February 10, 2004
PRR at Teledyne for MCM (They call it TMCM)
Robert Johnson – overview of tracker MCM
- Polyswitches to cut off shorted MCM. If an MCM shorts then loose everything above it but not below.
- 98% efficiency on track. Allocating 0.5% efficiency to bad MCM channels. Translates into 8 channels out of 1536.
- Want MCMs when delivered to read out 100% to left and right so no failures allowed at the MCM level after production. We maintain this requirement through delivery to Italy.
- Even 100% functional MCMs still need to meet dimensional and cleanliness requirements.
- Reports GTRC7 MCMs work from 1 MHz up to 25MHz and TOT bug fix has been verified.
- No outstanding design issues in MCM/SSD subsystem.
- Drawing package in release. Includes the design files (Gerber and so forth).
- Have to align pitch adapter, bias circuit, and two SSDs.
- Introduced concern about perpendicularity. Can wirebond – but studying speed of production.
- Introduced gluing fixture at G&A to control perpendicularity. Now this is considered a back-up that may be introduced into the process later.
- In summary, moving forward with current baseline. G&A confident they can accomplish the bonding but there may be a hit in the production rate.
Robert Johnson – Mini-tower lessons
- Mini-tower has 6 layers with prototype MCM. June 03 version used GTRC6.
- GTFE chip had problems with the output bias that have been fixed.
- GTRC problems corrected.
- Omnetics connectors were improved and introduced some new problems. Nick Virmani believes problems are fixable and the fix is known.
- Initial assembly had bias shorts due to fibers puncturing the bias circuit. These problems have been fixed on the second mini-tower.
- High frequency of GTRC and GTFE replacement. Now test 100% at wafer level, before the chips are diced. Also implemented a mil-spec die inspection. Better, but not perfect. Had to replace 1 die that tested good on wafer but failed at the first MCM test.
- Lost some wire bonds on the bias connection. Believe this was at the time of encapsulation. Have now doubled wire bonds from 2 to 4 for each of the bias connections.
- Had some flaking of the encapsulation with initial mini-tower. Had flaking of the second layer.
- Improved wire bond process control. Had significant loss of connections around the bend and improved the machining to make a better transition on the flex circuit. Appears to have solved the problem. Probing pre-production MCMs and have found zero broken connections in 6144, (4 MCMs) channels, even after thermal cycling and burn-in.
- Worst problems with the pitch adapters. On mini-tower the pitch adapters turned out to be too long. They look like they tend to shrink during processing. The last batch now looks too short. We have measured a bunch of the circuits at SLAC and are screening these. Throwing out any more than 100 microns out and 30 microns across a ladder. Thinks we will toss half of them doing this. Nick believes we can improve this if another buy is needed. Currently have the entire flight lot but this high rejection rate will likely require another buy. These are coming from Parlex. Jerry thinks the requirement on the drawings is 200 microns and that current lot meets this specification.
- Fixed problems with trace attachment.
- Trimming of the flex circuit process have been tweaked and now seems fine. Went to trimming beforehand. Also had some early problems with trimming too far or too close to the board. Circuit needs to trimmed cold and cutting set-up has been adjusted. Don’t seem any problems now with this. Will reject if we see this now.
- Had to move to Holt machining from DDI to get radius. Holt was doing a good job but had a problem with a worker getting fiberglass in his eye and now doesn’t want to machine any more. Have about 100, with about 750 to go. This just happened yesterday. Teledyne requested that they are notified when PWBs from new shop are inserted into the production run. Need to find another vendor and certify them with non-flight boards. Nick Virmani reports he has some references.
Robert Johnson on design status
- Design is complete and have drawing package in release.
- Have two drawings here, one for short and one for the long MCM. The tall MCMs go on 4 of the trays have thicker tungsten and thus one of these MCMs are “tall”. So only 4 of the 36 MCMs are tall per tower. These are discriminated only by part number.
Robert Johnson on MCM parts.
§ Two panels have failed coupon testing. Speed of coupon testing is an issue. Need to keep ahead of the production.
§ PWBs machined for the first flight lot and this is moving forward.
§ Selected next set of boards to be machined from panels already selected.
- Pitch adapters from Parlex. Still need part control approval for pitch adapter. Nick doesn’t see any problems with this.
- GTFE completed radiation testing and ready to go. Final approval by the parts control board awaits qualification testing at the MCM level. Flight quantities need to shipped to Teledyne.
- GTRC7. Need to repeat TID. Shipped enough parts now for 3 towers.
- More slides exist on the website.
Dave Rich on QA survey on the entire Teledyne facility.
- Performed on 14 11 04. Tracy produced the final report.
- All actions closed!
- Teledyne is ready to go.
- Using internal criteria, has been deemed equivalent to NASA standards. Conformal coating was the one exception. Unique requirements have been captured on the drawings.
- Darren reports no drop shipment on this program. He correctly identified this as a LAT issue.
Brian Caplen, Teledyne, on production flow, (charts also on the website – but no overhead slides).
- 25 units/week production rate.
- Gone to a mixing gun that mixes during the operation for epoxy. Eliminated problems with bubbles in the pre-mixed epoxy.
- Flex circuit bonding. Roll the flex circuit around the tool.
- Mandatory inspection point 1 to verify trimming operation. Still need to identify all the MIPs.
- Still some work to do to define SMT component attachment in production. Planning to put MCMs in a tray to be able to flux clean several MCMs at once. The basic problem they have is that these boards are longer than the usual jobs they are set up for. AK255 is solvent used in flux cleaner. More aggressive than alcohol. Should remove adhesive residue from tapes used.
- Planning automated ASIC attach for flight production run. Pre-production run used hand attachment. Nick asked about die attach epoxy quantities being defined.
- Working to improve a wire bond protection/carrier. Most damage to wire bonds are due to handling, not production. Had a fairly long discussion on possible changes to avoid handling problems with the many different inspection steps.
- Another issue is how much cleaning is really required. Plasma cleaner installation at G&A is belt and suspenders issue. It should improve bond strength. Nick Virmani also lead discussion on cleanliness. Teledyne wanted to take action to “remain sensitive” to this issue. I don’t think this is really an action.
- They do have a die rework step in the flow but still have some issues with the pads, need to reprogram to change the position of the first bond to allow for a rework area. They wish they had made the pads bigger to facilitate rework. Have enough rework to make it hard to live to without rework at the MCM level.
- Currently have 5 people part time on this project. These people will be full time as we transition to production. SMT is shared but the concept is to establish a “cell” for GLAST production. Wire bonding machine is dedicated to the project. Total production run will last 8 months. Because of the break-up Teledyne into profit centers we are actually a sizable job.
Robert Johnson on MCM Test Flow
- Darren brought up the burn in temperature. The current temperature of 85C is limited by the polyswitches. He didn’t have another position, just wanted to make sure all agreed with a burn in of 85C for 168 hr. Burn-in plan has been agreed to by the Parts Control Board. Nick Virmani says he is the only person who can change the burn-in plan. It was noted, however, that the final version of this document has not been released.
- There are two electronic MCM test systems. One is at Teledyne and one will be sent soon form UCSC to SLAC. The Teledyne system will be operated by UCSC people.
- EGSE system can’t test all addressing and MCM levels.
- 20 MCMs post screened MCM from flight lot. 4 sets of 5 taken through qualification testing.
- Resolved to use 20 Pre-production MCMs for qualification testing. These have two differences between the production run, the older version of the Omnetics connectors and the GTRC6 chip version. All test scripts should work. Really need to do this now to have some risk reduction before the production run begins to hit full swing.
š Nick took an action to work with GSFC to ensure this will be a valid qualification test but agrees the problems should be workable.
š Need to start planning for these tests at UCSC and SLAC today!
- Pitch adapter has been a big issue. Nick agrees with G&A in that they need a flat surface. He suggests the use of an autoclave to improve the quality of the pitch adapter bond surface. Teledyne needs direction to do this. This has been considered but not tried. Use of vacuum could allow better removal of bubbles and let the flex adapter lie flatter on the tooling. Nick is 100% sure this will improve the process.
Dave Rich on Inspection Points.
- Nick wants solder inspection to be done, definitely, and maybe some others. Nick wants to add some additional things to the process control document.
š Nick to provide updates to the inspection points.
Dave Rich on Schedule.
- Proposed to start production run 17 02 although Dave started to call this a goal. I don’t like this “goal” idea.
- Still have a bunch of stuff to inspect and ship to Teledyne.
- Omnetics should ship the connectors today.
Robert Johnson on NCR and Pre-production Status
- Planning to change bond on die off center to facilitate rework.
Rhonda Santiago on Purchase Order Status
- Have open issues to work on just released drawings.
- Some out of scope issues with SOW. These are likely to get resolved quickly.
- Rhonda thinks they are fully ready to go.