AFE Meeting, March 29, 2001 George Ginther, Stefan Grunendahl, Fred Borcherding, Paul Rubinov, Alan Bross, Harry Weerts, John Womersley, Hugh Montgomery, Laurent Duflot, Levan Babukhadia, Drew Alton, Don Lincoln, Paul Grannis/video (minutes). ** 1. Bare board production Sanmina started the production of bare boards on Mar. 28. We expect them to be delivered to the New Hampshire facility for stuffing on Monday April 9. We have told Sanmina that we wish to have control of the decision to proceed with stuffing. They have not come back to say that they heard us. Fred said he would contact Sanmina again Mar 30 to verify that they agree to us controlling the start of stuffing. ** 2. Spare boards 28 spare boards had been proposed to allow us to read out the full CFT stereo fibers in advance of the availability of AFE12 boards. We had agreed previously to order these boards as insurance against AFE12 delays. If AFE12 were not available for a long time, we could use AFE8's in their place at the expense of the dual range settings for preshowers. However, the MCM availability is such that we would need to reclaim MCM's from AFE8's to use on AFE12's -- something that would be very difficult to do at best, and likely to result in failures. We asked for an accounting of the inventory of MCMs, and the minimum number that would be needed if we read out only the CFT stereo on AFE8's (not the CPS axial). Fred agreed to provide this. Information presented at the meeting was that we presently have about 1900 working MCMs. We need 1100 for the AFE8 boards, and 744 for the AFE12's. Sanmina contacted us this week about the 28 extra bare boards, needing an answer within a day as they would add them to the production run. They asked for $1313 per board, making the 28 extras roughly the same cost as the full order of 146! Marvin told them that we would not want these at this price. Electro-Circuits has offered us RevB boards, reworked from the earlier unsuccessful production at $328 per board. The experts feel that using these RevB boards with the attendant green wires would be counterproductive. Buying spare AFE8 boards from an independent vendor was believed to be possible -- certainly at the large cost being asked by Sanmina. Fred felt that it would be possible to do a procurement from start to finish in 6 weeks by streamlining the bid process. Fred agreed to call Sanmina to see if there was room for negotiation pending decision from Harry and Mike Tuts on whether money was available. After the meeting, they determined that there was not money left in the project, even at the $10K level. It seems that the prospect for getting the insurance from the extra boards is getting dim. If so, this means that the AFE12 project is of critical urgency if D0 is to have a working fiber tracker. At present, John Anderson is leading the AFE12 project; Pat Sheahan is doing the schematic entry for the motherboard (very similar to the AFE8) and Donna Kubik is doing the schematic entry for the daughterboard. ** 3. Paul Rubinov discussed the state of tests at Lab3. The differential non-linearity (DNL) problem found last week is not solved. The problem is that there are strong peaks at roughly 8 count spacings for normal clock conditions (and some evidence for other periodicities like 32 counts). There is no guarantee that the previous belief that photopeaks were being observed is real-- they could have been artifacts of the DNL. Tests using larger SVX gain to space out the photopeaks seem to show evidence for photopeaks on one channel; Paul was unwilling to declare the photopeaks real however due to the evidence for other than 8 channel DNL peaks being present. It seems likely however that the photopeaks, at least on this channel are real. The problem seems to be worse at large signal sizes. A variety of tests with various clock changes have been done. Using an external clock operating at different frequencies shows that the size of the DNL effects can be made better or worse as one changes the frequency. Paul has a range of plots available, with brief indication of the conditions for each. Captions and plot list are linked next to the these minutes. The current plan is to take the tests of the DNL problem back to the test stand at DAB3; these stands have been down for the past week due to a variety of problems and upgrades, but should be available on Mar. 30 for use. Mike Utes will take charge of the SVX DNL testing at DAB6. Paul Rubinov will continue tests at Lab 3 with the VLPC setup. We should be in a position to do two board (LH and RH) tests by middle of next week; the cryo control functions are OK, but there is not a working temperature control yet. It would be possible to return to tests of the discriminator threshold variances, though without a good calibration using photopeaks, the accuracy of our tests would be compromised (by ~20%?). It should be possible to get some calibration from the one channel that has seen (we think) photopeaks with increased SVX gain and transfering the calibration to other channels. Some discussion ensued about operating the AFE without SVX data available. Though possible, our specs say we require the digitized output for calibration and we were unwilling to retreat at this point. ++++ note added Mar. 30, 2001 from Fred Borcherding: ++ ++ Today we ran board #5, the one under use at lab3, at the south test ++ stand in DAB3. We saw the 'bad' behavior seen at lab3. We then ran ++ board #1 under exactly the same conditions. We did not see any hint ++ of 'bad' behavior. This is a significant result, but it is too early ++ to jump to any conclusions. ++ ++ Paul has taken board #1 to lab3 to repeat some of the tests there. ++ Meanwhile we are inspecting board #5 using our checkout procedures, ++ and recording any 'differences'. We will also, later today, look at ++ a third board at DAB3 to see if there is any hint of bad behavior. ++++ ** 4. There was discussion of trying to find some technical help for Paul in the Lab 3 tests, perhaps through Ray Yarema. Marvin had felt that it may not be useful to add a new person at this stage as the learning curve is long. However, those at the meeting thought that with the work remaining before us, and the criticality of the Lab 3 testing, it may in fact be useful. Harry will discuss this by e-mail with Marvin. Next meeting will be Tuesday, April 3 at the usual time (4PM)