COTS PEMs Procurement

1
COTS PEMs Procurement Acquisition Concerns and
Issues

• Dr. Henning W. Leidecker
• GSFC/Code 562
• 301-286-918021 Nov 2002

2
See Also

• GSFC Plastic Encapsulated Microcircuit (PEMs)
  Supplement to 311-INST-011, released by the
  Parts, Packaging, and Technology Branch(Code 562)

3
How Much Should A Project Spend On Electronic
Parts?

• Baseline The electronics must work.
• Additional They must keep working, in space,
  without repairs! Reliability is needed. This
  costs extra.
• The launch vehicle sets a particular cost and a
  particular reliability this sets one level for
  the extra cost.
• The worth of the mission also sets a level.

4
The Material We Work With

• Old style --- Hi-Rel  manufacturing lot
  controls, defined performance with strong
  protections against fraud, notification of
  changes, ALERTS, qual and screening programs.
• New style --- commercial PEMs  no lot controls,
  vaguely defined performance with no recourse from
  "bad data", no notification of changes, no ALERTS
  (better not try it!), mysteries about qualing
  and about screening.
• We need several kinds of special care to
  incorporate commercial PEMs into spacecraft.
  Detailing some of these is the purpose of this
  talk.

5
Why Use Commercial PEMs?

• Mainly availability! Hi-Rel parts are less than
  0.5 of the market gt many manufacturers have
  switched to making only commercial PEMs. 
  Increasingly, we cannot find Hi-Rel parts.
• High performance in small low-power packages.
• Cost is almost never a driver. Commercial PEMs
  cost more per part than the Hi-Rel devices do,
  after qual and screening. (Swift-BAT an
  exception.)

6
COTS PEMs DANGERS (Part 1)

• The key idea commercial PEMs are being carefully
  aimed at a precisely defined market. And it is
  not the spacecraft market.
• Desiderata  The lowest possible piece-cost /
  Highest initial performance.
• What is sacrificed for Desiderata Reliability
  extending past one year (or a few years). 
• ESD hardness / EOS robustness.
• Not even on the table Radiation hardness /
  Operation in a vacuum.

7
COTS PEMs DANGERS (Part 2)

• High performance and low-cost
• Crowding circuits gt "hot electron"degradations
  .   
• Thinning traces gt electromigration-induced
  breaking of traces giving electrical opens.
• Thinning dielectrics gt dielectric breakdown.
• Passivation layers are being omitted.
• All the above can lead to "wear-out" a failure
  probability that says low until after a threshold
  time, and then rapidly increases.

8
COTS PEMs DANGERS (Part 3)

• The ESD/EOS guard circuits are gone in some
  cases.
• The physical sizes of the parts are shrinking
  profoundly.
• Hence, the capacitance of each pin is decreasing,
  so that a given amount of charge deposited onto a
  pin will increase its potential more than used to
  be the case. 
• The dielectric thicknesses are decreasing, along
  with their breakdown voltages.
• ESD/EOS damage is now affecting many devices.

9
Qual Can The Part-design work?

• Possible failure modes are explored using highly
  accelerated tests. These usually use high
  temperature, and may use high voltage, and high
  humidity.
• Potentially damaging --- these parts are
  discarded.
• New parts gt new failure modes --- What are they?
  What tests would reveal them?
• Cannot increase temperature enough!
• End-game failures, like radiation softness.
• Advertising the failures of commercial parts?

10
Screening Was This Lot Made Correctly?

• We do find problems with lots (e.g., memory).
• Burn-in what temperatures and times?
• C-SAM many PEMs delaminate during burn-in.
• 50 sockets for 50 cent parts, and other
  screening expenses.

11
Screening House Adventures

• Do not read testing requests.
• Do not test whether all parameters are OK.
• Destroy parts using bad burn-in settings.
• Destroy parts using bad equipment.
• Station a friendly Watchdog at the house!

12
What To Do?

• Work closely with the manufacturers get data.
• Be vigilant for changes in the parts being
  supplied different wafer supplier, different
  mask, different foundry, different wire bonder,
  different plastic composition, different
  encapsulation parameters.
• Check that the flight parts work in their
  circuits.
• Set aside extra time for the qual and for the
  screening.
• Ensure alternative choices --- be cautious about
  committing to a unique part because it is sweet.