The relation between the high voltage (HT) and the gain of the photomultipliers was determined from a set of dedicated "Laser Runs" with varying high voltages in Februar (runs 5012-5026).
The gain, defined as normalized laser amplitude, versus high voltage relation was fitted with a power law. The results of such a linear fit to ln(GAIN) vs ln[HT(Volts)] for one (the best..) module look like
Red lines denote the high voltages and corresponding (normalized) gains in September (higher HT) and June (lower HT). For forward rings the gains in June were up to 4 times smaller than in September (large extrapolation needed).
All gain fit values are stored in a single file with the format
MODULE SEP/JUN sigma power sigma HT_SEP HT_JUNwhich reads like
CSI_0101_R 1.8108 0.0034 5.891185 0.018580 1043 943 CSI_0102_R 1.8191 0.0036 5.792681 0.019037 1019 919 CSI_0103_R 1.7199 0.0029 5.721974 0.017565 1106 1006 CSI_0104_R 1.7217 0.0019 5.608264 0.011676 1083 983
Here, the 1-st column is the module in question, 2-nd the calculated ratio of "SEP/JUN" gains, 3-rd its sigma, 4-th the power from the power law fit, 5-th column gives the standard deviation of "power", and 6-th and 7-th the HT in September and in June respectively (in Volts).
Usage of this file in general is straightforward. The ratio of 2 gains (G1/G2) for a given module knowing the corresponding high voltages (HT1 and HT2) is just:
G1/G2 = (HT1 /HT2)power
It may be interesting to notice that tubes with 8 dynodes have the powers around 6 and the tubes with 10 dynodes have the powers around 8.
/d/kp3/bck/s185/pmt_gain_vs_hv/gain_fits.dat