AFE Meeting April 5, 2001 Harry Weerts, John Womersley, Fred Borcherding, George Ginther, Jianming Qian, Thomas Nunnemann, Levan Babukhadia, Manuel Martin, Mrinmoy Bhattacharjee, Paul Grannis The main purpose of the meeting was to consider a decision to buy extra AFE8 boards to allow full CFT (and thus CPS) readout. News items at the beginning of the meeting: 1. Board 4 has now been readout at Lab3 with a 50 MHz clock (25 MHz SVX readout, though has been read out at 40 MHz at DAB3). 2. The code for temperature control has been installed on Board 4; debug is underway. 3. Tests by P. Rubinov and M. Utes adjusted the maximum count that the the SVX could read out. Such tests were performed for SMT at SiDet earlier. For example, when the maximum was set at 45 counts, many results of 46 counts were found despite the presence of an input signal. This finding further suggests the differential nonlinearity problem is in the SVX. 4. Board 1 was broken at Lab3; is now back at DAB for fixing. 5. Attention at Lab3 will turn to tests of the discriminators with two board operation. During Paul Rubinov's absence this weekend, cassette test and certification will proceed. 6. Thomas Nunnemann has worked on fits to the large gain run from last week to see if adequate fits to photopeaks could be obtained. He found that some adjustments to account for the DNL ripple had to be made, but fits were sensible and consistent with Poisson distribution of pe's. 7. Jerry Blazey talked with Sanmina today; Sanmina understands not to proceed with assembly until a signoff from DZero. They need final certification of production parts, and an engineering OK. We believe these can be made shortly (pending approval based on our Lab3 tests). Sanmina expects to start stuffing on May 7, with a 5 day turn around. Sanmina confirmed that the price for extra boards would be the same as the first order -- ~$1300/board. However we are puzzled, since the per board cost of the 146 board order is less than that! 8. John Anderson has said that we are ready to go to an alternate vendor, and that we may expect better prices than Sanmina. Discussion then turned to a decision to buy extra AFE8 boards to outfit the CFT stereo and thus all of CPS axial and stereo. Slides shown on this issue can be found at the main AFE discussion page d0server1.fnal.gov/users/grannis/afe/afe.html The suggestion was made that current optimistic estimates for AFE12 completion put it at the end of 2001; this is much later than DZero wants to have a complete CFT tracker. The CFT of course can operate with AFE8, so the issue is whether reasonable CPS operation can be retained with AFE8 single threshold, single digitization channels. The desire for dual range CPS stems from the need for large dynamic range to avoid saturation on showers but allow calibratration with MIPs. Dual range triggering is desired to allow J/Psi and trilepton triggers in addition to W/Z triggers. The standard AFE8 configuration causes saturation of the SVX at roughly 20 MIPs on the preshowers, much below the desired 60 MIPs. However, it is possible to reduce the input charge in two ways -- by reducing bias voltage to the VLPC (can get perhaps a 50% reduction this way) and by reducing the 100 pF input capacitor to the SIFT. Previous tests have shown that reduction of this cap to 10 pF works fine; this reduction would give a factor of 4 reduction in input charge. With the reduced charge input, a solution for preshower operation can be found that gives approximately 60 MIP dynamic range. A single MIP gives between 4 and 8 counts on the ADC for standard SIFT gain. While this may be too small for reliable calibration, special runs (thought to be about an hour of data) with large SIFT gain would give adequate MIP signals for operation. We imagined that the CPS single threshold trigger would be operated at a threshold appropriate for J/Psi or trilepton triggering. This would give a somewhat smaller rejection factor for W/Z but here the calorimeter (with some preshower/tracker information) should be able to handle the rates at all but the very highest luminosity. The FPS operation with single channel AFE boards would be comparable to the CPS, so it is judged feasible to go to single channel boards. However, the FPS cassettes are constructed with crossover channels of shower layer strips from one module of VLPC's to different MCM modules. The sorting of VLPC batches is thus not fixed to module boundaries. Special provision for bias and threshold settings for FPS channels was to be made in the AFE12, and if we are to retain good FPS performance, some modification would be necessary to the CFT/CPS AFE8 board. Several variants of these can be imagined -- from full functionality of all channels in AFE12 boards, full functionality of all channels in AFE8 boards and one threshold, to solutions dropping some of the outer FPS strips from the readout, to compromise solutions where some strips may not have sensible efficiency. The group made a clear recommendation to the spokesmen that the experiment proceed to buy extra AFE8 boards to service the full CFT and CPS. We would expect to operate these boards for RunIIa (not replace with AFE12 even if available). The exact number of extra boards remains to be settled, but with 146 boards produced by Sanmina, and a need for 166 total, and spares at approximately 20%, the number of extra boards needed is about 50. We are open to adding the extras to the Sanmina order, but do not really imagine this will be the best solution, as we do not want to slow the first order up. Procurement of the extra needed parts must now start, and may not be done in time for the Sanmina assembly. Thus the effort to identify alternate bare board and stuffing vendors must begin now. We must settle on the number of extra AFE8 boards to buy. Work on the AFE12 boards should stop for now while the AFE8 tests and new procurement plans are done. We need to study the appropriate solution for the FPS -- AFE8prime or AFE12 in a timely way so as not to delay FPS unduly. We need confirmation that the AFE8 solution for CPS outlined in the meeting is truly feasible (dynamic range, calibration etc.) Jianming Qian agreed to do this. We need to decide what input cap to the SIFT is appropriate; this requires measurements of the SIFT to SVX gain on the prototype boards at Lab3.