC.
Microsemi’s low power flash devices meet Level 0 LAPU; that is, they can be functional prior to VCC
reaching the regulated voltage required. This important advantage distinguishes low power flash devices
from their SRAM-based counterparts. SRAM-based FPGAs, due to their volatile technology, require
hundreds of milliseconds after power-up to configure the design bitstream before they become
functional. Refer to Figure 17-4 on page 313 and Figure 17-5 on page 314 for more information.
312
R e vi s i o n 5
ProASIC3 nano FPGA Fabric User’s Guide
VCC = VCCI + VT
Where VT can be from 0.58 V to 0.9 V (typically 0.75 V)
VCC
VCC = 1.575 V
Region 4: I/O
buffers are ON.
I/Os are functional
(except differential inputs)
but slower because VCCI is
Region 1: I/O Buffers are OFF
below specifcation. For the
same reason, input buffers do not
Region 5: I/O buffers are ON
and power supplies are within
specification.
I/Os meet the entire datasheet
and timer specifications for
speed, VIH/VIL , VOH /VOL , etc.
meet VIH/VIL levels, and output
buffers do not meet VOH/VOL levels.
VCC = 1.425 V
Region 2: I/O buffers are ON.
I/Os are functional (except differential inputs)
but slower because VCCI / VCC are below
specification. For the same reason, input
buffers do not meet VIH / VIL levels, and
output buffers do not meet VOH / VOL levels.
Activation trip point:
Va = 0.85 V ± 0.25 V
Deactivation trip point:
Vd = 0.75 V ± 0.25 V
Region 3: I/O buffers are ON.
I/Os are functional; I/O DC
specifications are met,
but I/Os are slower because
the VCC is below specification
Region 1: I/O buffers are OFF
Activation trip point:
Va = 0.9 V ± 0.3 V
Deactivation trip point:
Vd = 0.8 V ± 0.3 V
Min VCCI datasheet specification
voltage at a selected I/O
standard; i.e., 1.425 V or 1.7 V
or 2.3 V or 3.0 V
VCCI
Figure 17-4 • I/O State as a Function of VCCI and VCC Voltage Levels for IGLOO V5, IGLOO nano V5,
IGLOO PLUS V5, ProASIC3L, and ProASIC3 Devices Running at VCC = 1.5 V ± 0.075 V
Revision 5
313
Power-Up/-Down Behavior of Low Power Flash Devices
VCC = VCCI + VT
where VT can be from 0.58 V to 0.9 V (typically 0.75 V)
VCC
VCC = 1.575 V
Region 4: I/O
buffers are ON.
I/Os are functional
(except differential inputs)
but slower because VCCI is
below specification. For the
same reason, input buffers do not
meet VIH / VIL levels, and output
buffers do not meet VOH / VOL levels.
Region 1: I/O Buffers are OFF
Region 5: I/O buffers are ON
and power supplies are within
specification.
I/Os meet the entire datasheet
and timer specifications for
speed, VIH / VIL , VOH / VOL , etc.
VCC = 1.14 V
Region 2: I/O buffers are ON.
I/Os are functional (except differential inputs)
but slower because VCCI/VCC are below
specification. For the same reason, input
buffers do not meet VIH/VIL levels, and
output buffers do not meet VOH/VOL levels.
Activation trip point:
Va = 0.85 V ± 0.2 V
Deactivation trip point:
Vd = 0.75 V ± 0.2 V
Region 3: I/O buffers are ON.
I/Os are functional; I/O DC
specifications are met,
but I/Os are slower because
the VCC is below specification.
Region 1: I/O buffers are OFF
Activation trip point:
Va = 0.9 V ± 0.15 V
Deactivation trip point:
Vd = 0.8 V ± 0.15 V
Min VCCI datasheet specification
voltage at a selected I/O
standard; i.e., 1.14 V,1.425 V, 1.7 V,
2.3 V, or 3.0 V
Figure 17-5 • I/O State as a Function of VCCI and VCC Voltage Levels for IGLOO V2, IGLOO nano V2,
IGLOO PLUS V2, and ProASIC3L Devices Running at VCC = 1.2 V ± 0.06 V
314
R e vi s i o n 5
VCCI
ProASIC3 nano FPGA Fabric User’s Guide
Brownout Voltage
Brownout is a condition in which the voltage supplies are lower than normal, causing the device to
malfunction as a result of insufficient power. In general, Microsemi does not guarantee the functionality of
the design inside the flash FPGA if voltage supplies are below their minimum recommended operating
condition. Microsemi has performed measurements to characterize the brownout levels of FPGA power
supplies. Refer to Table 17-3 for device-specific brownout deactivation levels. For the purpose of
characterization, a direct path from the device input to output is monitored while voltage supplies are
lowered gradually. The brownout point is defined as the voltage level at which the output stops following
the input. Characterization tests performed on several IGLOO, ProASIC3L, and ProASIC3 devices in
typical operating conditions showed the brownout voltage levels to be within the specification.
During device power-down, the device I/Os become tristated once the first supply in the power-down
sequence drops below its brownout deactivation voltage.
Table 17-3 • Brownout Deactivation Levels for VCC and VCCI
VCC Brownout
VCCI Brownout
Deactivation Level (V) Deactivation Level (V)
Devices
ProASIC3, ProASIC3 nano, IGLOO, IGLOO nano,
IGLOO PLUS and ProASIC3L devices running at
VCC = 1.5 V
0.75 V ± 0.25 V
0.8 V ± 0.3 V
IGLOO, IGLOO nano, IGLOO PLUS, and
ProASIC3L devices running at VCC = 1.2 V
0.75 V ± 0.2 V
0.8 V ± 0.15 V
PLL Behavior at Brownout Condition
When PLL power supply voltage and/or VCC levels drop below the VCC brownout levels mentioned
above for 1.5 V and 1.2 V devices, the PLL output lock signal goes LOW and/or the output clock is lost.
The following sections explain PLL behavior during and after the brownout condition.
VCCPLL and VCC Tied Together
In this condition, both VCC and VCCPLL drop below the 0.75 V (± 0.25 V or ± 0.2 V) brownout level.
During the brownout recovery, once VCCPLL and VCC reach the activation point (0.85 ± 0.25 V or
± 0.2 V) again, the PLL output lock signal may still remain LOW with the PLL output clock signal toggling.
If this condition occurs, there are two ways to recover the PLL output lock signal:
1. Cycle the power supplies of the PLL (power off and on) by using the PLL POWERDOWN signal.
2. Turn off the input reference clock to the PLL and then turn it back on.
Only VCCPLL Is at Brownout
In this case, only VCCPLL drops below the 0.75 V (± 0.25 V or ± 0.2 V) brownout level and the VCC
supply remains at nominal recommended operating voltage (1.5 V ± 0.075 V for 1.5 V devices and 1.2 V
± 0.06 V for 1.2 V devices). In this condition, the PLL behavior after brownout recovery is similar to initial
power-up condition, and the PLL will regain lock automatically after VCCPLL is ramped up above the
activation level (0.85 ± 0.25 V or ± 0.2 V). No intervention is necessary in this case.
Only VCC Is at Brownout
In this condition, VCC drops below the 0.75 V (± 0.25 V or ± 0.2 V) brownout level and VCCPLL remains
at nominal recommended operating voltage (1.5 V ± 0.075 V for 1.5 V devices and 1.2 V ± 0.06 V for
1.2 V devices). During the brownout recovery, once VCC reaches the activation point again (0.85 ±
0.25 V or ± 0.2 V), the PLL output lock signal may still remain LOW with the PLL output clock signal
toggling. If this condition occurs, there are two ways to recover the PLL output lock signal:
1. Cycle the power supplies of the PLL (power off and on) by using the PLL POWERDOWN signal.
2. Turn off the input reference clock to the PLL and then turn it back on.
It is important to note that Microsemi recommends using a monotonic power supply or voltage regulator
to ensure proper power-up behavior.
Revision 5
315
Power-Up/-Down Behavior of Low Power Flash Devices
Internal Pull-Up and Pull-Down
Low power flash device I/Os are equipped with internal weak pull-up/-down resistors that can be used by
designers. If used, these internal pull-up/-down resistors will be activated during power-up, once both
VCC and VCCI are above their functional activation level. Similarly, during power-down, these internal
pull-up/-down resistors will turn off once the first supply voltage falls below its brownout deactivation
level.
Cold-Sparing
In cold-sparing applications, voltage can be applied to device I/Os before and during power-up. Coldsparing applications rely on three important characteristics of the device:
1. I/Os must be tristated before and during power-up.
2. Voltage applied to the I/Os must not power up any part of the device.
3. VCCI should not exceed 3.6 V, per datasheet specifications.
As described in the "Power-Up to Functional Time" section on page 312, Microsemi’s low power flash
I/Os are tristated before and during power-up until the last voltage supply (VCC or VCCI) is powered up
past its functional level. Furthermore, applying voltage to the FPGA I/Os does not pull up VCC or VCCI
and, therefore, does not partially power up the device. Table 17-4 includes the cold-sparing test results
on A3PE600-PQ208 devices. In this test, leakage current on the device I/O and residual voltage on the
power supply rails were measured while voltage was applied to the I/O before power-up.
Table 17-4 • Cold-Sparing Test Results for A3PE600 Devices
Residual Voltage (V)
Device I/O
VCC
VCCI
Leakage Current
Input
0
0.003
工商网监
湘ICP备2023018690号
