Trip Report Teledyne 1/27 to 1/28

 

 

The primary purpose of this trip was to perform bonding tests to see if electroplated nickel Pitch Adapters can be used in MCM Production.

 

 

Participants

 

NASA/Swale Paul Baird

NorthStar SciTek Richard Fernholz

SLAC: Richard Gobin

Charlie Young

Teledyne Brian Caplen

Peter Hansen

Jose Luis Saldana

Manuel ?

UCSC: Robert Johnson

 

 

 

Material for Bonding Tests

 

We have 11 PWBs that passed visual inspection at SLAC on the radius. Jose Luis also inspected them. He rejected one.

 

We have PAs from three vendors: Parlex, Dyconex and Titan. Parlex PAs are from an R&D run. No electrical tests were made, and dimensions were not checked. They are serialized differently from production PAs by panel number (P5, P6, P7, P8 and P15) and circuit number 1 through 3 within a panel. Two panels were plated to between 50 and 100 minch of nickel, and the other three were placed to between 100 and 200 minch of nickel. An electrolytic plating process was used. One thin (P5-1) and two thick circuits (P7-1 and P8-3) were tested earlier on mandrels for trace cracking. We decided against using them for bonding tests, in case the pre-bend leads to abnormal failures.

 

There are three R&D PAs from Dyconex. They have not been electrically QCed or checked dimensionally. We assigned S/N Y0004 through Y0006. The nickel was plated using an electroless process. Thickness is unknown.

 

We received 15 PAs from Titan. They have not been electrically QCed or checked dimensionally. The vendor assigned S/N 0001 through 0015. Nickel plating was done electrolytically. Thickness is unknown.

 

 

Teledyne Bonding Fixture Tension

 

We experimented with three configurations. One is normal tension. We follow the nominal bonding procedure.

 

Another is referred to as zero tension. The PA is cut in the long direction on the Teledyne side. The location of the cut is chosen so the Teledyne tail is not captured by the movable arm, and therefore will not apply tension directly. The top of the movable arm sweeps the PA around the 1-mm radius and constrains the tail so it cannot swing back out.

 

The third is so-called half tension. The PA is cut as above. The two sides are then taped back together with Kapton tape. The adhesive allows the sides to come apart a little, resulting in intermediate tension. The actual tension is uncontrolled. It can vary along the length and can also vary over time.

 

 

Bonding Tests

 

We decided to concentrate on PAs with nominal nickel thickness, i.e. ignore the two Parlex panels with thin nickel layer. The first test used two Parlex and two Titan PAs under normal tension. The second test used only Parlex PAs, 2 at zero tension and 2 at half tension.

 

Test No.

Fixture

1

2

3

4

1

PWB S/N

2206

2393

2329

2186

PA Vendor

Parlex

Parlex

Titan

Titan

PA S/N

P8-2

P7-2

0001

0001

Tension

Normal

Normal

Normal

Normal

2

PWB S/N

2189

2208

2188

2205

PA Vendor

Parlex

Parlex

Parlex

Parlex

PA S/N

P7-3

P8-1

P15-1

P15-2

Tension

Zero

Zero

Half

Half

 

There are some voids in all cases. Jose Luis judged them to be at a not unusual level.

 

One Parlex and one Titan from the first test have no cracked traces. We found a few cracked traces in the other two. The numbers of cracked traces are below the QC cutoff, and the boards would have been accepted.

 

All four MCMs from the second test have no cracked traces. The Italian side has more waviness at half tension. We speculate that it is caused by tension variation along the length.

 

Excess PA was trimmed off and identified. We can examine the alignment holes for distortion.

 

 

Dyconex PA Bonded on G&A Fixture

 

Prior to this trip, Robert bonded Dyconex PA S/N Y0003 to PWB S/N 2121 at SLAC using the G&A fixture. We trimmed off the excess PA at Teledyne, and identified the pieces for further investigation.

 

 

New Fixture Design Review

 

Richard Fernholz presented his design to Baul Baird, Brian Caplen and Peter Hansen. Robert Johnson and CY were also present. There is some concern about alignment of traces needed at the mil level. Teledyne has been using the four small holes at the corners of the PA rather than use the large holes as is proposed here. Teledyne has to adjust those alignment pins from PA lot to PA lot. We confirmed that the new fixture will be smaller than the existing Teledyne fixtures, and can therefore fit at least four into the oven.

 

o      Allow for adjustment of flex location. There should be two degrees of freedom: distance between the two pins and location of the two pins relative to the rest of the world.

o      Add (rough) guides to the base to ensure that the movable part stays within the capture range of the alignment pins.

o      There will be no mix-and-match of parts among the fixtures. Everything will come as matched sets.

o      Hard anodize black for short and red for tall.

 

Richard will distribute drawings on Monday 1/31/05.

 

 

Discussion on Encapsulation Test Results

 

Teledyne had attached ASICs to MCMs that failed bonding with massive trace cracking. These were then wired bonded, encapsulated and conformal coated. Three such MCMs were sent to GSFC to see if we have fixed the problem with delamination of the encapsulant. We have received the report, showing large regions of void. A Teledyne expert on Sonoscan joined this discussion.

 

Our primary interest is in the region where Teledyne wire bonds to the PA. It was explained to us that the operator has to set a narrow timing gate to pick out this depth and avoid signal from voids at other depths. We were therefore confused by features in other areas that can only be at different depths. For example, the images show the weave of the PWB material throughout the board. They also show large voids in the controller chip areas. These chips were not on the boards. We wanted to contact the person who made the measurements, but were not able to find him. We will follow up next week.