1 / 14

Briefing: Independent NASA Test of RTSX-SU FPGAs Summary of Industry Tiger Team Results

Briefing: Independent NASA Test of RTSX-SU FPGAs Summary of Industry Tiger Team Results. Aerospace Actel Team Team Lead : Larry Harzstark 2-16-05. Problem Statement.

Download Presentation

Briefing: Independent NASA Test of RTSX-SU FPGAs Summary of Industry Tiger Team Results

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Briefing: Independent NASA Test of RTSX-SU FPGAsSummary of Industry Tiger Team Results Aerospace Actel Team Team Lead: Larry Harzstark 2-16-05

  2. Problem Statement • Several Contractors Reported Failures of a New Generation (RT54SX-S/A54SX-A) of Actel FPGAs Subsequent to Successful Programming • These Devices Are Ubiquitous and Are Used by Nearly Every Program (SMC, NRO, NASA, …) With Recently Built Hardware • Rapid Identification of Root-cause and Remediation Needed to Keep Programs on Schedule

  3. Planned Activity • Collect Data on FPGA Failure Conditions and Successful Operating Conditions • Determine root cause of failures • Perform Destructive Physical Analysis (DPA) of Normal and Failed Devices • Gain Physical Understanding of Failure Mechanism • Establish a Better Reliability Database and Determine Effect of Electrical Stress on Reliability • Screen Out Infant Mortality • Understand Long Term Reliability • Develop a potential screen (if possible) • Assist Programs in Assessment of Fly-as-is Risks

  4. Phase 1: MEC Old Algorithm • Original Hypothesis • Failures Caused by Users Operating Parts Outside of Actel Specification • Small Sub-population of Antifuses Fail Quickly Under Electrical Overstress • All Contractors Were Operating Devices With Out-of-spec Transients • All Reported Failures Occurred in First 200 Hours • Aerospace-led Tiger Team Developed a Test Plan • Perceptive Test Vehicle for Antifuse Degradation • Measure Failure Rate With Increasing Electrical Overstress • NRO Program Sought Early Results and Sponsored Subset of Tests - Testing Began 27 May • All testing performed by Actel • Aerospace Developed New DPA Techniques to Understand Failure Mechanism

  5. Colonel’s MEC Test Plan Old-Algorithm, Started 27 May SSU-Stressed Stress-free 4B1 @ 25 MHz 17 I/Os switching 12.5% I/O toggle rate -1V undershoot 4B2 @ 50 MHz 70 I/O’s switching; 50% toggle rate -2V undershoot Project 4B1 550 parts Project 7 600 hrs Project 4B2 2000 hrs Project 7 – 30 failures observed in 600 hours Project 4B1 – 5 failures observed in 2000 hours Project 4B2 – 3 failures observed in 2000 hours Results proved original hypothesis of user overstress as the cause of failures not valid

  6. Colonel’s Project 4 Failures • Failures consistent with continuation of Project 7 Weibull curve • No evidence that stress (SSU) caused any additional failures * After 600 Hours in Project 7

  7. Phase 2: New Algorithm • Actel Released Modified Programming Algorithm on 19 May to Eliminate the Early Infant Mortality Failures • Actel Believed Old Algorithm allowed too much power to be delivered to antifuse during the initial breakdown phase of programming • Actel Had Performed Experiments Using Their Standard Qualification Test to Compare Old and New Algorithms using A54SX72A (2X dynamic antifuse over industry RT54SX32S testing) • Old Algorithm: 16 Failures in 623 Devices in 168 Hours • New Algorithm: 0 Failures in 705 Devices in 1000 Hrs • Actel Subsequently Used More Perceptive Project 7 Test (A54SX72A) • 5 Failures in 231 New-algorithm Devices in 124 Hours • All Low-current Type Antifuses • Tiger-team Tests Observed Failures in Both High-current and Low-current Antifuses • DPAs Showed Same Characteristics As Old-algorithm Failures • Lower Failure Rate Than Old-algorithm, but Still Unacceptable MEC New algorithm not recommended by Aerospace due to high failure rates

  8. Tiger-Team MEC Parts Testing Started 31 Aug • 4B2 Old Algo • 25º C • 4B2 Old Algo • 85º C Vcca=3.0 500 parts • 4B2 New Algo • 25º C Electrical test points at 0, 1, 25, 49,168, 500,1000,1500, … hours

  9. Tiger Team Project 4 Test Failures New algorithm did not eliminate failures

  10. Statistical Analysis **Note: estimates for the old algorithm are based on the Tiger Team Project 4 tests (which has a much smaller sample size than the Colonel’s test) and could be as large as 8.8% according to the Colonel’s data

  11. Phase 3: UMC Foundry • Actel Announced UMC Version of Radiation Tolerant FPGAs on 19 Jul • Several Years of Successful UMC Experience With Commercial FPGA Designs • Millions of Parts Shipped • Received MIL-Q Certification From DSCC on 1 Sep • Over 1.3M Device-hours Without Antifuse Failure Using Actel Qualification Test Vehicle • Longest Test Time Was 2000 Hours • Actel Has Tested With No Failures Using Tiger-team P7 and P4 Test Vehicles (see latest Actel data) • Over 250K Device-hours on 666 Devices • Longest Test Time Is 1268 Hours on 100 Devices UMC test results very promising

  12. The Way Ahead • Most SMC and NRO Programs Have Switched to UMC FPGAs or ASICs • A few experimental programs still evaluating options • Three Test Programs for UMC FPGAs in Process • NASA Tests – Rich Katz to discuss • Aerospace Life test on 240 Commercial UMC Parts • Space Qualification Program

  13. End Aerospace

More Related