Decision of bad or hot in FMS

- Goal: set the standards for bad/hot channels to be excluded before calibration

- Conclusion:
   a. Judging hot / bad from # of entries or ADC spectra alone looks inappropriate
   b. I suggest at the beginning of the iteration (calibration) exclude only "apparently bad" cases,
        then judge / exclude additional bad channels based on 1st reco. mass along with ADC spectrum


  1. Hot channels

    1-a. Conventional bad/hot channels decision

    - Collected ADC vs. channel by using "physics stream + single MB trigger (st_physics + VPDMB30) ".
    - Saved ADC range: " 3 < ADC < 250 ". To cut out pedestal tail & To see the spectra clearly
    - For the decision, use "# of ADC entries of a cell / # of trigger (left)" and "Each cell's ADC spectrum (right)".

    1- b. Complication

    - High online calibration case: a cell with relatively large # of entries does NOT guarantee it's hot.
       i.e. a cell with high online gainCorr -> more frequently triggered, ends up with more entries,
       but it can be properly corrected in offline calibration. Simply masking it out can be dangerous.

    - Possible instability in pedestal subtraction: out pedestal subtraction precision isn't 100 %.
       If a cell has relatively long pedestal tail, it is likely the cell's # of entries will be larger than its vicinity.

    - Channels with high bit shift: if I try to judge hot channels by comparing each ADC spectrum,
      channels with positive (+) bit shift will have relatively larger entries in same ADC bin.

  2. Bad channels

    2-a. Channels apparently bad

    - Some channels' ADC distribution looks utterly bad. These cells definitely need to be excluded
    - Following plot shows marked "apparently bad" in currently ongoing iteration (Apr. 29)
    - Channels d11_209 and d11_210 were decided bad by independent QA

    2-b. Channels with high bit shift >= 4

    - BS = 4 channels in RUN15 day 80-81, ADC (left) and reco. mass (right, ongoing iteration)
    - Cells with gainCorr = 1.0 means they were reintroduced to check if they can be used
    - 10 / 14 look decent to be used

    - BS = 5 channels, in RUN15 day 80-81
    - 11 of 14 look bad