Trip Report – 2/10/05 Visit to Teledyne

 

 

The primary purpose of this trip is to resuscitate MCM production.  Production was started last Thursday 2/3/05 with an agreed upon goal of starting five units per day through 2/11.  We achieved that for the first three days; however, there were no new MCM’s on either 2/8 or 2/9.  The immediate cause is failure to deliver any parts to the factory floor. 

 

I arrived on site at 8:30 am, and there were still no parts available.  Material and travelers for 25 MCM’s showed up shortly before 9.  Jose Luis, Carlos and Rigoberto had the first group of 4 boards in the oven by around 10:30, and the first of these were trimmed by 12:30.  The second group of 4 was prepared while the first was in the oven.  This group came out of the oven shortly before 15:00.  There was not enough time to cool and trim before the end of their work day.  I conclude from this that if material can be made available at start of shift at 7, we can do two complete bonding and trimming cycles a day and prepare material for the third cycle. 

 

Ana C. worked in wire bonding.  She called for help because of operational difficulties.  The automated program would not bond to the Pitch Adapter, and she had to manually guide it in for each trace.  She was part way through one board, and started on a 2nd one to find the problem was not unique to one board.  Steve(?) looked at it, and concluded that the traces were too narrow to bond to.  Under high magnification, it was obvious that traces are narrower than the gap between them.  Width looks uniform.  We measured one trace on MCM S/N 2082 to be 72 mm.  Another board looked similar but we failed to record either the S/N or the actual width.  Both used PA from Titan.  We measured a board used for wire bond pull tests, and found trace width to be 103 mm.  Trace width and gap are approximately the same size as one would expect.  We also measured a new MCM that has just been bonded, and found a trace width of 102 mm.  This one used Parlex PA made last year.  All these measurements were made on the same machine with the same technique.  I called Richard Gobin and asked for a check on new Parlex PA.  He measured several traces using calipers and a microscope, and found trace width to be around 120 mm.  Traces and widths are about the same size.  Tom Nakashima has been asked to make more precise measurements.  I conclude that the problem is with the Titan PA.  There is no reason to panic regarding the new Parlex PA.  I would suggest that we do not measure trace widths as a routine part of Parlex PA acceptance.  Spot checks may be useful.  

 

The initial diagnosis was the traces were too narrow to bond automatically.  It would have to be done by hand at a rate of roughly one board per day, compared with one every 30 minutes when done automatically.  Brian Caplen came to a different conclusion when he looked at the problem.  He believes that there is more than enough room to put down the wire.  What failed was the pattern recognition step.  The machine was looking for a trace ~100 mm wide, could not find it, stopped and waited for operator intervention.  A relatively simple reprogramming could resolve this problem.  The current program is saved and will be reused for Parlex PA.  We have 10 MCM’s with Titan PA.  Five are ready for wire bonding; five are scheduled for die attach next Monday.  Teledyne is to provide us on 2/11 with a cost estimate for the reprogramming.  There is of course no assurance that it will resolve our problems without additional tweaking.  Assuming the cost is not exorbitant, I am inclined to go ahead.  We can use these suspect parts to continue trailblazing.  We have to be careful it does not turn into a blackhole that sucks up lots for resources for 10 boards.  Brian and I are scheduled to talk about this Friday morning. 

 

Jerome Lepulu and I met with Elvia Paez and Ayub Patel to understand Teledyne’s production plan.  Elvia is confident that they can deliver 25 finished MCM’s a week with current resource levels.  This has all yields factored in, so more than 25 units have to be started each week.  They are working on increasing the rate beyond this, and will present us with a proposal next week.  They gave us the schedule for finishing the first 10 MCM’s.  Note that this was made up before we were aware of the wire bonding problem.  Also, Teledyne is likely to shut down on 2/21 for Presidents Day.  We should modify this accordingly.  In particular, Teledyne had requested Albert Nguyen to be there for MIP-2 on 2/14 and 2/16.  This may have to be rescheduled. 

 

Item

Date (1st 5)

Date (2nd 5)

SMT

Done

2/10

Connector

Done

2/11

Die attach

2/10 (1 unit, others done)

2/14

Lead bonding

2/10 – 2/11

2/15

MIP-2

2/14

2/16 (starting noon)

QC

2/15

2/17

Encapsulation

2/16

2/18

Conformal coating

2/17

2/21

Final Test

2/18

2/22

Final Inspection

2/21

2/23

Ship

2/22

2/24

 

Ayub expects the 25 new units to be ready for MIP-1 inspection on Monday 2/14.  He is not sure about the dates for each step.  He will keep us updated regularly.