123456789012345678901234567890123456789012345678901234567890123456789012 CALICE MAPS Meeting, RAL, 01/03/10 ================================== Present: Jamie Crooks, Paul Dauncey, Jan Strube, Renato Turchetta, Mike Tyndel, Nigel Watson, Gary Zhang Phone: John Wilson, Tony Price Minutes: Paul Minutes of last meeting: No changes. Matters arising: o Paul started trim tests but the system was unstable so these are mainly still to be done. o The EUDET interface firmware is not yet done. Rui will be coming over to RAL this afternoon to give his changes a first test. o Nigel said there has been no progress on the LCIO work for the EUDET telescope alignment, track reconstruction and analysis. o Paul contacted Felix Sefkow and we have at least two named people who can help us out; they are Mark Terwort and Clemens Guenter. o To avoid confusion, Paul went through Jaap to get an xy stage. Jaap says a 20kg stage may be hard, while 5kg stages are quite easy to borrow. It will be difficult if we cannot move the stack easily so this needs to be chased up. o Sensor #50, which was half dead, has been replaced and the new sensor trimmed. o Marcel has not yet set up the paper svn repositories. DESY beam test: Paul went through a long to-do list: Jamie has checked the records for all trimmed and working TPAC1.2 sensors. Sensor #46 died from overheating before CERN. Sensor #37 is a hi-res 12mu but needs to be trimmed; this will be done over the next few days. This means the total sensor list for DESY is Standard = 8 total: #29, #32, #39, #41, #43, #44, #47, #48 Non-DPW = 2 total: #38, #38 Hi-res 12mu = 3 total: #21, #37, #42 Hi-res 18mu = 2 total: #26, #50 Of these, five are unpotted. Jamie ordered four watchglasses which should arrive today. These are plastic and are 3.8mm high. He will check if there is adequate clearance from the wire bonds if using these. If so, he will order more; if not then the fallback is to use the black plastic covers, as before. Either is likely to be higher from the PCB surface than the potted sensors, but mostly the critical spacing is for the standard sensors which are the first or last two. All sensors should be run in the system before leaving RAL so that we can check for configuration errors, bad pixels, etc. The equipment will be packed into the van on Mon 8 March. Nigel will then return to Birmingham and then drive out to DESY starting the next day, arriving on Wed 10 March. The CAEN cable boxes are effectively ready, although only one large SCSI-type cable is available at present. Gary does not know how to power up the CAEN so this will require Marcel. This should be tried this afternoon. It is not clear whether our scintillators should be included in the EUDET telescope trigger. This depends on the limiting factors for the telescope and Fortis readout, which are not yet known. If they are to be used, then the LVDS outputs from Xen's board will be used for this. The TTL outputs will be used for the TPAC system time tags in either case, exactly as done for CERN. Nigel brought down the Xen's older PMT control board today. There are two PMTs with control boxes at RAL already; the third set is still at Birmingham for testing the new version of Xen's board. This new board has different components to make it more stable but is otherwise very similar. The two PMTs should be installed and included in the readout this afternoon. There should be a Confluence area for the DESY beam test, similar to that for CERN. Jan will set this up today. Nigel brought down a black cloth which is 3x3m2; this should be easily big enough as Paul estimated 2x2m2 would be the minimum. Paul has got a 50x50x50cm3 frame being built in the Imperial workshop; unfortunately it was not ready on Friday for him to bring today. He will get this to RAL by Monday next week so it can be packed with the other equipment. The black cloth will prevent airflow and there are suggestions that the sensors getting hot causes configuration errors. Hence, it would be good to have a pipe with a fan forcing air into (or out of) the covered stack. Nigel will look into this in the afternoon. Jan has not yet started on the monitoring changes as discussed in the last meeting but he will get going on these today. As stated above, Rui will come to RAL to test his EUDET interface firmware this afternoon. The 2TByte disk for calicedaq4 is not yet installed. The idea is that this should be inserted and used as the main data-taking disk. It will then need to be nfs mounted on tpacmon so the monitoring can see the data files. In emergencies, calicedaq4 can then be used as a standalone DAQ system as it requires nothing else. There is a 1TByte disk in calicedaq3; this PC will not be normally used but will be a spare to take over from calicedaq4 in case of problems. In the system at present, there are just exactly the required number of USB_DAQs, one master and six slaves, so there are zero spares. We need to get more. Rui has one for his firmware development which can be used as the master spare. There is one in the trimming system (in the clean room lab) and one being used for the Fe55 measurements, both of which could also be taken out as slave spares. However, there should be 12 total, so this implies there are two unaccounted for. Twenty brand-new USB_DAQs were made around a year ago, to use up funds at the end of FY08/09. Paul brought three of these to RAL today; they have not been tested in any way and should be powered up and programmed with the slave firmware so they can be tested. If they work, then they can act as spares. Note, these to not have adapter boards so only act as the base board spares. Paul asked about how the substrate should be connected for the hi-res sensors. The options are to ground the substrate (in principle to either the analogue or digital ground, although the former is the more obvious), allow the substrate to float, or to apply a bias voltage, again relative to the analogue ground. Mike had found a few plots from Marcel which seemed to contradict Renato's expectations of a resistive connection between the deep p-well and the substrate with a resistance of around 20ohms. The exact method to make these plots should be clarified with Marcel. Previous runs have all been with the substrate connected to the analogue ground and this should be the default. It would be worth checking the pedestal and noise dependence on the bias voltage, in case the sensors cease to function at some bias. As Paul and Jamie will be trimming sensor #37 over the next few days, it would be easy to do some more sets of runs with various bias voltages, as this would give a direct comparison of these quantities. Each run takes six hours, so this would still allow the trimming to be completed before next Monday. Gary will also try to measure the relative rate of Fe55 hits in the full-energy peak (around 200mV) as a function of the bias voltage. He should use sensor #42 for this. Gary has been tracking down the instabilities in the system over the last week. Using the scope, he found no missing triggers on the cable. He also found that a USB_DAQ which gave poll timeouts (and hence C++ exceptions) did not recover for the next bunch train. However, by removing the trigger cable briefly, Gary found that boards which did genuinely miss the trigger did recover. Hence, he concluded the error was not due to a bad cable but more likely due to grounding. He found attaching scope leads made the system more stable. On Thursday, he made up a set of lemo-ended ground leads which are attached together and to the crate at one end and plugged into the brass PMT lemo inputs (as a good ground point) at the other end. The system has not crashed at all in the intervening four days. He will also add filters to all the power leads today. Paul is still seeing some level of configuration errors, O(100) pixels for one or two sensors. He also sees some rate of hit corruption; this is identifiable as the group number (0-6) is stored in three bits so there is one invalid group = 7. This occurs at a lowish rate (~one in a thousand bunch trains) but the associated hit data do not look sensible so the whole word may be spurious. This also occured at CERN and these hits were ignored, so they are not critical but it is annoying. The frame for the black cloth will be metal and so might disturb the grounding, although it will not be in direct contact with the system. It is likely to be anodised and so it may not be easy to ground. Paul will check on this with the workshop. Nigel has organised the van hire but only he and John will be insured to drive it. Nigel has had one tungsten holder plate modified so it can be inserted without removing the long threaded screws in the stack. This should be tested in the stack this afternoon and the rest then machined in the same way. Tony showed some slides on simulation results for the beam test; see slides on the usual web page. He has generated all combinations of 1-6GeV beam energy in steps of 1GeV times 1-15X0 of tungsten in steps of 1X0; a total of 6x15=90 runs, each with 10k events. These will need to be digitised, although this step is very much slower; currently Paul reported it takes 20sec per event. Tony's slides 6-11 are for 1-6GeV respectively, and slide 12 is for hits in layer 3 with 6GeV and 5X0. One important conclusion is that many of the low energy beam particles will scatter in the ~10m of air before they reach the stack, so we may only see ~10% of the beam particles in the sensors. This may mean we need to run for much longer at low energies to get the same size data samples, although this clearly also depends on the relative beam intensities too. CERN beam test: SPIDER has been assigned two separate weeks in the SPS schedule for 2010; Tue 25 May - Mon 31 May and Mon 20 Sep - Mon 27 Sep. We may not take these up but they would clearly not be very convenient if we did. Papers: Jamie reported that there are valid data for the TPAC1.0 paper from his laser scans of the test pixels. However, these are on a PC which he does not have access to so he will need help from Marcel. He should be able to get the raw data to allow some 2D plots of the response vs x and y to be made. Unfortunately, he thinks the 5mu vs 12mu data are not usable. Nigel has not yet got the digitisation code from Anne-Marie. He would like to try to get some results on PFA using Pandora and has two final year project students working on this. It may be that it would be better to produce two papers on this (on the EM and PFA resolutions separately). This would also mean the basic EM results would not need to wait on the more complicated PFA studies. Mike went through some of Jamie's timing results using the laser on the test pixel. It is not clear whether these results are easily interpreted and more data might be needed. It will not be possible for Jamie to work on this until after the Cherwell chip is submitted. Next meeting: This will be arranged after the DESY beam test.