Post on 28-Dec-2015
NASA EEE Parts Challenges for Spaceflight Electronics
National Aeronautics and Space Administration
www.nasa.gov
Quality Leadership Forum
March 21, 2013
Michael J. Sampson
NASA GSFC
Safety and Mission Assurance Directorate (Code 300)
michael.j.sampson@nasa.gov
301-614-6233
SMILE – It’s NOT that BAD
2MJS QLF 3/21/2013
Commercial Off-The-Shelf (COTS), Electronic Parts• The obvious driver is COST• NASA uses many COTS parts and always has• MIL/Hi Rel parts can have a selling price 1000 X COTS• COTS parts can also be more powerful, faster, more
volumetrically efficient• The challenge is to know enough about the COTS parts to
reach an acceptable risk – Total Cost of Ownership• The position stated here is not new, see NEXT SLIDE• What may have changed are options for how and when to
develop the knowledge to manage the risk• Changes to NASA’s ways of business are imperative for
greater use of COTS and cost savings from COTS
Only One Challenge Matters - COTS
3MJS QLF 3/21/2013
INHERENT RISK INVERSELY PROPORTIONAL TO KNOWLEDGE
• If a part is KNOWN to be high risk, this knowledge can be used to avoid its use or take appropriate actions to move to medium or low risk
• Lack of knowledge means good parts cannot be distinguished from bad
• Obtaining reliable knowledge about COTS EEE Parts requires:– Expertise
– Time
– Vendor visits
– Testing and Analysis
– BIG BUCKS
ONLY a LIMITED number of COTS part types can be reliably deployed in any one system
EEE Parts Risk Assessment - Risk versus Knowledge
From QLF July 2001
4MJS QLF 3/21/2013
Uncertainty Cost
Risk
A Notional View of the Key Tradeoffs
5MJS QLF 3/21/2013
Uncertainty Cost
Risk
A Notional View of the Cost Tradeoffs
Class S
NASA Level 3
COTS
Class B
6MJS QLF 3/21/2013
Inherent Risks - for EEE Parts
• Manufacturing Factors– Spec– Vendor– Maturity/Qualification Status– Knowledge of Changes– Radiation Sensitivity– Traceability (added 3/1/2013)
• These are risks inherent to the part regardless of:– Redundancy– Derating– Mission Requirements– Mission Budget
From QLF July 2001
7MJS QLF 3/21/2013
Traceability?From QLF
March 2003
8MJS QLF 3/21/2013
S
• Manufacturer hi-rel catalog part based on MIL-PRF-38534 Class Level S• GSFC project suffered hybrid failures traced to LEDs used to trigger a
photodetector, hybrid failure rate ~ 2%, LEDs ~ 1.5%• Very unusual failure involving total bond pad lift• Poor Traceability Records mean lack of ability to identify the specific
wafer lot, impeding ability to determine root cause
Traceability is Important
9MJS QLF 3/21/2013
DYNAMIC BURN-IN RESULTS
Dynamic Burn-In Per the Application is Recommended
• Value added step when done in conjunction with a data review for part performance and reliability. More effective than static burn-in for many part types.
Part Type ss Vendor Hours BI Temp Rejs(25C) Functional Parametric Critical Parameters
A/D 254 A 440 +85C 1 0 1 ICCD
Multiplexer 250 B 168 +125C 7 0 7 Ron, I+VEN,IAL,IAH
Op Amp 253 C 400 +105C 1 0 1 VOS
Reference 252 D 168 +125C TBD TBD TBD TBD
Amplifier 230 E 168 +125C 1 1 0 Gain ERR,VOO
NEPP/NEPAG FY03/04 – COTSPlastic Encapsulated Microcircuits Evaluation
Example of COTS PEMS Evaluation Findings
•Dynamic Burn-In Screening Captures Failures in 4 out of 5 Lots Tested
10MJS QLF 3/21/2013
Goals & ObjectivesLOT VARIATIONS EXAMPLE
Demonstrated: COTS Products May Demonstrate Distinctly Different Lot to Lot Parametric Variations Post Burn-In
Recommendation: Evaluate ALL Lot Date Codes Procured to Determine Flight Worthiness and Application-Specific Acceptability
LDC 0112
LDC 0122LDC 0127
Three Lot Date Codes of One Part Type From A Single Source
NEPP/NEPAG FY03/04 – COTSPlastic Encapsulated Microcircuits Evaluation
11MJS QLF 3/21/2013
Impact of Radiation Assurance With Upgrading On Parts Costs (incl. parts cost)
Cos
t in
$1,
000,
000’
s
12MJS QLF 3/21/2013
• Learn a lot about parts before installation– Works well for one-off, long duration, low risk missions– Minimizes post installation costs in correcting parts
problems• Learn a lot about parts after installation
– Works well for mass production, short duration missions – risk?
– Limited variation, standardization of key common functions: attitude control, communication, power, etc.
• Perhaps there is a middle ground for NASA?– Basic qualification at the part level– Limited part types used for standard functional modules– Screening for compliance to key parameters– Extensive testing at the module level– Automotive parts?
So What About Options?
13MJS QLF 3/21/2013
Time at which defect is caught
The later a defect is caught, the more:
• Layers have to be removed
• Work has to be “undone”
• Testing has to be redone AND
• Likely the project will fly with residual risk.
Co
st to
re
mo
ve a
sin
gle
def
ect
Launch
Date
Mission
Cost+ Cost of a Single Defect
14MJS QLF 3/21/2013
Microcircuit Manufacturer Statement (3/3/2013)
Restructuring:“… reduce our staffing by 18%” “… our Space Products (High Rel business), … was NOT affected by the announced action. In fact, X has added R&D dollars to the high rel group to support our space customers with new products. Rest assured, High Rel/Space products continue to be a core competency that X will continue to develop and grow.
Part availability issues are overstated
15MJS QLF 3/21/2013
16MJS QLF 3/21/2013
Xilinx Package Change
Future Package
Current Package – Now Obsolete
17MJS QLF 3/21/2013
A 1752 I/O Column Grid Array
From: Reliability of CGA/LGA/HDI Package Board/Assemblyby
Reza Ghaffarian Ph.D, JPLAvailable at http://nepp.nasa.gov
18MJS QLF 3/21/2013
Amusing Metal Whiskers
MJS QLF 3/21/2013 19
Image Courtesy of Peter Bush, SUNY, Buffalo
Image Courtesy of Lyudmyla Panashchenko NASA GSFC
MJS QLF 3/21/2013 20
Image Courtesy of Peter Bush, SUNY, Buffalo
QUESTIONS
21
http://nepp.nasa.gov