Notes for 2015 Time Calibration Service Task

Contents

Time Calibration Overview:

For each period and pass, we need to create a timing calibration map to correct the cell-by-cell timing differences that are inherent in the detector (due to relative position within the detector, length of readout wire, etc.) This is accomplished using a two step averaging technique, which uses the following steps:

1. Separate the events into 4 groups by (Bunch Crossing Number)%4 (has to do with the fact that bunch crossings occur every 25 ns, while the internal detector clock is on a 100 ns cycle). The rest of the steps are applied for each BC%4 group separately.
2. Choose only high energy, photon-like clusters, and use these to compute the average time that each cell fired at.
3. Use this average to do a secondary timing cut, only taking the high energy, photon-like clusters that occur within a 20 ns window around the first average. Use the events that pass this cut to compute a new average time.
4. This average time for each cell (for each BC%4) is stored in a histogram within the OADB file. When used using AliEMCALRecoUtils, each cell is corrected by this offset, as well as a 100 ns offset for each BC%4 group, so that each group is well separated when looking at the timing.

The method ProduceCalibConstants within the Time Calibration task produces two different files for each BC%4 (and separated for low/high gain): a TH1F containing the calibration offset (average), and a TH1F containing the RMS values. The errors in the calibration are determined for each cell as: σ²_i = (x²_RMS,i - x²_ave,i)^1/2

Passes & Periods currently in OADB:

List of Currently Missing Periods/Passes

Note: passes in bold have had the time calibration run on them.

Current Progress on Missing Periods/Passes:

2016-02-08

• LHC10e: Calibration looks good for BC0/BC2, but BC1/BC3 have very low statistics, making calibration incomplete and noisy.
• LHC10f: Need to run comparison between 10f and 10d to see any overall offset differences. Of note are structures that appear in 10f BC0 (BC2) that appear at ± 150 ns (and ± 75 ns). Need to do run by run analysis to see if I can identify where these are coming from. Also, similar to 10e, BC1 and BC3 have very low statistics, making calibration incomplete/noisy.
• LHC10g: 10g has very few good runs, and very few runs with EMCAL on. Most likely this was a calibration run before heavy ion started (10h), so this period will be ignored for now.

2016-01-29

• LHC10f pass 4: Step 2 of calibration complete, offsets look good for BC0/BC2, but not for BC1/BC3. Now need to compare with LHC10d OADB offsets just to check.
• LHC10g (all passes): Very few good runs, only a few with EMCAL on. Not sure if worth messing with (or if timing information is even available).

2016-01-27

• LHC10e pass 2: Calibration is complete, however there are strange structures in the raw time vs. cell data that are causing many of the cells (within certain SM's) to have inaccurate calibration offsets. Not sure how to handle these structures, or if this pass will be useful since pass4 Calibration seems OK.
• LHC10f pass 4: Step 1 of calibration complete, producing reference root file in order to evaluate and launch step 2 wagon.

2016-01-25

• LHC10e pass 2: Step 2 calibration complete, running final merging of results to check for comparison
• LHC10e pass 1: Keep getting error on accessing TOF timing, possibly the way timing was stored has been changed. Haven't been able to find an old AliPhysics version that will work with the dataset.

2016-01-12

• LHC10d pass 4: simple comparison with each of the other passes (per BC%0) completed.
  • Notes: Calculated values are within ~1 ns of other passes, except 3 specific cells which didn't have calibration offsets in previous passes, but now do in pass4. Only BC%4 = 0 and BC%4 = 2 avaliable.
• LHC10e pass 4: simple comparison with each of the 10d passes (pass1, pass2, pass3 - per BC%0) completed.
  • Notes: Calculated values are similar to LHC10d, except shifted by ~25 ns (as well as several cells that now have calibration values whereas 10d did not).

2015-12-21

• LHC10d pass 4: step 2 of calibration is complete, need to plot it and compare it with pass 1-3 that are already in OADB file
• LHC10e pass 4: step 2 of calibration is complete, need to plot it, do same for pass 1-3 and compare with each other

Images for Missing Periods/Comparison with Current OADB

• LHC10d pass4:

Time Offsets for each EMCAL cell for LHC10d pass4 data (BC%4 = 0).

• Comparison with LHC10d pass 3 (in OADB)

Difference in Time Offsets for LHC10d pass4 and the LHC10d pass3 offsets found in current OADB file.

Magnified view of Time Offset differences between LHC10d pass4 and LHC10d pass3. Most cells are within 2 ns of OADB values.

• LHC10e pass4:

Time Offsets for each EMCAL cell for LHC10e pass4 data (BC%4 = 0).

Difference in Time Offsets for LHC10e pass4 and the LHC10d pass3 offsets found in current OADB file.

Magnified view of Time Offset differences between LHC10e pass4 and LHC10d pass3. We see an overall difference of ~25 ns compared to the LHC10d offsets.

Relevant Links:

• EMCALQAPeriodbyPeriod - a rundown of EMCAL QA for 2013 and 2012, period by period. 2015 should be accurately reflected on RCT
• CaloQA - a rundown of EMCAL QA for 2011 PbPb data (potentially, only LHC11h is up to date). Also has 2010 good run list, need to double check accuracy
• EMCALTimeCalibration - a overview of the Time Calibration task, as well as how to access OADB offsets from the file.
• ALICE.EMCalCode:HowTo#Energy_and_Time_calibration - how to actually implement time callibration offsets using AliEMCALRecoUtils
• PWG1EvSelDocumentation - overview/documentation for the AliPhysicsSelection class.
• AnalysisTrains - overview/guide for train operations.
• ALICE.EMCal - general twiki for EMCAL detector
• EMCalOffline - twiki for EMCAL Offline group
• EMCalQABadChannels2 - page for EMCAL Bad channel status/results by period

-- JustinThomasBlair - 2015-12-21