On June 3rd Dave built up the FORTIS + EUDET telescope system.
More details found in EUDET Manual: http://www.desy.de/~gregor/EUDET/Manual/manual_M26_2010.pdf
Our "Detector Under Test", mounted between the two parts of the EUDET telescope, is the FORTIS 1.1 sensor. Its schematic details can be found at:
We noticed a feature that some columns are behaving badly.
We run with 120 GeV/c pions. Beam file H6A.712. The telescope is reasonably aligned. As the beam file might change tomorrow morning, we will not optimize it for now.
We have taken some initial data to test the DAQ and it seems fine. Real data taking starts with 9318. We have until tomorrow morning 9am until ATLAS upstream people might change beam files etc.
The layout for all running with ATLAS pixels (i.e with the FORTIS and ATLAS sensors all under the same box) is shown here: fortis-cern-h6-atlasbox.ppt
Issues during data taking
23:37 June 4th
Bug in pivot pixel marking fixed. In previous runs, this went wrong in the second run after starting the DAQ. This results in errors in the data integrity check. Data should still be useable. First new run is 9356.
02:25 June 5th
Just found that the wrong config file was loaded. The data will contain real hits and pedestal events. Of course noticed when beam is away... First new run is 9396.
05:00 June 5th
Beam was gone. No new counts. Checked current in wire chambers. Checked magnet status. Magnets lost programming. After refreshing magnet status a couple of times the magnets came back and thus the beam as well. can only solve this from little control room next to ours. Happened after 106k events in run 9401.
08:00 June 5th
The French have dumped the beam. Finished in run 9407 after 94828 events.
11:00 June 5th
Moved to taking data with C1-C4 vectors. First run with new config = 9414
02:00 June 6th
Atlas given up for today. Now taking data with FORTIS only. First run is 9449.
21:00 June 6th
Moving to D-row.
06:30 June 7th
Move back to C-row. D looks particularly dodgy.
01:00 June 8th
Swapped stacks. This solves the problem of the weird columns. Now taking data
08:00 June 8th
Swapped sensors. Took out the DPW (L2-8AC2) and put in the high res (L5_17AC2).
17:00 June 8th
Swapped sensors. Took out the high res(L5_17AC2) and put it a reference one (L1_1AC1).
01:00 June 9th
Swapped sensors. Took out reference (L1_1AC1) and replaced with the DPW (L2-8AC2). Taking data with large pixels.
01:10 June 10th (JW on shift with Florian from ATLAS)
Run 9693 in progress. Watching the beam scalers increment burst by burst, we notice that scaler 1 counts less than the others. Averaging over 5 bursts, rates per burst are: scaler 1 = 16k; scaler 2 = 49k; scaler 3 = 49k; scaler 4 = 52k; Particles (ie the 4 fold coincidence = 220.127.116.11) = 10.1k .
Conclude that 2,3 and 4 look completely consistent but that 1 has significantly lower rate. The number of "Particles" is being limited by scint 1. At our last interventions, all the counters looked reasonably well aligned (excpt that 3/4 looked displaced relatively by a few mm). We check the magnets and collimators - all are at nominal values. I was not aware of this effect but it may be longstanding; the event rate is 70Hz and a run takes about 25mins which has been the norm over the last few days, we think.
We decide to run on and not to intervene in the area to check alighnment etc. Will discuss problem with next shift.
09:50 June 11th (JW on shift with Koloina from ATLAS)
At the beginning of R9783, Particles and Triggers were accumulating normally with beam singles = 10k/burst; rate = 97Hz. But "Events built" remained frozen at 97709 and File bytes at 956Mb. This problem of not writing data at the start of run had appeared several times during the night. I think. We stop the run, XKILL and restart' then all looks fine again. We shall need to watch out in case this happens again.
17:30 June 11th
Swapped L2-8AC2 (DPW) for L5_17AC2 (High res). Got about 13M tracks with DPW, after acceptance correction we should have 4k tracks per 5x5mu bin on the large area one. Hence swapped to high res to measure the difference.
06:45 June 12th (JW on shift)
Several significant problems were present during this shift:
1. Beam intensity is about a factor 6 lower than in previous days (see comment above).
2 For every run, the number of "Particles" is a little over half the number of "Triggers" and of "Events built". (These latter two are always identical). This discrepancy between Particles and Triggers remained when runs were stopped/restarted, when the Mac was reset; when the TLU and APIX PC were rebooted. It has been a constant feature during the shift. The timestamps appear to be correct however - the expected correlation plot etc are seen.
3. Occasionally at run end, the "File Bytes", "Evebts built" anf "Triggers" remain frozen at their values from the end of the run. The new run starts with Particles incrementing, the beam scalers counting but "File Bytes", "Events built" and "Triggers" not changing. This effect seesm to happen at about every 5th run.
18:00 June 12th (JW on shift)
Run 9873 in progress. We are now in a new (perhaps final) mode of running. Several changes have occurred during the day shift.
1. Mathieu (upstream in H6A) has changed the beam profile. He is now using H6.812 insteadh H6.712.(H6.812 is aprofile specifically to be used for H6 when H8 is running on protons). Mathieu had been surprised that the N Hall expert had not installed this proper file when the switch to protons in H8 was made on Friday. The results are good; beam intensities are back to their pre-Friday values:
beam scalers = 16.5k, 50.2k, 54.2k, 56.0k per burst ; Particles = 10.7k per burst data taking rate = 82Hz.
Now Events built (= Triggers) comprise about 30% of Particles ie the anolalous ratios of earlier (Triggers = 2 * Particles) have disappeared !
2. ATLAS pixels have stopped running. In winding up their HV beyond 500V, they began drawing current and have burnt something out (I think). They hope it is just a front end readout element - not the sensor itself. So we are now running standalone - from R9869 (ie from about 1610hrs).
05:20 June 13th (Joel on Shift)
Run 9904 stops accepting triggers after about 24k events (particles still incrementing during spill). Have to restart run.
17:20 June 13th
Final changes of plans. Stopped D-E structure on high res. Now moving to C-structure on HighRes. First new C-row run is 9934.
18:03 June 13th
Made a small magnet scan to improve trigger rate. But we are at the maximum. Nothing to do but wait until runs complete....
19:30 June 13th
Exchanged hi-res sensors: L5-17AC2 removed and replaced by L5-17AC1. Still reading the C row, C1-C2-C3-C4. First run with new hi-res L5-17AC1 is R9948. Last (very short) run with L5-17AC1 was R9955.
23:30 June 13th
Hi-res sensor L5-17AC1 removed after 8 runs (-> 800k events) . Reference sensor L1-1`AC1 was mounted in its place. With the reference sensor, we continue to read the C row ro check against earlier data. (Runs 9961, 9962 and 9963 (not complete) were taken with this reference sensor, C1-C2-C3-C4.(250k events approx).
00:50 June 14th
Run 9963 (C row with reference sensor L1-1AC1) was stopped as there were very many warnings "Mismatch in event number". These errors continues in next run 9964. So we stop this (very short) run and change from the configuration from C1-C2-C3-C4 to D1-D2-D3-E of the reference sensor. First run with reference sensor, parts D and E is R9965.
Take 1M events with C-struc on HighRes. Then swap sensors. Take 1M on C-struc Ref and then as much as possible on DE for Ref.
Took huge sample with Deep Pwell. Now taking data with High res to compare. Will run until the end with high res.
Swapped stacks to check whether noise etc where better. Made large improvement. Decided to stick with same Deep Pwell sensor for now. We want to take enough data with the C-section of each sensor to measure position resolution. After that want to measure charge collection variations in the large pixels. We do not have enough time to measure this for all types, so need to pick one.
To measure position resolution, will need 10k evts on a pixel type. Our pixels are ~4mm^2 and the scintillators 200 mm^2. So we need to collect 500k events per sensor. One run contains roughly 100k events.
Done Deep Pwell.
Done High Res.
Move back to old plan. D-row is crap.
Have taken a large data sample with the C-row. Now changing to D/E row. So please use the correct configuration file: eudet2-beam-6boards-in-1crate-FORTIS-D1-D2-D3-E1
Mechanics mounted in Atlas mounts. Cannot run together as Atlas DAQ keeps busy high.
Keep taking data until 09:00 with either C2 or the entire C-column. At 9am Harris will dump the beam to install new sensors. We will use this time to make a survey of the setup. Then data taking will continue until 18:00. Then the ATLAS colleagues will install their new system which means that we will move to the new mechanics.