To:  Distribution

From: R. Gobin, C. Young, B. Craig

Re: Parlex Site Visit and Actions

Date:  January 21, 2005

 

 

We visited Parlex to discuss the results of the Teledyne testing performed earlier this week that clearly indicated Parlex pitch adapters produced after April of 2004 were the sole cause of the massive failures at Teledyne.

 

After presenting these results to Richard List , Phong Ho and Ferdinand Fong we discussed the implications of our findings, i.e. that Parlex pitch adapters produced in April of 2004 (Lot A) were acceptable and those produced in Lots B and C in September and November of 2004 failed catastrophically.   We walked through Parlex’s thinking on potential issues and what they had done to date.  As that discussion progressed it became clear that there was considerable uncertainty, and no readily available documentation, on the parts of the process performed outside Parlex’s facility.  It was not clear, for example, which nickel coating firm (Coatex or ExpressTech or perhaps another company) had performed the nickel coating on Lot A nor whether an electroless or electrolytic process was used.  Phong had done some research that indicated that electrolytic was considerably less brittle but they were unable to tell us what process was used on any of the lots in question. 

 

We talked over the effects of the Cu material from Dupont  (Dupont AP8525R roll annealed .5 oz copper in 36" X 24" sheets (copper/kapton/copper)Each lot has lit# for traceability. IPC spec designation: IPC-4204/11-E1E200/0-CU-W7-HS/HS. Copper Foil: IPC-4562) and learned that they were well aware of the increased performance of flex circuits aligned with the grain of the rolled Cu material but only tended to use it in that orientation only in dynamic applications when the circuits underwent a considerable number of cycles.  Otherwise it is more economical to manufacture with the grain perpendicular to the traces on the pitch adapter as it increases the number of parts that can be produced per piece of raw material.  It did not appear credible that there were any changes in the raw material although Parlex will produce certifications as part of their documentation effort as called out in action item 1 below.

 

We focused on a discussion of other changes in the process, beyond the possibility of another Ni plater (and type of Ni deposition).    After walking through the process, step by step, we found that one significant change was made in the basic procedure between the Lots.   In the Lot A parts, the areas outside the ‘fingers’ was masked and did not get plated with Ni or the second of two Au coatings.  In later lots, this mask (tape) was not applied and the whole part was coated with Au, Ni and Au.  It wasn’t clear why that change was made, although Richard List believed it might have had something to do with the aesthetic appearance of the parts that didn’t have the final Au deposition.  The effect of this is to considerably stiffen the portions of the pitch adapter used to apply tension to the part.  Perhaps this doesn’t affect anything but it does change the part mechanically in a significant way and is a real change.  Given that it cost them significantly more to coat the entire piece it was unusual not to have had a motivation for this change.

 

We found no evidence of changes in process at Parlex itself and did find that they were very willing to cooperate with resolving the problem.   We also found that that their documentation, particularly for parts produced before the summer of ’04, appeared inadequate.  The following action items were agreed to

 

1)     Determine the details of the successful April ’04 recipe: Parlex to examine their records and determine the actual process recipe for the Lot A parts and contrast them with the more recent Lots.  Documentation to be supplied to SLAC.  Due date: Monday January 24.  Actionee:  Richard List

2)     Measure the thickness of the plated material on a sample from all lots:  SLAC to supply 3 pitch adapters to Parlex, one each of Lot A, B and C.  These are then to be measured along the ‘bend’ portion of the circuit using X-ray fluorescence measurements at ExpressTech.  Up to 1 hour of diagnostic time will be expended per part with a minimum of 5 sites and a desired number of 20+ sites to be measured and reported.  Due Date:  Parts delivered to Parlex on Monday morning, test results by Wednesday January 26.  Actionee:  Richard Gobin to deliver parts.  Phong Ho to obtain measurements in concert with ExpressTech.

3)     Recreate the plating recipe from April and contrast it with the later recipe: Parlex to take existing ½ oz material and perform the processing steps through imaging on a number of parts.  At this point, several Ni/Au coating sequences to be performed, duplicating the April recipe and the recipes from Fall of ’04 and, as possible, some reasonable variations on these.  These parts, without final gold overcoat, will be tested on the SLAC 1mm bending fixture.  Test target date is Wednesday January 26 at SLAC.  Coating steps at Coatex and ExpressTech to be witnessed by a SLAC representative.   Due date: Wednesday January 26th.  Actionee:  Phong Ho, Ferdinand Fong for Parlex, Richard Gobin, Charlie Young, Bill Craig for SLAC.

4)     Prepare for resumption in production as early as Wednesday January 26th:  Assuming success once the plating recipe is resurrected the goal would be to immediately resume production with full process documentation on all steps, including grain orientation and all plating steps at Parlex’s subcontractors.   A remaining issue on dimensional tolerance must be resolved by transferring Charlie Young’s metrology into adjustments in the artwork and the numerical controlled drilling machine program.  Raw material is available now at Parlex for another project.  One half of this existing stock will be reserved for use on the pitch adapters, this is sufficient for several hundred pitch adapters.   Due Date:  Wednesday January 26th.  Actionee:  Richard List to reserve material, Ferdinand Fong to work with Charlie Young to ensure the correct changes are made to artwork and program. Phong Ho to document process and ensure no changes in plating or process without written permission from SLAC. 

 

After these actions are complete and a decision made to resume normal production SLAC QA to be informed in advance of all key steps in the production of the next lot and will be afforded the opportunity to be present throughout.  Richard Gobin will obtain an upper and lower limit on the Ni thickness spec and pass this along to Parlex.  SLAC will require continuing thickness measurements from each lot with the details of number and location of measurements to be determined after the analysis of the data from action item 2 above.