The Jan/2016 test beam sessions were used to commissioning the DUT readout integration with there newly installed CALADIUM EUDET telescope. Two old FE-I3 pixel modules from the dissembled old cosmic telescope are used as the DUT. The initial readout system was still the Gen-1 RCE and old HSIO as used for 2014. 

Detector Configuration

DUT namemodule IDHSIO linkPosition
C3M51273712Back, vertical
C4M51285511Front, horizontal

Sensor positions (mm). Beam from left to right. Blue planes are EUDET telescope. ATLAS pixel module are red DUT above. C4 can be seen in picture on the right.

Left diagram: only drew C4 DUT, while C3 DUT is behind the Al plate to the right. 

Right picture:

  • Top DUT stage has the PP0 box with HV.
  • Middle shelf has L->R: DVDD PSU (2.1V); AVDD PSU (1.9V) and HSIO.
  • Bottom shelf: 12V PSU for PP0 box HV.

PP0 box ground, HSIO ground strap are tied together to the AVDD PSU ground.

DAQ Configuration

  • Datataking driven from ar-eudaq logged  in a tfuser, with telescope operation from the operating directory  /home/tfuser/eudaq
  • The DAQ configuration files are at /home/tfuser/eudaq/conf on ar-eudaq, with names such as ni_fei3_c4only.conf
  • The logged data are at /home/tfuser/data/ on ar-eudaq with names like run000xxx.raw
  • DUT DAQ was with Gen-1 RCE + old HSIO controlled from esa-user02 with module config default directory at e.g. for C4:

Event Log

  • Jan/5: Matt modified one of the old cosmic telescope plate with two modules C3/C4 to mount them within CALADIUM DUT space and roughly alignment with the EUDET sensors. The DUT plate is support on a manual XY stage.    
  • Jan/6: First round connection of PSU/readout to the DUT modules. The Gen-1 RCE DUT readout node esa-user02 could not boot with /u1 disk failure.
  • Jan/8: Matthias replaced failed /u1 disk and reloaded RCE DAQ software on /u1 and brought in another spare Sun server. Network config still not right. 
  • Jan/11: After getting instructions from Zen, Michael Dunn, reconfigured the network patch panel to switch ssh-061b-nw01 connection inside ESA, esa-user02 is reestablished at its previous private network IP of 
  • Jan/12: Obtained configuration files of FE-I3 modules from Martin and picked two random FE-I3 module's config space for C3 and C4 to start trying calibration, and created new testbeam20160112 directory for user tbslac onesa-user02 as the default DUT configuration directory. Initial calibration failed still at first stage. Joined Carsten for the ESA walkthrough before first beam. One front section of beam pipe missing, but at least cleaned out heavy items from beam line and used thin flanges at the breaking point just before the beam started. Learned the initial rough laser alignment: works pretty well for vertical, but horizontal alignment was cumbersome as it was very difficulty to get a exactly vertical layer plane. Learned EPIC controlled XY-stage adjustments and remote control of power strips (very handy for RCE shelf in particular). First play with beam using just the EUDET telescope: 6 GeV beam and very broad (larger than EUDET sensors). ~10 particles per crossing going through EUDET sensor capture.            
  • Jan/13: Matthias installed Xilinx dongle driver Impact tool on esa-user02 and reprogramed HSIO from V3->V8. Now started to see working calibrations for C3, but C4 still fails. Not bothered with C4 and just using C3 to try integration with EUDET. Got RCE producer connecting to the ar-eudaq run control node. 
  • Jan/14: Continued integration test all day. Had to get an proper example from Marco Bomben as the 2014 ATLAS run used a computer brought from CERN which was taken away with all the ESTB config files. Started to see DUT data online, but no obvious correlation with EUDET data. Calibration looks poor and module noisy. Poked around with raised GDAC etc. Tried some different try delay values and still no obvious indication of timing peak. Last beam at ESA for this week and no beam until at least next Wednesday.  
  • Jan/15: Not clear if the DUT data were logged properly in the Jan/14 tests. FE-I3 pixel modules do not have a standard module type for test beam config. What did the DAQ do with that for logging ? Investigating raw data dump offline utility. 
  • Jan/19: Back to ESA and identified the real module IDs: C3=512737; C4=512855. Swapped C3/C4 link channel on HSIO: C3 can work with either link 11 or 12, while bizarrely C4 can only run with HSIO link 11 but not link 12. Poked about with connector on PP0 box and swapped sables back and forth still consistently seeing this effect. Kept C3 on link 12 and C4 on link 11 so we have both modules working. Also edited the config file to point to the original calibration directories of these modules, sensible threshold calibration results were seen.
  • Jan/23: Some play of integration indicated that DUT index should be packed with consecutive active channels starting from 0 to ensure overall active sensor count to be correct. Index gaps in DUT can cause global EUDAQ monitor plots to miscount number of planes and garble the other plots. Still cannot get C3 at Link=12 to generate any hits, so continued with just C4 at Link=11 as single DUT. Took a bunch of timing runs 96-104 until beam ended at 10pm. These run unfortunately didn't hit the DUT timing (later realized beam trigger delay was ~5ms off)  and beam particle rate also very low.    
  • Jan/25: Final attempt on timing. Switched to 10.5 GeV electron and Tonee found a setup with momentum slit tighted down, gap=0.125mm, C24 left at 8mm that gave a few particles per trigger in EUDET aperture (eudaq monitor hits/event ~230, clean MIMOSA planes show ~12 pixels/event). Carsten came suggested to go for higher rate beams with slit gap opened to 1mm and much higher rate. He pointed out the "Trigger" panel in the ESTB EPICS GUI contains the beam trigger gate delay and width. We played with this instead for timing scan at high beam intensity. Based on hit X,Y distributions, adjusted XY stage to X=72; Y=91 (detectors may be still a bit high in Y, but very broad beam anyway). After jumping 300ns steps back and forth eventually found the magic spot:  Experimental Triggers panel channel 10 (Silicon Tracker) Delay (ns)=-104950 (originally at -100000), in conjunction with ni config TrgDelay=232, Latency=232 will put the C4 DUT data time centered at tick=8. Increasing TrgDelay and Latency in config together by same amount results in same observed time tick for the data. This confirms the config latency value for FE-I3 is counting forward. EUDET readout didn't seem to care much about the beam trigger delay change. The beam trigger width was initially at 10000, but dialing down to even 100ns everything still works. Brought beam back to the normal 1 particle/trigger rate and took a long Run=134 for 10K triggers, with configuration ni_fei3_C4only.cfg. Tried briefly to bring in C3 on Link 12 and still doesn't work. We are at the end of this beam period. Powered off DUT PSUs, HSIO and RCE ATCA shelf with remote control. Ready to dismantle for the end of this period and XYstage upgrade.              

Test Runs and Results

Run typeRunDateRun categoryModulesResultsRemarks
Calibration182Jan/19Threshold scanC3,C4threshold_000182.pdfFirst sensible calibration after settled both C3,C4 to there own calib folder
Calibration183Jan/19Digital scanC3,C4digital_000183.pdf"
Beam134Jan/22 C4 Final benchmark long run with DUT timed in. 20K events. 10.5 GeV electrons at ~ few particles/trigger. 
  • No labels