CALICE MAPS Meeting, RAL, 16/07/08 ================================== Present: Jamie Crooks, Paul Dauncey, Owen Miller, Andrei Nomerotski, Marcel Stanitzki, Jan Strube, Renato Turchetta, Mike Tyndel, Nigel Watson Minutes: Paul Marcel introduced Jan Strube, who has just started working in RAL/PPD and will be half time on ILC. Minutes of the last meeting: Jamie pointed out that the foundry test device does not contain every possible component, but just a wide variety of them. Paul also said there was a typo in the splits; the last in the list should be non-deep p-well. Sensor 1.1 design: Because the hi-res wafers will not be available in time for the shuttle run, then the fabrication will be done in two parts. Three of our six "standard" wafers and all the non-hi-res splits will be done first. The other three standard wafers and all the hi-res splits will be later. Doing the standard wafers in two parts allows us to check there are no differences in these processings. The first set of splits should be returned ~eight weeks following the shuttle deadline (July 21), which means towards the end of Sept. The timing for the hi-res splits is less well defined but will be between two and six weeks following this. Andrei asked about the hi-res wafer resistivity; Renato said it would be somewhere between 1-10 kOhm cm. Jamie said he still has a few minor design issues. There is a conflict with the deep p-well and a resisitor in the pad protection circuit. This may just mean the resistor value will be less certain (which is not a problem) but this needs to be checked. The RC extraction to simulate the circuit also gives minor errors but these could be left until after submission. He hopes to submit the design within a day if the final checks do not show any further issues. He is anyway very confident of making the deadline of July 21. Conferences: Marcel reported we have been granted a talk at TWEPP (Naxos, Sep 15-19) of 15 mins. We also have a poster at IEEE (Dresden, Oct 19-25). As TWEPP is more focussed on electronics, then it was decided Jamie should do this talk. This fits in well with the expected fabrication return date. Pixel 2008 (FNAL, Sep 23-26) and LCWS08 (Chicago, Nov 16-20) should be added to the conference list; the websites are linked from the usual meetings web page. Paul is a convenor for the calorimeter sessions of the latter. Talks from Giulio (TWEPP in 2007) and JohnW (ECFA) are missing from the web and should be added. Sensor 1.0 analysis: Paul showed some plots of Fe55 runs; see directory linked from the usual web page. Here, runs 476519 and 476520 are with and without the source respectively and have a threshold scan range of 200-399. Runs 476691 and 476690 are the same but over a range 300-499. Plots labelled "raw" hits are as a function of the nominal threshold and "real" hits are as a function of the threshold corrected for the individual pixel pedestal. The distribution of pedestals is shown in the MeanQuad0 plot. All plots are for quadrant 0 but are similar for quadrant 1. The Fe55 has a gamma energy of ~6keV which should give 1600e- in the silicon. When right next to the diode, all this charge should be seen. If the observed noise of 5TU is the best possible from the circuit simulation of 20e-, then 1600e- would correspond to 400TU. If the 5TU of noise corresponds to a worse figure in e-, then the Fe55 edge would be at a smaller threshold. Hence, the above plots should contain the edge unless it is below 200TU, as which point it may be hidden in the noise. The critical issue is that no falloff at the Fe55 endpoint is seen; the distribution smoothly falls off throughout the regions scanned. The quadrant 0 gains have an RMS of ~12% so e.g. a drop around 300TU would be smeared out by ~40TU, but this should still be visible. Without finding the endpoint, these data cannot be used for calibration. The RAL source is orders of magnitude more powerful. It may allow a calibration from a single pixel, so the pixel gain variation would be irrelevant. Marcel will set this up with a trimmed sensor and check the Fe55 availability. Non-source noise runs are needed and could be done before the source is available. Paul should run his mask threshold scan to find the pedestals as these are needed for the best precision and it would be useful to compare with Owen's results. The most reliable method to get the absolute calibration still seems to be using the Fe55 peak seen in the test pixel to calibrate the laser in the same system, and then using the laser to measure the signal in the bulk pixels. The sensor used for the previous Fe55 measurements is no longer available so the source measurements probably need to be redone first. The laser will give a different signal in the test pixel depending on its position. Hence, a rough position scan will be needed to centre it on the pixel first. The simulation results should then be used to correct for the charge loss, although this will in principle cancel out in the bulk pixels if the laser is again centred on the pixels there. Around 100 bulk pixels (from both quadrants 0 and 1) need to be calibrated with the laser to get a good idea of the pixel spread. Jamie said he would try to do this when the submission is complete although this should be automated, as discussed in previous meetings. The DAQ<->laser PC communication was debugged a year ago so the issues with stage control are purely internal to the laser PC LabView program. As discussed previously, the test structure laser scan shown at IEEE will need to be redone. Preferably, this would be automated by reading out the scope via its GPIB port. A full scan and a quantitative comparison to the sensor simulation results is needed. Another beam test for sensor 1.0 is probably not sensible at this point but doing one for sensor 1.1 early in the New Year seems a good plan. Sensor 1.0 papers: Renato submitted the INMAPS paper to Sensors. Paul will set up a web page to store this and future papers. Paul went through the outline of the bigger paper(s); see the draft on the usual web site. The critical issues are to ensure all the analysis work required is covered, as most plots will need to be redone. At a general level, figures can include colour but we should work with a common ROOT style file; Nigel will get the CALICE one and send it to Marcel for any modifications. A discussion on the outline produced the following: o The paper will focus on shapers, only showing sampler results if easy or if needed (e.g. the test structure). o Sec 4.1 should have a plot showing the expected gain from the circuit simulation o The effect of the reset is not so interesting for this paper. o The trim bit sizes for each pixel should be shown. o The gain variation for several sensors should be shown, but a complete gain map of one sensor is not worth the time to do. o The charge spread data need to be done more carefully with per pixel gain and better alignment. o A bad pixel count per sensor would be useful. o It is not clear what we can say about twinning as the effect is not yet understood. The hitOverride results should be included. o The Sr90 beta source is the only reliable handle we have on 5mu/12mu epitaxial layer differences and so should be included. o For the beam test, only positions correlations/residuals should be shown, not absolute efficiencies. o The no-harm work from JamieB should be included if it was done in a consistent way to the ECAL plots in Sec 6.2 and 6.3. The other PFA work was not done in a consistent manner and so will not be included. The following people (were, in their absence) committed to providing the plots (and later text) for the following sections: Sec 2.1: Anne-Marie Sec 2.2: Paul and Nigel Sec 2.3: Paul Sec 3.1-3.8: Jamie and Renato Sec 3.9: Paul Sec 4.1: Jamie Sec 4.2: Anne-Marie Sec 4.3: Jamie and Marcel Sec 4.4: Paul and Owen Sec 4.5: Paul Sec 5.1: Jamie and Marcel Sec 5.2-5.3: Paul Sec 5.4: Paul (Sr) and Marcel (Fe) Sec 5.5: Paul and JamieB Sec 5.6: Nigel Sec 6.1: Nigel Sec 6.2-6.3: Anne-Marie Sec 6.4: Paul and JamieB These people should have a plan to complete these sections by the next meeting. The paper must be in good shape to go with the proposal to the PPRP in Sept. Next meeting: This will be on Tue 29 July at 1pm.