UW-River Falls

O.M. Signals, Peak to Valley Studies

IceCube 2007 Home

Gain Studies

Introduction

An L.E.D. light source was used to do tests to determine gain curves over varying input voltages for the the optical modules.

Equipment

Optical Module
Hamamatsu BB-7250, containing P.M.T. type R5912-02
Oscilloscope
LeCroy WaveRunner 6050
Signal Generator
Stanford Research Systems, Inc. Model DG535 Four channel digital delay/pulse generator
L.E.D. Blue

Setup

The L.E.D. is at the bottom of the black tube that is held vertically in the picture to the right. At the other end of the tube is a variable diffraction filter on its darkest setting (it still lets a little light through). The filter is used to dim the L.E.D. as much as possible, while still running the L.E.D. at operational voltages. The O.M. is directly above the L.E.D. facing it.

A pulse generator was used to light the L.E.D. The generator also sent a trigger signal to the oscilloscope. The scope then used a qualified trigger, which first required the signal from the generator and then a signal from the O.M. This eliminated from the averaging any zeros signals from the O.M.
These zeros were common because we were trying to see one photon at a time. In an attempt to get one photon from the L.E.D. to the O.M, the signal generator was set to flash the L.E.D. at 5 Hz, for a duration of 2.5 ns, and an amplitude of 4V. These numbers were arrived at by keeping the amplitude at 4V while lowering the pulse duration until the O.M. could no longer see anything. We assumed this threshold was were one photon per pulse was getting through to the O.M. This is a risky assumption as it is pivotal to the way we calculated the gain and we do not have anything else to confirm this assumption.

Calculating the Gain

To calculate the gain using the above assumption we used the average signal area. By rearranging Ohms law, V= RI, we can get the equation (V*t)/R=Q. Once we have the average charge per signal we can divide it by the charge per electron to get the number of electrons, Q/e^-= N_e. If it is true that the O.M. saw only one photon per pulse the number of electrons per signal is equal to the gain, N_e= G.

Results

We tested from 1000V to 1800V inputs and averaged about 1000 signals per data point. (The first two data points are an exception. Noise levels were too Ten signals were averaged for each of these.) The data for BB-7250 turned out to be very precise as it fits the y = ax^b model with an R² = 0.99991. The resulting equation was used to calculate the input voltage for a 10⁸ gain. The tests were later done on BB-7256 and BB-7085. They had R² values of 0.987 and 0.970 respectively. Only the graphs where the resistance is assumed to be 50 ohms are shown. They are identical in shape to those where the resistance is assumed to be 130 ohms.