CALICE MAPS Meeting, RAL, 17/05/06 ================================== Present: Jamie Crooks, Paul Dauncey, Yoshi Mikami, Mike Tyndel, Nigel Watson, John Wilson Minutes: Paul Minutes of last meeting and of PDR: No comments. Preliminary Design Review: There was a brief discussion of the outcome of the PDR which took place on 05/05/06. It was concluded that the design schedule is mainly paced by the wafer run dates. These are not yet defined but will be similar to this year, when they were Jan 15 and Feb 14. These are the dates to send the tape to the foundry. Mike suggested setting an internal date for our own use; this will then avoid problems with Christmas and allow planning for the second Interim Design Review which should actually approve the submission. A date of Dec 1 was suggested although we should iterate on this (but then fix it) at the next meeting. Renato has some detailed on costs for Foundry B and these may restrict the technical implementation, in particular stitching. Stitching will reduce the yield and this may be the major cost driver for the final system. Hence, it is likely we will not use stitching in this project. The important thing will be to work with a definite cost for the two fabrication runs. This should be discussed further when Renato is present. There is an issue of large energy deposits in one pixel. In Atlas, these arise from neutrons knocked out by the hadronic showers. These thermalise and are then captured by boron, which has a huge neutron reaction cross section, giving out slow, highly ionising nuclear fragments. The numbers at LHC are around 10^13 neutrons/cm^2/year, meaning 10^6 neutrons/cm^2/sec. As these arise from physics jets, it is hard to see than the rate would be anywhere near this high in an ILC detector; the physics rate is over a million times lower. However, there will be thermal neutrons from machine backgrounds which will dominate over physics, but the rate is not known. There could also be effects from low energy photons depositing all their energy in or near the epitaxial layer. Even 100keV photons would give signals more than an order of magnitude higher than MIPs. Jamie can look at how long a pixel would be in saturation for large signals and Yoshi can study the rate of at least the photon component in the simulation. The neutron contribution may be harder to estimate. The PDR final report and minutes should in principle be send to the external referees for approval; Paul will forward them on. Paul will also sign off the Project Specification, Technical Specification and Testing Specification after the meeting. The Project Management Plan is internal to RAL Technology and will be signed off by Renato; this still needs to be updated. Sensor design: Jamie showed two sets of slides (see usual web page). The first described the SRAM column idea which he and Renato had discussed before the PDR. Here, the memory and control is concentrated into three dead pixels which control 48, giving 6% dead space. The control takes one of the three pixels while the memory occupies the other two. This should make much more efficient use of memory as the statistical fluctations over 48 pixels is proportionally smaller than for one pixel. Also, SRAM can be used; the physical limit seems to be around 20 SRAM locations in two dead pixels. It would be useful to know the distribution of the numbers of pixels hit in physics events in lines of 48 contiguous pixels and this is something which the simulation can now start to be used for; Yoshi will look into this. In addition, the Poisson fluctuations on the noise will probably determine the total number of storage locations needed; Paul will try to look into this. As Jamie is mainly concentrating on the analogue pixel circuit first, a decision on the memory and control is not critical now. Jamie's second set of slides reported on various other items which arose since the PDR. He has talked to Vicky Davies (in Paul Sellars' gold stud bonding facility) and it is likely normal wirebond pads could be used, which can therefore allow stud bonding with effectively no extra work from us. However, the actual assembly would have to be hand aligned and could take a significant amount of effort. He also reported on analogue summing of diodes; this seems feasible but may need careful setting of the working point. Single large diodes are not feasible as, for a fixed signal charge, the signal voltage seen is inversely proportional to the capacitance; this requires the diodes to have the minimal 0.8x0.8 mu^2 size. Even then, the voltage change is only around 10mV. This week, Renato is attending a conference which will include a report on a purely NMOS amplifier, which might help improve the signal size. However, as it is not known which noise source will dominate, this may also simply amplify the noise by the same factor, giving no gain. Sensor simulation: Giulio was not present but had reported to Paul that the disks for the simulation farm had now been installed and the simulation seems to run. He thinks the swap space should be adjusted; he should contact the IT person in RAL Technology and the fault should be logged in the requests system in the proper way. As decided previously, Giulio should cc Renato on all emails to IT Support. Giulio had also reported that he could start on the mesh simulations for Jamie soon. Jamie thinks he should include circuit n-wells from now on as these could have a large impact. The optimal geometry of the circuit n-wells and charge collecting diodes is not obvious and several suggestions were made; the simulation should be able to start looking at these ideas. Physics simulation: Yoshi showed some slides on the GEANT4 simulation (see the usual web page). He has varied the pixel size between 25mu and 100mu to study changes. The results he sees are hard to interpret and will need further study. For electron showers, he seen no sign of a plateau at smaller pixel sizes in the number of hits but does see the average energy deposited per hit is independent of energy. For muons, he sees the number of hits is roughly twice the number of layers, for both the original (500mu) and MAPS (15mu) case. The hit energy spectrum shows a very clear MIP peak for the 500mu case but no peak is seen (at 3% of the energy) for the 15mu case. The implication is that even muons are accompanied by a lot of soft hits and these may swamp the MIP signal in the MAPS epitaxial layer. This clearly needs to be clarified and using the event display may help. Next meeting: 1pm on Tue 6 Jun, in R76 as usual.