CALICE MAPS Meeting, RAL, 08/02/08 ================================== Present: Jamie Crooks, Paul Dauncey, Anne-Marie Magnan, Yoshi Mikami, Konstantin Stefanov, Renato Turchetta, Giulio Villani, Nigel Watson Phone: Owen Miller Minutes: Paul Beam test: Due to other commitments, Owen presented his work on sensor efficiency first; see usual web page. The plots show runs both with and without beam; runs 490048, 490063 and 490065 are without beam, while 490061 and 490064 have beam. All these runs have no threshold scan so there is no particular layer defined as the "scanned" sensor. This means Owen's definition of efficiency (number of 4-layer tracks/sum of number of 3- and 4-layer tracks) may not give the layer efficiency due to the four possible combinations of 3-layer track hits; this should be checked. The runs used did have tungsten plates, so the comparison with JamieB's results in the last meeting is not trivial, as he used runs with no converter. With plates, if the electrons start converting, then there could be many tracks in each layer. The analysis only requires at least one in-time hit in each layer and so the apparent efficiency could be higher for showers. Again, this should be checked; this is easiest done by using no-tungsten runs. Sensor simulation: Giulio was also constrained for time and so then showed the results of his completed simulations of the bulk pixels; see documents on the usual web page. The deep p-well is 4-fold symmetric but the n-well circuit is not. This leads to an asymmetry for charge in the four pixel corners of O(1%) for the deep p-well case but O(10%) for the non-deep p-well case. Giulio also finds a 3mu substrate contribution which may balance the 3mu loss of charge due to the deep p-well. For the deep p-well, the lowest charge collected is still in the corner, but for the non-deep p-well, it is the centre of the pixel. The non-deep p-well results shown did not look like those seen with the laser on the test structures, as shown at IEEE. This is assumed to be due to the spread of the laser, which means the results here should be convoluted with a Gaussian of some width. The laser shutter size was 4 or 5 mu for the IEEE data, which did not seem enough; it was thought a spread of 10's of mu would be needed. This sounds unrealistic, but there could be significant contributions from non-spectular reflections, particularly from the metal layers. Giulio will finish the test structure simulations within a couple of weeks. It is hoped that the measurements can be redone by that time as well, through automating the system. Minutes of last meeting: Jamie stated that the problem with the later PCBs was due to the wirebonds hitting the seal ring, not the passivation layer. Otherwise, no comments. Planning: Paul went through the latest ideas for the plans for FY08/09; see slides on usual web page. One issue raised was whether the foundry shuttle run submission dates would fit into the proposed schedule. From Jamie's talk at the last meeting, it was seen that there is a submission date on 14 Jul, which would be around two weeks later than assumed in this plan, but still feasible. We still need to aim for a publication on the results from the first sensor. The PP people thought NIM would be regarded by the PPRP as more prestigious while Renato thought an IEEE paper would be regarded as better by engineers. We should consider this more when we are closer to having the results. It was decided that we would commit to any second sensor being compatible with the existing PCBs, so we should go ahead and make another 30 PCBs and base plates. We should also order more stand-offs and screws if needed. We may need Matt's external SRAM power supply circuits for some of these, but only if used with the first sensor; this problem should be fixed in the second round. The test structure on a second sensor would have to be matched to the pixel type in the bulk close to it, as running the supply and control lines would be too complicated otherwise. On effort, it was thought that there would be enough for testing in the short term (to prepare for the next design) but the period at the end of the year, when the second sensor would have to be tested rapidly to feed back results to the PPRP, would be marginal. In particular, it is now known that both Anne-Marie and JamieB will have moved onto other projects by that time. Sensor design: Jamie showed some slides on aspects of the sensor status; see the usual web page. There are six unbonded PCBs remaining. The plots of response have an x-axis labelled by "numele", which is the number of electrons on a single diode; the total signal is therefore four times this, so e.g. the Fe55 signal of 1600e- would be numele=400 on this scale. The measurements shown on 3/12/07 and discussed further on 18/1/08 no longer agree with the simulation and so Jamie suspects they may have been incorrect, although it will be hard to be sure. The laser seems to be very non-linear below 50% intensity so the statements in the last meeting that the laser is accurately linear are now thought to be incorrect. The difference between the 5mu and 12mu epitaxial layer response was not understood, as the latter seems to be ~1.4 of the former, not the ~2.4 naively expected. There are uncertainties in the actual epitaxial layer thickness from fabrication although the size of these uncertainties was not known. The deep p-well will take up around 3mu of the layer, reducing the response but this should be counteracted by the charge contributed from the substrate, which Giulio (earlier in the meeting) showed was equivalent to around 3mu extra thickness. Beam test (part II): Nigel returned to the beam test, showing some results produced by his project student, Tim Martin; see link to Tim's results from the usual web page. These results are again from runs with the tungsten plates inserted. One comment was that it is likely the beam was lost during run 490060 so this should be checked. Source tests: Paul showed some plots from JamieB and himself from Fe55 source runs at Imperial; see usual web page. They see a clear signal with the source which is absent without, but there is no rate falloff corresponding to the 6keV end-point. This might be due to saturation in the discriminator response and JamieC will investigate this on the test structure. Anne-Marie showed some simulation results for Fe55 which show no 6keV peak after charge spread, but this may be due to using Giulio's charge spread simulation for MIPs, where the charge is deposited throughout the epitaxial layer depth. For the visible Fe55 peak, it is thought the visible peak is due to deposits right under the diode, i.e. in a very limited region in depth. Anne-Marie also showed some checks on Giulio's latest charge spread results, which seem to be close to the symmetry expected if the n-well absorption was symmetrical. This means the full simulation (which assumes this symmetry) should be reasonable. Conferences: The Symposium on Radiation Measurements and Applications in Berkeley (SORMA West) will be on 2-5 Jun 2008. Giulio may attend this independent of any MAPS presentation. No one else expressed an interest to go, so Giulio can give the talk on MAPS. The abstract submission deadline is Fri 15 Feb so he will circulate a draft of this next week. It should be vague as it is not clear what results will be available at that time. In the meeting on 17 Sep, we previously approved Yoshi to give a talk at the TILC08 meeting in Sendai, Japan on 3-6 Mar 2008. He will give a practise talk for this by phone at 10.00 on 22 Feb. Next meeting: At RAL, on Fri 29 Feb at 13.00. In addition, Yoshi will give his TILC08 practise talk by phone on Fri 22 Feb at 10.00.