Microsemi Corporation

Microsemi — Space Forum 2017

24 slides from Microsemi's conference presentation, 2017

  1. Slide 1: RTG4 Radiation Update, Space Forum 2017

    Power Matters.TM RTG4 Radiation Update Space Forum 2017  J.J. Wang, Chief Engineer  Nadia Rezzak, Staff Engineer  Stephen Varela, Engineer © 2017 Microsemi Corporation. Company Proprietary 1

  2. Slide 2: Agenda

    Agenda  Introduction  Chip Level TID and SEE Update  Single Event Effects Update • Fabric Circuit Heavy-Ion Testing • PLL and SERDES Heavy-Ion Testing • Reprogramming in Space Flight  Summary and Further Radiation Testing © 2017 Microsemi Corporation. Company Proprietary Power Matters.TM 2

  3. Slide 3: RTG4 FPGAs

    RTG4 FPGAs  Radiation tolerant Flash-based FPGA manufactured by UMC 65nm technology  High-Speed Signal Processing • 300 MHz • 150K LE (STMRFF) • 5 Mbit SRAM (EDAC) • 462 Multipliers (DSP) • RT-PLL • 24 x 3.125 Gbps SERDES  Hardened for both TID and SEE • TID > 100 Krad • SEL immune • SEU/SET/SEFI

  4. Slide 4: Flash-Cell Radiation Hardening

    Flash-Cell Radiation Hardening © 2017 Microsemi Corporation. Company Proprietary Power Matters.TM 4

  5. Slide 5: TID and SEE Hardened C-Flash Cell

    TID and SEE Hardened C-Flash Cell 10 DATA Propagation Delay Change (%) Commercial 9 N-Flash FPGA 8 Switch 7 VDD(2.5) 6 5 RTG4 P-Flash 4 Radiation Tolerant 3 C-Flash FPGA VDD 2 1 PMOS 0 -1 VSS 0 25 50 75 100 125 Total Dose (krad)  NOR Flash architecture VDD  Charge storage for N-Flash or P-Flash is

  6. Slide 6: Fabric Circuits Radiation Update

    Fabric Circuits Radiation Update © 2017 Microsemi Corporation. Company Proprietary Power Matters.TM 6

  7. Slide 7: Fabric FF (STMRFF) SET Filter Enable―2X Error Reduction

    Fabric FF (STMRFF) SET Filter Enable―2X Error Reduction SET Filter Off 600ps Filter On  No error observed at 1 MHz  Flip-flop TMR works to eliminate SEU at 1 MHz  Errors observed at 50 MHz, 100 MHz, and 200 MHz were all SET and not SEU  Filter reduced SET by half Device Family Error Rate for GEO

  8. Slide 8: LBNL―RTG4 Mathblock SET Filter Testing

    LBNL―RTG4 Mathblock SET Filter Testing  Mathblock design with SET Filter enabled vs disabled are tested • Parallel Math Block Chains (25 stages) – Configured using cascade mode uses dedicated math block routing—Not going through the fabric – Frequencies : 50 MHz and 100 MHz – SET filter enabled (60

  9. Slide 9: Math Blocks SET Filter Enable―10X Error Reduction

    Math Blocks SET Filter Enable―10X Error Reduction 100 MHz Cross Section (cm2) 1.00E-07 100MHz SET Disabled 1.00E-08 100Mhz SET Enabled 1.00E-09 0 5 10 15 20 25 30 35 40 LET (MeV-cm2/mg) 50 MHz Cross Section (cm2) 1.00E-07 50MHz SET 1.00E-08 Disabled 50 MHz SET Enabled 1.00E-09 0 5 10 15 20 25 30 35

  10. Slide 10: Global Clock Buffer

    Global Clock Buffer  No SET observed in Global Buffer or Row Global Buffer for LET < 30 MeV-cm2/mg Notes: Data points indicate testing limits. © 2017 Microsemi Corporation. Company Proprietary Power Matters.TM 10

  11. Slide 11: PLL and SERDES Radiation Update

    PLL and SERDES Radiation Update © 2017 Microsemi Corporation. Company Proprietary Power Matters.TM 11

  12. Slide 12: PLL

    PLL  PLL Single Event Functional Interrupt (SEFI) is defined by PLL loss of lock  Heavy-Ion SEE Testing Results • PLL lost lock and self-recovers at LET < 65 MeV-cm2/mg • PLL lost lock and can be recovered by reset at LET = 65 MeV-cm2/mg • Lock loss < 100 us  More testing is planned Cross Section

  13. Slide 13: SERDES SEE Test and SEFI Auto-Recover

    SERDES SEE Test and SEFI Auto-Recover Hardened 1. Controller in external master chip manages the testing system. 2. Configuration register in SERDES is SEE hardened but has upset. 3. When SERDES SEFI occurs, controller detects the event based on looped back data. 4. A command is sent to on-chip Core

  14. Slide 14: SERDES SEFI Results

    SERDES SEFI Results  SERDES SEFI is defined by loss of link • Several signals were monitored • Loss of link is when returned data is invalid for 2+ consecutive cycles © 2017 Microsemi Corporation. Company Proprietary Power Matters.TM 14

  15. Slide 15: SERDES Summary

    SERDES Summary  No DEVRST_N was required to recover link loss  Most SEFI were self-recoverable • Others required reinitializing SERDES configuration registers using CoreABC  Link loss was recoverable with duration in sub-millisecond range  Plan for next testing: • Improve SERDES test design and

  16. Slide 16: In-Flight Reprogramming―In-Beam Reprogramming Test

    In-Flight Reprogramming―In-Beam Reprogramming Test © 2017 Microsemi Corporation. Company Proprietary Power Matters.TM 16

  17. Slide 17: In-Beam RTG4 Reprogramming

    In-Beam RTG4 Reprogramming  FlashPro used by customers Off-Beam Reprogram Pass Fluence until Prog Fail Reprogram Functional Reprogram Attempts Reprogram Passed  Reprogramming in beam often gets interrupted Fluence (Ion/cm2) Effective Flux Effective LET (Ion/cm2/s)  No damage at LET ≤ 30.5 Run  R

  18. Slide 18: In-Beam Reprogramming Error

    In-Beam Reprogramming Error  Example: Run 24, Bit stream error programmer '87254' : Scan Chain... programmer '87254' : Scan Chain PASSED. programmer '87254' : Executing action PROGRAM programmer '87254' : EXPORT ISC_ENABLE_RESULT[32] = 00004404 programmer '87254' : EXPORT CRCERR: [1] = 0 programmer

  19. Slide 19: In-Beam Reprogramming SEFI―Register Upset

    In-Beam Reprogramming SEFI―Register Upset Run 25, Nominal, Tilt: 0 LET: 1.23, Fluence: 6.00E+05, Temperature: Room Run 27, Nominal, Tilt: 0 LET: 8.17, Fluence: 1.41E+06, Temperature: Room 0.38 1.4 0.385 1.4 0.36 1.2 0.365 1.2 0.34 1 0.345 1 Current (A) Voltage (V) Current (A) Voltage (V) 0.32 0.3 0.

  20. Slide 20: In-Beam Reprogramming Soft SEFI―Cross Section and Error Rate

    In-Beam Reprogramming Soft SEFI―Cross Section and Error Rate 1.0E-04 Cross Section (cm2) 1.0E-05 1.0E-06 GEO MIN Rate = 2.8  10-3 event/device/day 1.0E-07 0 5 10 15 20 25 30 35 LET (MeV-cm2/mg) © 2017 Microsemi Corporation. Company Proprietary Power Matters.TM 20

  21. Slide 21: In-Flight Reprogramming Guidance

    In-Flight Reprogramming Guidance  Preliminary guidance • Highly unlikely that a destructive event will occur during programming in space • Probability of success for programming in GEO is estimated ~ 99% or higher • It is highly likely that in space, no ion will disrupt programming • If an ion stri

  22. Slide 22: Prompt-Dose/Dose-Rate Testing

    Prompt-Dose/Dose-Rate Testing Contact Microsemi for more information  Ken O’Neill Director of Marketing, Space, and Aviation 408-643-6179 ken.oneill@microsemi.com  Minh Nguyen Senior Marketing Manager, Space 408-643-6283 minh.u.nguyen@microsemi.com © 2017 Microsemi Corporation. Company Proprietary

  23. Slide 23: RTG4 Radiation Summary

    RTG4 Radiation Summary Total Ionizing Dose Stays within parametric limits > 125 Krad (Si) Single Event Latch-Up No failure at facility limit of 103 MeV-cm2/mg, 100°C Configuration Upset No failure at facility limit of 103 MeV-cm2/mg, 100°C Flip-Flop SEU 2.6E-12 errors/bit-day, GEO solar minimum, 1MH

  24. Slide 24: Thank You

    Thank You Microsemi Corporation (Nasdaq: MSCC) offers a comprehensive portfolio of semiconductor and system solutions for aerospace & defense, communications, data center and industrial markets. Products include high-performance and radiation-hardened analog mixed-signal integrated circuits, FPGAs,

Microsemi — Space Forum 2017 on 6ix