Tower Gain Matching Summary

The tower gain matching procedure was iterated 5 times. This process involves taking 100-300K real beam events, finding the slope of each tower spectrum, correlating this slope to the gain of the tower and then predicting the desired gain needed to match the real events to the MC events. At each iteration new high voltage values were calculated in order to achieve the desired gain and new HV input files created.

    Summary of Q files and Gain data
    used HV file used data file output gains &
    new HV     		cuts for
    new HV     		plot of
    gain factors (R)     		remarks
    q# ( default ) crate 3 : mar1/run1001 : 80/19
    crate 3 & 4 : mar1/run1004 :70/19 70/19
    crate 4 & 5 : feb25/run7 : 80/19 40/19     		gainsDef.dat t0 -
    q#_A crate 3 (80/19), 4 (70/19), 5(40/19) gainsA.dat t0 HV set A
    q#_B mar17 run0006+0009 crate 3 (80/19), 4 (70/19), 5(40/19) gainsB.dat t1 HV set B
    q#_C Mar18 run0010+1001+1000 crate 3 (80/19), 4 (70/19), 5(40/19) gainsC.dat t2 HV set C
    q#_D Mar18/run01003 : crate 3 (80/19), 4 (70/19), 5(40/19) gainsD.dat t2 HV set D
    q#_E Mar19/run1012+1013
    crate 3,4,5 : 80/19,70/19,40/19
    		gainsE.dat
    gainsE.gif     		- sec05 , sec06 ,
    sec07 , sec08     		details
    q#_E replay of the same data but:
    - fixed in DB mapping for even channels in 8TE
    - use pedestals only few hours old
    - masked out known dead towers 5TC12, 5TD10, 6TA12, 6TB04, 7TD08, 8TB02
    - masked out towers with sticki bits 6TA07, 8TB07
    - rejected corrupted events using Jim's algo
    - in the isolation algo ignore towers with ADC>pedestal+5
    - use target slope from the source scan as the target slope
    - distinguished tubes & towers
    		outSetE.dat
    plot--> ( gif )     		- sec05 , sec06 ,
    sec07 , sec08     		details
    updated 4/3/03
    q#_F April5, minDaq run ..42-..46, 304K eve
    crate 3,4,5 : 80/19,70/19,20/19
    ( first pp data in 2003)
    Note, timing of crate 5 is 20 instead of 40 , it is marginal     		gainsF.dat
    plot--> ( gif )     		- sec05 , sec06 ,
    sec07 , sec08     		details
    updated 4/5/03
    q#_H April 25 run03002, 170K event
    crate 3,4,5 : 80/19,70/19,40/19
    questionable tubes no longer masked out
    compared good source tower fits in 3002
    to same tower fits in 3003 (q* files) and
    fits are same within error. From now on
    hold voltage for tube 6C-D,7B-E const
    use newslope.C instead of plSoloMip.C
    used hvmod.C with power=1/5 instead of 1/10
    		 gainsH_HV1.dat
    plot--> ( gif )     		- sec05 , sec06 ,
    sec07 , sec08     		details
    updated 4/26/03
    q#_H April 25 run03002, 170K event
    realized using power 1/5 caused bigger oscillations
    instead of smaller. Use same data from q*_H setting
    and predict new voltages using power=1/13 instead of 1/5
    Produced q*_J HV files from this iteration
    Manually changed tubes in q8_J 8TE02 from 969->850
    Manually changed tubes in q8_J 8TE04 from 1128.6->900
    Manually changed tubes in q8_J 8TD09 from 744.6->600
    		 gainsH_HV2.dat
    plot--> ( gif )     		- sec05 , sec06 ,
    sec07 , sec08     		details
    updated 4/30/03
    Column description:

  1. input file Input HV file used in data run
  2. output file Output file contains slopes of all towers, old and new gain, old and new HV setting
  3. cuts : let G = desired_gain/actual_gain of tower
  4. R spectrum : R = desired_gain/actual_gain of tower

    The following plots compare the ratio of the HV and gain used in the default q files as compared to those used in set q_E.
  5. Old/New HV
  6. Old/New gain