The MGAINSEL parameter was changed in assembly 1DA the "long" side of that assembly (RDO=1,ARM=0 apvs 0 to 4 have the nominal value (MGAINSEL=2) and the corresponding neighbors on the short side had the parameter set as is shown below:

RDO=1    ARM=0    GROUP=0    APV=0    VPSP=46         LAT=87  # run 12: 1DA
            APV=1    VPSP=46         LAT=87
            APV=2    VPSP=46         LAT=87
            APV=3    VPSP=46         LAT=87
            APV=4    VPSP=46         LAT=87
            APV=5    VPSP=46         LAT=87 MGAINSEL=2  (no change)
            APV=6    VPSP=42         LAT=87 MGAINSEL=0 (lower gain by 20%)
            APV=7    VPSP=45         LAT=87 MGAINSEL=1 (lower gain by 10%)
            APV=8    VPSP=42         LAT=87 MGAINSEL=3 (raise gain by 10%)
            APV=9    VPSP=46         LAT=87 MGAINSEL=4 (raise gain by 20%)

 A comparison is made for each of the APV that had their gains modified with their neighbor on the "long" side as shown in the figure below:

The figures below have the distribution of adcmax, the maximum value of pedestal subtracted adc in the pulse. All pulses accepted for this study had a good fit to Len's function (chi2/dof <2.)

Figure 1  No change in gain MGAINSEL = 2   APV=4 and 5

Slope for APV=5  -4.29X10^-3 ± 3X10^-4                   Slope for APV=4 -4.28X10^-3 ± 4X10^-4

The red histogram correspond to the APV=5 MGAINSEL=2 (no gain change)

<hr />

Figure 2  MGAINSEL = 1 applied to APV=6  neighbor APV=3

Slope for APV=6  -6.19X10^-3 ± 7X10^-4     MGAINSEL=0                     Slope for APV=3     4.43X10^-3 ± 4X10^-4

The red histogram (APV=6) correspond to pulses that were reduced by 10%

Figure 3      MGAINSEL=1 gains reduced by 20%   APV=7 and APV=2

APV=7 Slope:  -5.44X10^-3 ± 1X10^-3       MGAINSEL=1                      APV=2 Slope: -3.88X10^-3 ± 8X10^-4                            

Red histogram (APV=7)  has pulses whose gain was lowered by 20%

Figure 4    MGAINSEL = 3     APV=8 and APV=1

APV=8 MGAINSEL=3,   Slope: -4.46X10^-3 ± 4X10^-4                         APV=1 Slope:    -5.14X10^-3 ± 9X10^-4

Red histogram: pulses from APV with gain raised by 10%

Figure 5   MGAINSEL=4   gains raised by 20%   APV=9 and APV=0

APV=9 MGAINSEL=4 Slope: -2.11X10^-3 ± 6X10^-4                       APV=0  Slope:   -3.03X10^-3 ± 9X10^-3

Red  histogram filled with pulses from APV=9

 Barring some normalization problems, I would claim that at this point we see the 20% change in gain. The smaller change of 10% is harder to see just comparing the slopes of the max adc distributions. But fitting the distribution to exponentials brings up the difference.

In conclusion, MUXGAIN will be a good parameter to equalize the gains.