CALICE MAPS Meeting, RAL, 03/12/07 ================================== Present: Jamie Ballin, Jamie Crooks, Paul Dauncey, Anne-Marie Magnan, Yoshi Mikami, Owen Miller, Matt Noy, Marcel Stanitzki, Konstantin Stefanov, Renato Turchetta, Mike Tyndel, Nigel Watson By phone: John Wilson Minutes: Paul Minutes of previous meeting: no comments. Sensor basic tests: JamieC reported that he had found that when a monostable fires, it could draw enough current that the neighbours would also fire, which then spreads and causes the whole sensor to light up. He has limited the current drawn by increasing the bias resistors on the IOUTBIAS1 and 2 inputs and this has fixed the problem. This requires a hardware change to the sensor boards but following this, the shaper pixels look OK. However, to get clean behaviour, the samplers had to be disabled. For the samplers, JamieC sees a large oscillation on VRST when it is connected to the DAC, where it is set to 1.0V. The oscillation is much reduced when jumper J24 is changed to connect it to the 1.8V power supply. In addition, he has found the shaper and sampler resets were being released in the wrong order. Paul tried swapping these and the samplers looked more consistent, with no noisy bunch trains. In addition, the shapers now work without having to disable the samplers. The thresholds scan for the samplers shows noise hits out at much larger threshold values than for the shapers. JamieC stated that the gain for the samplers is expected to be approximately twice that of the shapers, so this behaviour is to be expected. JamieC also observed a large common mode on the threshold VTH+ and VTH- lines which should in principle not matter for the threshold but should be investigated. He showed a list of tasks to be done to optimise the sensor before the DESY beam test; see usual web page. It was thought that optimising just on the noise might be dangerous as this could be reduced by killing the gain. The signal/noise ratio would be a more appropriate measure and this would have to be done using the laser as it gives a reproducible signal. We should try to achieve at least the top two sections of JamieC's list before the beam test. JamieC also showed some slides of temperature measurements with a thermal camera. For a quantitative measurement, to remove the effects of the different emissivity of the different materials, photos with and without the sensor powered should be subtracted. Alternatively, all the materials should be painted with black paint (which would not be possible in this case because of the wire bonds). Hence, these pictures are qualitative but do show a very uniform temperature distribution over the sensor. The latest four sensors have been bonded to PCBs and these were given to Matt to implement the latest PCB changes at Imperial. These are sensors #11, 12, 15 and 16 and are all 12mu epitaxial thickness with deep p-well. Information on all the sensors will be kept in an Excel spreadsheet; see the usual web page. The central copy will be kept on the sensor web area. If you need to modify this spreadsheet, contact Paul and then download a copy from the web, change it, and then send it back to him. Laser tests: Marcel reported that he had seen laser hits in the bulk pixels for the first time last week. The hits moved as expected with the laser motion and were clearly correlated with the intensity. One odd effect was that with the laser focussed to 2x2mu2, a line of three or four pixels in x registered hits, although these were all in the same row (i.e the same y value). He also sometimes observed two groups of pixels firing, separated by ~20 pixels. These effects may indicate the mapping of the readout to physical conversion is not yet correct. Following this, a threshold scan was done, with and without the laser; see JamieB's talk below. For H&S reasons, the Fe55 source needs to be kept in Konstantin's laboratory, so a second system will be assembled there for source tests on the test structure pixels. This will then allow the laser to be calibrated using JamieC's simulated gain curves. [Note added after the meeting; the gain curves have been added to the usual web page for this meeting.] The priorities are then for the laser to be calibrated using the Fe55 source, the laser system to be used to step along a row and confirm (or debug) the readout to physical conversion, and for the laser to then be used to scan the critical DAC biases as detailed by JamieC above. This will require two DPW sensors. The Fe55 measurements can use sensor #1, which has a faulty readout column. The other sensor in RAL/PPD is #4, which has no DPW. This should be swapped with #8 which will be potted at Imperial and then brought to RAL within a few days. Marcel also showed some slides of noise measurements on the laser system; see the usual web page. The noise seems to depend on some external influence and is hard to make reproducible. Given the results shown for the test structure, it would also be useful to check if there is an observable rate difference from noise in the bulk pixels with the sensor 5V supply delivered from the USB_DAQ or from an external power supply. Source tests: JamieB showed some slides on the source tests; see the usual web page. It would be useful to make plots in terms of the pixel types, i.e. grouped by quadrants rather than regions, in future. Both the laser and the source measurements show a signal rate in the shapers effectively independent of the threshold over the range scanned. A rough estimate of the size of a MIP near the corner would be 400e- which was thought to give about 120mV output (see the gain curves mentioned above). Hence, the nominal threshold of half this would be at 60mV, which with a threshold LSB of 0.4mV gives the nominal threshold setting of 150. This would mean a full MIP signal in a pixel of 800e- would give a signal of around 600 threshold counts, well above the range currently scanned. For the samplers, it would be a factor of two bigger again. Beam tests: It was thought essential to have an electronic log book and it was decided to use the CALICE one previously used for other beam tests. Anne-Marie will contact Sven Karstensen and Erika Garutti to get this set up. The problem with PMT readout at Birmingham has still not been solved and Matt will take back a spare PMT to Imperial to debug this. Owen showed plots of the data (see URL on usual web page) showing the code is in place to select hits with timestamps close to the PMT coincidence hits. This means that when the PMT problem is solved, he should be able to identify cosmics and beam events quickly. Yoshi has also got some first plots from George Mavromanolakis's semi-online monitoring code and so will continue to work on adding more plots to this. Nigel showed some photos of the Birmingham system, showing the NIM crates modified to hold the USB_DAQs and the power supplies; see the usual web page for the URL. He also showed photos of the mechanical structure, which is being finalised. One issue is with grounding as the whole structure has been anodised, making the surface non-conductive. The structure and the tungsten plates must all be grounded together. The tungsten plates will have to be mounted on top of the aluminium holding plates rather than recessed into them. This will then violate the 2.5mm stay-clear and so require the next sensor PCB to be inserted into the next-but-one slot. As the tungsten plates will only be used for the beam and not the cosmics, this should not be a major issue. Matt reported that we would not need to have potentially four extra power supplies to provide the 2.6V SRAM bias, which might need to be adjusted for each sensor. He has produced some small regulator boards to be added to each sensor PCB, which provide the 2.6V directly. These can be individually trimmed by adjusting their resistor values. The regular boards can be attached to the sensor PCB through one of the mounting screw holes; we should make sure the hole used is not one of the four required by the Birmingham mechanical structure. Matt reported on the firmware status. The most critical item for the beam test is the synchronisation of multiple USB_DAQs. The logic for this has been implemented and tested with all the required functionality. However, the jitter on the distributed clocks is worse than expected and needs to be fixed. Currently, the slave USB_DAQs can lose clock-synch during a bunch train and so disconnect from the USB port. Matt has several potential ideas on how to fix this and he will try these out over the next day or so. He has also cured a problem with the USB_DAQs apparently randomly enumerating as "full speed" rather than as "high speed" USB devices, which can cause problems with the DAQ. The next release of the firmware needs to be coordinated with the software as it will not be compatible with older versions. Matt will also test the performance through a USB hub, which may be an easy way to effectively extend the length of the USB cables from the PC to the sensors. JamieB will update the java versions on the PCs before releasing his latest GUI version. DESY organisation: We will plan for the equipment to arrive late on Sun 9 Dec and so the system can be put together first thing on Mon 10 Dec. A non-CALICE group have the beam from Mon 17 Dec, so we should be able to run from 10-16 Dec inclusive. How we will share the beam with the European DHCAL group is unknown as they have not yet replied to emails, but we should assume we get ~half the beam time. All accomodation for the people on shift should be booked through Susan Kettles via Marcel asap. He is also organising safety training and so the required information should be sent to him straight away. The training will take place on Mon 10 Dec morning. JamieB will miss this and should try to get trained later, although it is not essential. Anne-Marie and Matt will drive all the equipment out to DESY in a hired car next weekend. Matt and JamieB will drive it back at the end of the beam run. The equipment will have to be collected from Birmingham on either Thu 6 Dec afternoon or Fri 7 Dec. It should be packed and ready for collection when they arrive. Yoshi will fly out to DESY on Sun 9 Dec so that he will be the person who knows how to check and assemble the system when it arrives. He was at the previous beam test also and so knows the DESY area. For offices, we can use Susan Kettles's office as well as Zeus and H1 UK offices. Marcel wants to start short (15-30min) daily organisational meetings from Wed 5 Dec until the end of the run. These will be at 10am UK time; he will circulate the phone details later. The first item is to make a prioritised run plan, starting from the list in the minutes of the previous meeting. It is critical that the DAQ PC within the beam area can communicate with the outside world through the DESY network. This requires it to be registered and this must be done beforehand. Marcel will organise this and it requires the MAC address to be sent. He should also register the Imperial PC as it would be useful to have this for analysis outside the beam area, in the control room. Conferences: JamieC's and Marcel's IEEE proceedings contributions have been submitted and are on the MAPS conferences web page. Several results have been released outside the group recently after being distributed by email. While this was understandable in the circumstances, we should try to only show results outside (including collaboration meetings) after they have been discussed within one of the MAPS meetings. This gives the best chance of being sure such results are correct. Oversight Committee: The next OsC document needs to be submitted within the next two weeks. A proposed schedule for the rest of the project was distributed by email following the last meeting and was again discussed. The main changes needed were - Following the FDR, there is 4 weeks until submission. This was thought to be tight, so 6 weeks should be allowed. - The length of time allocationed for the second fabrication is the same as for the first. However, in the next round, the sensors will have to be diced following fabrication and this could take around two weeks. Hence, it is likely the fabrication time should be lengthened by this amount, but Renato will check. [Note added after the meeting; the total time needed is seven weeks.] - The second round of basic tests should continue until the end of 2008, which is when the funding for JamieC's effort will expire. The PDR is currently scheduled for early in the New Year. This is too early for us to know all the necessary results, in particular about whether to make the next sensor with shaper or sampler pixels. This means it is likely the PDR will in fact be split into two parts, as happened for the first round sensor, with the second day in Feb or even Mar. An alternative would be to postpone the outstanding decisions until the IDR in Apr. We also need to show some of the MAPS results in the OsC submission document. The test structure measurements shown at IEEE should be included, together with simulation if available for comparison. Also, JamieB's results on the source should be shown, and any results from the samplers added if available. Some events from the test beam would also be ideal but this will be very tight for time as the document must be submitted by Tue 18 Dec. All figures should be sent to Paul in eps format asap. Next meeting: This was not fixed but is likely to be early in the New Year. An exact date will be set by email.