Introduction

Back about five years ago we performed the original evaluation tests of candidate pmts from several vendors in an effort to choose the best pmt for Milagro. During that series of electrical performance tests we measured the linearity/saturation of the R5912, our Milagro pmt. Figure 1 shows a plot of the light level hitting the pmt versus charge out of the pmt. These plots clearly show that the evaluation pmt (with its 1.87 Mohm base) is linear out to about 150 PEs or so.

Since those tests, information from David Williams, and from Isabel Leonor's laser calibration data seem to indicate that the pmts (with there current 20 Mohm base) are saturating at much lower light levels, perhaps below 10 PEs.

I have measured the linearity/saturation of five production Milagro pmts to try to quantify this change from the evaluation R5912. The methods and results are discribed below.

Measurement Methods

I have used our pmt acceptance test facility to make the measurements of "expected" charge out versus "observed" charge out. Please refer to Milagro collaboration meeting notes dated Oct 1995 for schematic of test facility and layout (note on the electronic layout: the 20 dB attenuators have been moved to a position before the amplifiers. This insures that the amps will not saturate up to the highest light levels achievable with this facility). The test facility uses a Nitrogen laser at 337 nm with a ~0.3 ns pulse width. The light passes through a quartz fiber to each pmt. Between the quartz fiber and the laser is a beam splitter which also holds a neutral density filter in the beam path. The charge out of the pmt is measured using a LeCroy 2228A camac ADC unit.

First I calibrate a set of neutral density filters. This was done using our calibrated 2" Thorne-EMI pmt and low (5-30 PEs) light levels. Two filters were used, one with a transmission coefficent of 0.301 +/- 0.008 and one with 0.105 +/- 0.006.

The general method is to set the light level such that with the filter in the path the pmt outputs about 3-5 PEs of charge. Then the filter is removed and the number of output PEs is measured. This is the "observed" PEs. The "expected" PEs is determined from the first light level and the transmission coefficent of the filter.

The pulse-to-pulse FWHM of the laser output was about 25%. The time stability of the laser output was <5% over about a 5 minute interval. The pulse rate on the laser was about 17 Hz, so each measurement took about 1 minute

Results

Below is a table of the raw results for two of the five pmts tested. Each measurement was the mean from 1000 laser shots. "PEs with" is the pulse height with the filter in the beam. "PEs without" is without the filter in the beam.

          SA0860                          SA0817
T        PEs with  PEs without     PEs with    PEs without
0.301    3.89      10.41           2.89        10.47
0.105    3.37      27.72           3.15        31.07
0.105    4.76      38.59           4.79        45.45
0.0316   2.32      61.11           2.17        76.38
0.0316   3.88      82.0            3.39        105.9
0.0316   5.70      114.2           5.74        152.3
0.0316   8.29      137.3           8.47        181.9
0.0316   14.5      187.0           13.8        216.0
0.0316   21.4      229.0           22.5        269.1
0.0316   32.8      280.7           34.0        335.2
0.0316   51.1      345.4           57.8        420.6
0.0316   73.8      419.3           88.3        502.5

Figure 2 shows a plot of "observed" PEs versus "expected" PEs for the five pmts tested on a log-log plot. Figure 3 shows the same data on a linear scale. Figure 4 shows the same data on a linear scale with the "low" pulse region expanded. Figure 5 shows a plot of "observed" charge versus "expected" charge (in # of electrons) for three of the five pmts (gains are: SA0783 = 2.33x10^7, SA0817 = 1.72x10^7, and SA0860 = 1.95x10^7).

Conclusions

From the results of this study it seems that our pmts are indeed linear out to about 80-100 PEs which is reasonably consistant with the evaluation pmt which was tested several years ago. Also there seems to be significant variation in the saturation characteristics from pmt to pmt. This variation can't be "normalized" out by just considering the pmt's gain as in Figure 5 above.

Any and all comments about these results are welcome. Additional measurements/tests can be done assuming time allows.

Your humble servant, Tony...