Trip to Teledyne on 12/29/04

 

Richard Gobin and Charlie Young

 

 

MIP-1 inspection was the original motivation for this visit to Teledyne.  All recent MCM’s have been rejected due to excessive cracked traces; therefore, we focused on this problem instead.  Three people from SLAC participated in the meeting:

 

Richard Gobin

Jerome Lepulu

Charlie Young

 

There were three people from Teledyne:

 

Brian Caplen

Jose Luis Saldana

Rhonda Santiago

 

 

Cracked Trace Problem

 

Teledyne re-started production on 12/20/04 with an initial batch of six MCM’s.  [This was reported last week to be 8, but their log clearly shows 6.]  A second batch of 8 was made on 12/22/04, and a third batch of 4 on 12/28/04.  All 18 failed inspection after bonding the PA.  Jose Luis reports that every board has a large number of cracked and open traces, spanning more than half the length of the board in some cases.  The log does not contain the actual number of bad traces, but a number of entries say “open and crack at U3 – U26”.  Some only have notation of broken power traces.  The last 4 pieces were still in their gluing fixtures and had not been trimmed.  There were no notes yet in Teledyne’s log.  Visual inspection of samples from all 3 batches confirms Jose Luis’s observation of massive numbers of cracks.  This is qualitatively different from earlier runs. 

 

 

Observations

 

Teledyne had seen large numbers of cracked traces in the past when the radius on the PWB was not well done.  They inspected this group of bare boards and found “ridges” on the board as illustrated below.  Richard and Charlie were not able to discern the features even with the aid of a microscope.

 

 

Teledyne then showed us examples after PA’s have been attached.  The ring of light around the microscope’s lens becomes effectively two point sources in cross section.  If the radiused part is uniform along the length, these point sources show up as straight lines.

 

 

The following two pictures are from a “good” MCM.  It was rejected in any earlier run, but there were no problems with cracked traces.  We can make out the last two digits of its serial numbers and will refer to it as S/N xx39.  The pictures are from around U20 and U13 respectively.  Note that the two bright lines are fairly straight.  The traces fade away in the direction orthogonal to the bright lines due to poorer reflection.  This fading is similar for all traces. 

 

 

 

 

The next set of pictures is for S/N 2258, one of the 4 made on 12/22/04, from locations near U17, U19 and U24 respectively.  The bright lines are wavy and non-parallel.  The traces do not fade away at the same rate.  This is interpreted to mean that the radiused corner has variations along its length.  Notice the cracks in many traces, and sometimes multiple cracks in each trace.  If they were caused by “ridges”, then several ridges would have to be present. 

 

 

 

 

 

The next set of photographs is for S/N 2167, also made on 12/22/04.  The first two are from area near U12, and the third comes from U22.  The traces go from bright to dim abruptly and the transition is at a different point for different traces.  Note the number of cracked traces. 

 

 

 

 

 

Known Process Changes

 

We are aware of only one change that might be related to this problem.  The boards are radiused by Diamond Tools with a custom tool bit.  Small irregularities were observed, and a step was added in early November.  SLAC manually “deburred” the bare boards before shipping them to Teledyne.  We did not control the radius while sanding, relying on the operator’s finger to conform to the board contour.  The applied pressure depended on the operator.  It was speculated this could have resulted in localized corners and flats, which led to traces cracking.

 

 

Other Possibilities

 

The PA’s may have come from a different production run than the ones used earlier, and may be more prone to cracking.  The radius may have been put on incorrectly, either from a worn bit or from operational error. 

 

 

Action Plan

 

Teledyne was convinced that the problem was with the “ridges”.  We were willing to treat it as a non-exclusive working hypothesis.  Here are some of the things to be pursued. 

 

 

 

Other Items

 

 

Rework

 

Teledyne said most of the boards we sent back several months ago for rework were done.  They were not shipped earlier because there was no Source Inspector to give OK.  Jerome’s presence will take away that excuse.  Teledyne promised a full accounting by middle of next week.  We told them we expect to hear “everything has been shipped to SLAC”. 

 

 

Production Schedule

 

The schedule that Teledyne provided last week is no longer valid.  Production will re-start when the problem of 100% rejection is resolved.  It is anticipated that the delivery schedule will slip by that amount. 

 

 

PA Pull Test

 

As reported before, the pull tests have been done.  Brian has not finished analysis of the data. 

 

 

Encapsulation Test

 

We will use some of the 18 rejected boards to see if replacing silicone based Kapton tape with acrylic based tape improves the bond.  Teledyne will attach dead dies, and wire bond 4 boards.  [Having dies means we don’t need to modify the flow to compensate for the volume.  Wire bonds may be useful during QC.]  They will be cleaned in the usual way and then encapsulated.  Rhonda will contact Dave Rich for permission to carry this out under the $10K testing contract.  They will be subjected to the usual handling at SLAC, including temperature cycling and burn-in.  Quality of encapsulation will then be evaluated with sonogram, sectioning, and other techniques.  

 

 

Cover Tests

 

The first test was successful.  We will continue testing as a back-up solution to encapsulation.  SLAC will get new covers made, using an acceptable material and with more precise dimensions.  Teledyne has no action until the new covers are available.

 

 

Vacuum Bagging

 

When shipping from Teledyne to SLAC and from SLAC to INFN, the MCM’s are vacuum bagged.  We are not sure why this is done.  We do not know why this should not be done with the current design.  On the other hand, if we were to use covers instead of encapsulation, vacuum bagging may deflect the covers and damage wire bonds. 

 

 

Clean Room

 

There was a leak in the roof near our production area.  Water was being collected in a bucket. 

 

We observed no production activity in the clean room during this visit; there was not a single operator there.  The construction area has grown.  It is separated from the production area by a sheet of plastic.  We observed workers going through the production area in street clothes. 

 

Teledyne said these people were involved with construction and/or repair of the roof, and it was impractical to make them suit up.  We were examining rejected parts, and no potentially accepted parts were exposed.  12/29/04 was scheduled as a clean-up day to scrub the clean room in preparation for production on 1/3/05.  It was not obvious that the construction and repair would be done by then. 


Appendix I – Serial numbers.

 

 

Date

Serial Number

Fixture

Trace Problem

 

MCM

PA

>8

power

 

12/20/2004

958

2134

1

x

x

 

 

2346

2138

2

x

x

 

 

2248

2137

3

x

x

 

 

2402

2131

4

x

x

 

 

2017

2147

3

N/A

N/A

 

 

2397

2146

1

N/A

N/A

 

12/22/2004

2175

2148

1

x

 

 

 

1364

2132

2

x

 

 

 

2270

2112

3

x

x

 

 

2130

2159

4

x

x

 

 

2057

2160

3

x

x

 

 

2140

2173

4

x

x

 

 

2167

2172

2

x

x

 

 

2258

2171

1

x

 

 

12/28/2004

2097

2187

1

N/A

N/A

 

 

2011

2185

2

N/A

N/A

 

 

2283

2182

3

N/A

N/A

 

 

2053

2196

4

N/A

N/A

100% reject

9/7/2004

11292

674

1

 

 

 

 

11293

677

2

 

 

 

 

11296

681

3

 

 

 

 

11298

707

4

 

 

 

9/8/2004

861

373

1

 

 

 

 

11186

368

2

 

 

 

 

11234

388

3

 

 

 

 

11295

423

4

 

 

 

 

11201

334

1

 

 

 

 

11203

335

2

 

 

 

 

11227

336

3

x

x

 

 

11231

338

4

 

 

 

9/9/2004

11188

346

1

 

 

 

 

11228

342

2

 

x

 

 

11229

343

3

 

 

 

 

11233

348

4

 

 

 

9/13/2004

869

574

1

 

 

 

 

913

572

2

 

 

 

 

12239

576

3

 

 

 

 

11237

573

4

 

 

 

 

989

879

2

x

 

 

 

992

863

3

 

 

 

 

11000

843

1

 

 

 

 

11352

884

4

 

 

 

9/14/2004

11275

690

1

 

 

 

 

11239

697

2

x

 

 

 

11272

735

3

 

x

 

 

11284

748

4

x

 

 

9/16/2004

11243

788

1

 

 

 

 

11248

798

2

 

 

 

 

11266

809

3

 

 

 

 

11353

816

4

 

 

 

9/17/2004

11178

210

1

x

x

 

 

11213

239

2

 

 

 

 

11297

259

3

x

 

 

 

11300

631

4

x

x

 

9/20/2004

11269

320

1

 

x

 

 

11235

321

2

 

 

 

 

11242

324

3

 

x

 

 

499

325

4

 

 

 

 

11003

326

1

 

 

 

 

11199

327

2

x

 

 

 

11037

328

3

 

x

 

 

11195

329

4

 

 

 

9/22/2004

995

337

1

 

 

 

 

11019

322

2

 

 

 

 

11093

716

3

 

 

 

9/27/2004

764

672

1

 

 

 

 

11440

671

2

x

 

29% reject