CALICE MAPS Meeting, Birmingham, 16/07/09 ========================================= Present: Paul Dauncey, Owen Miller, Nigel Watson, John Wilson Phone: Jamie Crooks, Barnaby Levin, Marcel Stanitzki, Renato Turchetta Minutes: Paul Minutes of previous meeting: No corrections. Paul had discussed Diamond with Chris Parkes the day previously. Chris had said things go smoother if the Diamond staff think the sensors may be useful for their beam monitoring. Chris had used a beam focussed to around 4mu and then scanned over their sensor using an xy stage. This is very similar to what we would want to do and Chris said we could use his software to control the stage if we wanted. One thing which was not clear was the energy range. Chris used 8-20keV photons, while we would want 1-2keV. However, this definitely seems a useful facility which we should consider seriously. It would allow us to calibrate the bulk directly rather than using the "fake" Fe55 peak and it would also allow a direct measurement of the charge fraction collected as a function of position within the sensor. It is clear it would not be sensible to try to do a run at Diamond on the same timescale as the beam tests, so sometime around the end of the year would seem an appropriate time to aim for. Sensor testing: Jamie showed some slides of data taken by Barnaby; see the usual web page. The non-zero configuration error level shown by sensors #37, #42 and #45 correspond to one bad column, so this cannot be fixed by changing the external voltage level. Sensors #42 and #44 were on the edge for the standard voltage setting. Sensor #44, which showed errors when trimmed by Paul, appears to have no errors following the modification to the external circuit. Sensor #42 showed few errors initially during the trim but the number of apparent bad columns increased during the three days of the trimming process. This may have been due to temperature, although it is not clear that the temperature varied significantly throughout the period. Paul had considered this was because when loading all trims at zero, as done initially, then a "bit slip" during the configuration load would have no apparent effect. However, as the trim built up more and more non-zero bits, then errors from bit slips would become more observable. Barnaby plans to test the configuration load error dependence on temperature by placing a sensor in the environmental chamber. Paul will retrim sensor #44 when the sensor currently being trimmed (#43) is completed, which should be tomorrow. Barnaby showed slides on data taken during the trim runs by Paul and on new trim runs he had done; see the usual web page. He sees an odd effect when repeating the trim scan, where the sensor apparently gets stuck in a state where hits are only seen at a fixed threshold value. The sensor seems to draw excess current in this state and is only reset by a power cycle. Paul has never seen this behaviour, despite running multiple trims for several weeks. Hence, it is likely this is an effect either of the particular sensor (#47) used, or of something specific to the CALICEDAQ2 system. Barnaby will try to diagnose this further. Marcel showed a plot of the substrate current vs applied voltage on a hi-res sensor; see usual web page. This is expected to be reversed biased but he sees behaviour much more like a forward biased diode. This could be confirmed by seeing if the current has the usual temperature dependence. Meanwhile, he will check the PCB connections are as expected and try reversing the polarity. Marcel is also looking at the Fe55 response as a function of the applied substrate bias, where it would be expected that the fake peak should be sensitive to this voltage. Beam test preparations: Mike had recently announced that we now have confirmed bridging funds for this FY, at the level needed for the beam tests. However, there are no project codes yet so any purchase costs should be buffered for now and claimed later. Matt has produced a version of the master USB_DAQ firmware that is a better starting point for Rui. It is untested and Paul will check this to see what remains to be done. Rui is still unable to build the current version of the firmware so it is hoped Matt's version will help speed up the final development when this becomes possible. Paul has ordered parts for three more sets of the thick cables. These have arrived at Imperial and should be made up within a week. Marcel would like to have another set of cables at RAL to install permanently in the environmental chamber. It is not clear if this requires ordering yet more cable and connectors but the beam test should take priority if there is a shortage. We will also need custom cables for the clock and startBunchTrain distribution and for the PMT inputs; we should probably also make spares for these. To make them, we need to decide on lengths and the signal positions on the connectors, which probably requires the firmware to be completed. Birmingham should get more sensors so it has four usable ones. This will require potting as it would be dangerous to continuously handle them while checking possible geometries in the mechanical stack without protection. Jamie will find out of the potting can be done at RAL. If so, he will get the four "perfect" trimmed sensors potted and then sent to Birmingham. Ideally the potting would be around the bonds only, so the sensor centre would still be usable for source studies. However, given that we have a reasonable number of good sensors, it would not matter if these four are potted with the sensor completely covered. If RAL cannot pot them within a short time, then they need to be brought to Imperial, who did the previous ones. For mechanics, Nigel thinks the best approach would be to have the PMTs mounted outside the main body of the mechanical structure, so as to leave the maximum amount of space inside. This would allow six sensors and possibly some converter to be inserted. One downside is that it would not be trivial to lie the stack on its back with a PMT mounted there, so the rear PMT might need to be removed whenever the stack is laid down. The PMT alignment to the sensors is not critical at the sub-mm level. Nigel will need to know if any temperature sensitivity of the sensors is found as this could set requirements on the number of fans needed for cooling. The PMT readout PCB has been returned to Birmingham and has been populated. Xen only had a quick look before he left for CERN but found it mainly worked, although there is some noise in the LVDS section which he will try to reduce when he is back next week. Nigel and Paul will discuss the network switch/wireless access with the Birmingham system manager today so as to order a sensible device soon. Paul would like to test all three "non-standard" sensors (i.e. 12mu hi-res, 18mu hi-res and non-deep p-well) in the beam, so at least one of each of these, and possibly two of the 12mu hi-res, should be glued centrally and then trimmed. Paul will leave instructions on the trim procedure if these are not completed before he leaves on holiday on 1 Aug. For disk space, Marcel has several 500GByte disks which can be mounted in the PCs and this should be done asap. Bristol have a large disk array which could be plugged directly into the local network. It is not clear if they aim to use this for the Fortis data; if not, then we could use it. Another option would be to buy an external 1.5TByte disk. Whatever is done, the disk has to be directly visible from the PC used to run the semi-online monitor. The data also need to be shipped out of CERN and more than one copy should be made. There is a large (20TByte) disk array at RAL which can be used for the primary storage although both Birmingham and Imperial will take local copies also. Nigel suggested storing the files on the Grid too. Paul showed some slides on the complications of the efficiency measurement; see the usual web page. A rough division of the tasks listed on the last slide was done, with Imperial taking on the alignment, RAL the track pattern recognition, Imperial the track fit, and Birmingham the extraction of the efficiency. Prior to the meeting, Paul had distributed a first MC file in the right format which should allow analysis code to be developed in advance. The MC file should also be used to develop the semi-online monitoring. This should contain plots which demonstrate a beam signal but does not need to do a full analysis. Specifically, the position coordinate correlation plots of different sensors, and the time correlation plots of the sensor and PMT hit timestamps, are probably very useful. DESY beam test: Given the funding is now confirmed, we need to fix dates for the DESY beam test. It was agreed that some weeks would be needed after CERN to understand the data enough before the second beam test. This would allow the DESY test to fill in any gaps in the data and also would allow us to be better prepared as the MIP efficiency should be known. Given also the Albuquerque ALCPG meeting in late Sep, this effectively means that the DESY beam test would have to be around the middle of Oct at the earliest. This is unfortunately during term time but it would not be good to wait until the Christmas break and it is not clear people would be more available during that vacation either. Paul has been told we can apply for EUDET travel funds for the DESY beam test, even though we are not using the EUDET telescope, and he will look into applying for money. He will also try to book a specific slot of two weeks at DESY for mid Oct. Next meeting: Paul is on holiday from Aug 1, which is only two weeks away. There are also weekly test beam phone meetings until the test in Aug. Hence, it was decided to not set a date for the next TPAC meeting but to stay in regular contact to ensure the work converges. We will formally meet again at some point after the beam test, i.e. in Sep.