CALICE MAPS Meeting, RAL, 07/11/07 ================================== Morning session =============== Present: Jamie Ballin, Paul Dauncey, Anne-Marie Magnan, Nigel Watson Minutes: Paul Simulation: Anne-Marie showed the status of the GEANT4 simulation for the source tests; see slides on the usual web page. This is not based on Mokka but is "inspired" by it. It would be easy to adapt it to the beam test also. She sees a significant shift in the MPV for the source energy deposits compared to MIPs. However, the MIP shape is different from the shape seen in the standard ECAL simulation and this needs to be understood. The GEANT4 versions used are the same in both cases. The spatial distributions also show the source particle generation will need some adjustment. The momentum direction is not done to be equal in solid angle and the source size is set to 1cm while the real source is a few mm. The simulation currently is done with 50x50mu2 pixels and this will need to be reduced to 5x5mu2. Data analysis: There was a long discussion on the format to be used for analysing data. This should be common to real data and MC. The alternatives are to convert the MC to the raw binary data format or the data to LCIO and the latter was thought best. The second issue is that the data come in bunch trains while LCIO is structured by LCEvents. The bunch train could be broken down into time slices around the PMT trigger tags or an LCEvent could include all the data from the bunch train. The former gives problems with noise runs where there are no external time tags. It would also not allow complete studies of efficiency loss due to filling memories earlier in the bunch train. Hence, it was decided an LCEvent should contain a full bunch train. The real data should then just be converted from the raw format to some LCIO LCIntVec list of hits labelled by x,y,z and t. The simulation is more complicated. The simulated events will need to be superimposed onto the bunch train. The number per bunch train will be Poission distributed and then their timestamps within the bunch train should be uniformly distributed. The rate at DESY is roughly ~500Hz so with a 2ms bunch train, the average number will be ~1 event per bunch train. The events to be overlaid will need to be Mokka-level output without digitisation. The digitisation including the time dependence will probably be different enough that it will have to be re-implemented specifically for this application. It needs to generate noise hits throughout the bunch train, add the physics hits and include the monostable lengths that cause the hits to be on for more than one timestamp. The SimCalorimeterHits and other truth information from each Mokka event can be stored in separate collections to distinguish them, but the data hits will in principle not be able to be separated unambiguously. Anne-Marie volunteered to start setting up the raw->LCIO converter and then will look into reimplementing the digitisation. Particle flow: JamieB showed some slides on his work on MIP-finding and the "no-harm" Pandora studies; see usual web page. He sees a significant improvement in the ECAL resolution by tuning the MIP-finder algorithm and it seems better than the standard ECAL value. Using the improved MIP-finder to feed into the no-harm studies, he now sees very similar resolutions for the Pandora output on hadronic events for both the standard and the MAPS ECALs. Afternoon session ================= Present: Jamie Ballin, Jamie Crooks, Paul Dauncey, Anne-Marie Magnan, Renato Turchetta, Mike Tyndel, Nigel Watson Minutes: Paul Minutes of the previous meeting: No comments or changes. Sensor tests: JamieC showed some plots of the test structure both with and without the deep p-well; see the usual web page. These were shown at the IEEE conference last week. The upper two diodes of the lower pixel (left plot) would be expected to be similar to the lower two diodes of the middle pixel (right plot) but these looks different. This may be due to the differences of the surrounding area. Giulio is doing a detailed simulation of the whole three test structure pixels to be able to compare in detail. JamieC will continue to work on the test structures to understand the loss of gain and where we are operating. He can get support from the Imperial group for quantitative measurements of the bulk pixels if this would be useful but they are not set up for studying the test structures. Paul showed some slides of the status of the basic and source tests at Imperial. There is a lot which is not understood but a signal (albeit with low efficiency) has been seen from the source. One issue is the comparison of the overall and trim thresholds, which does not show any sensible response yet. JamieC stated that the LSB for the general threshold should be equivalent to 0.4mV while for the trim threshold, it should be 2mV, giving a factor of 5 difference. Konstantin suggested masking off all pixels but one so there is no issue with filling the memories. JamieC also suggested checking the baseplate is adequately grounded as he has seen it changing potential significantly. To run the bunch train for some time before taking data, to remove the startup noise, JamieC suggested setting the MUX address to be fixed to the "park" value as this will prevent the memories filling. Paul reported that Matt has taken all five new sensors to Imperial for tests. (JamieC thought he might have also taken the non-deep p-well one which JamieC wanted to use for tests; this should be sensor #1 or #3 and Paul will check when back at Imperial.) JamieC had modified most of the PCBs before leaving for IEEE and Matt finished off the remaining ones. There are also now all ten USB_DAQs and adapter boards at Imperial although one may have a fault. Nigel reported that Birmingham have two PMTs with 1cm scintillator squares on them which they will use for the cosmics. They would need to use a light guide for the beam test to keep the PMTs clear of the beam. They will buy some smaller PMTs specifically for this. Paul handed over sensor #6 to Nigel for use at Birmingham. We should give the dimensions (in particular, the required height from the base stand to the beam line) and the weight of the beam test stack to Erika asap so that she can find us a support table. Schedule: We have been asked by the OsC to draw up a new realistic schedule for a plan-to-completion of the grant work. We can use the unspent Working Allowance to extend the project beyond the end of the official grant end of Mar09. This would ideally cover the MAPS in the EUDET beam test which will be in Jun09 or Sep09. Paul showed a draft of the modified Gantt chart for the rest of the project. The main comment was that the four months of redesign time for the second round was far too short and this should be more like eight months. This would be possible without JamieC's time falling outside of FY08/09, which is a constraint. It was also thought that a second round PDR would be required to decide what the sensor would be (shapers/samplers, etc). This would be in addition to the IDR and FDR. Hence, the schedule would be along the lines of: o Design Dec07-Jul08 inclusive o PDR early Jan08 o IDR early Apr08 o FDR late Jun08 The sensor fabrication should take the same time as previously and the rest of the tasks should be pushed back accordingly. Paul will ask Mike Green draw up a new Gantt chart with these changes to see how it looks. Next meeting: This will be on Mon 19 Nov starting at 13.30 (note the slightly later time that usual). Because of availability and the need to concentrate on the DESY beam test preparations, there will not be a software meeting in the morning. Renato booked the R76 meeting room.