Barrel Harness Status

1
Barrel Harness Status

• Harness Production Status in Taiwan.
• Harness Tests at RAL.
• Problems found.
• Resonant wire bond breaking.

2
Taiwan Status

• Batch process. Sub-assemblies well advanced
• LMTs and PPB1s completed in Ljubljana
• Opto-packages /352
• PPB1/LMT/if PCB
• MT terminated ribbons.
• 8 production harnesses sent to RAL.
• Full BER testing after thermal cycling in Taiwan.

3
Taiwan Status
4
RAL Harness Testing

• Full test on receipt of harness at RAL.
• Critical LMT and fibre lengths measured.
• PPB1 installed with TPP in service cage.
• HV insulation at 500V.
• Continuity open and short circuit test.
• Light o/p VCSELs.
• PIN responsivity.
• Data and TTC BER scans.
• Clock check
• Redundant clock and commands.
• Long BER test.

5
Harness Mounting on Barrel

• Trial with test harness on B3 using vacuum
  pick-up jig.

6
Harness Testing on Barrel

• Electrical Continuity and short tests repeated ?
  check for damage during mounting.
• Power VCSELs and responsivity PINs.
• Long term BER tests for data and TTC links.
• Use Minglees simple cheap BER test system to
  test 4 harnesses in parallel.
• Run overnight/weekend.
• Prototype for single harness test works fine.
• Minglee has parts for final system and will
  provide RAL with a system before next batch of
  harnesses arrive.
• Final verification of harness on receipt of
  barrel at Oxford.
• Loop back PCBs mounted on dogleg loop back TTC
  data/clock ?data links.
• Checks normal clock/data and redundant
  clock/data.
• Resistors used to check continuity VDD, VCC etc.

7
Status for 8 Harnsses

• Four perfect harnesses
• Passed all tests
• Two failures
• Both had one dogleg with no PIN responsivity
  probably ? broken fibre inside heat shrink tube.
• Send these harnesses back to Taiwan for
  post-mortem.
• Minglee will devise a better protection for the
  single fibre inside the heat shrink tube.
• Dont produce any more harnesses until this
  problem is understood.
• Two questionable harnesses
• One dogleg gave out a jittery redundant clock
  when the select was off. Not a select line
  problem but due to DORIC4A.
• This feature was not checked at wafer level or by
  Taiwan.
• Will be checked in future at single dogleg level.
• What do we do with these two harnesses ?

8
Resonant Wire Bond Breaking

• CDF discovered that wire bonds that carry
  currents that depend on L1 trigger can break in
  magnetic field.
• Failures occurred during anomalous trigger
  conditions
• Trigger torture tests.
• High occupancy mode in SVX3.
• These caused system to be readout at fixed
  frequency.
• CDF tests showed that wire bonds have high Q
  resonances in the 10-100 kHz range and that bond
  wires fail after minutes to hours if operated
  on resonance.
• Potential problem for VDC ? VCSEL wire bonds.
• Length of bonds 2mm, diameter 25 mm. (similar
  to CDF).
• Current 0?10mA for sending a 1. Readout after
  L1 consists of equal 0s and 1s ? equivalent to
  a pulse of 5mA amplitude for duration of readout
  (smaller than CDF).

9
Resonant Wire Bond Breaking

• First quick and dirty tests used
• bonds of correct length on a test PCB.
• Birmingham superconducting magnet B6T,
• Sine wave and square wave with current up to 40mA
  (Lorentz force an order of magnitude greater than
  for ATLAS operation).
• No bond wires broke.
• Second test.
• Used video camera to image wire bonds.
• B0.5T (fringe field of magnet deteriorated the
  resolution of the camera).
• Sine wave of amplitude 20 mA (?similar Lorentz
  force to ATLAS operation).
• Scanned frequency and found resonances around 15
  kHz (? photos).

10
Resonances
Clear resonance seen with current pulse on
Current pulse off