Title: A Test Bed for the Giant Air Shower Array Electronics
The test bed will be built around a Philips XP1901/03 phototube, powered by a passive base. Hand wired electronics will be assembled to test the single tube path. The dark box arrangement will allow testing with a small tile, testing with a fast two channel pulser, and testing with a slower multi-hit (up to 100) pulser. A set of neutral density filters built into the light box will allow amplitude measurements.
Figure 1. is a block diagram of the test bed.
The the XP1905/03 phototube will be mounted (Fig. 1a) so that it views a pair of blue LEDs though a set of neutral density filters. The LEDs can be driven from a fast or a slow pulse generator. The fast generator is driven by avalanche transistors and generates nanosecond pulses. One pulse can be slid +/- 100 ns around the other. The slow generator is limited in speed to about 20 MHz. There is a 100 position plug board which allows pulses in any of 100 positions (as long as the operators fingers hold out) if hooked up as a single pulse train generator. It also allows 2 pulse trains of 50 pulse positions if hooked up in the dual pulse train mode. Through another mode selection, adjacent pulse positions can be either 50% or 100% duty cycle. The 100% duty cycle mode would allow simulating closels spaced hits merging so as to appear as a single pulse. The phototube will be powered by an external high voltage supply.
The output of the phototube will be connected to a simple JFET (4 transistor version) integrating amplifier (Fig. 1a). The prototype of this amplifier requires only 45 mw. The noise level of this amplifier is expected to be of order 10,000 electrons (to be measured). Using an amplifier C of 100 pf, assuming 5 photo electrons as a typical signal, and a design amplifier output signal of 1 mv, a PM gain of 125,000 is needed. This is easily achieved with the XP1905 (even with a stage removed). The amplifier noise will be well below any signal. The 1 mv amplifier output will allow a dynamic range of about 2000 simultaneous hits. By reducing R below that needed to discharge the amplifier for the relatively slow average hit rate, higher instantaneous rates can be accommodated. This can produce and undershoot error which can cause under-measurement or missing pulses. Ref. [2] discusses a technique that can reduce this error.
The shaper (Fig. 1a) uses a 20 ns delay line to convert the step output of the JFET amplifier to a 20 ns wide pulse whose amplitude is proportional to the charge collected by the PM tube. The shaper prototype requires 45 mw of power and has a gain of 3. Since the shaper is a differential device which looks at the input and the output of the delay line, it is possible to have different gain in the two paths to implement the "trick" of reference [2].
The output of the shaper (Fig. 1a) is supplied to the trigger path and drives a 50 ns lumped constant delay line. The output of the delay line is supplied to the analog and the digital measurement path.
The output of the shaper (Fig. 1b) is amplified by a x8 buffer amplifier so that a reference 5 photo electron hit will produce a 24 mv, 20 ns wide pulse at the discriminator input which is 1.2x the tested threshold.
The discriminator (Fig. 1b) requires 40 mw and when set for a 20 mv threshold, has a 2x - 10x time slew of 7 ns. The output is a 20 ns wide 5 volt CMOS pulse with rise and fall times of order 2 ns. Note that it would be easy to increase the gain of the PM to reduce the time slewing. This, however, will reduce the dynamic range of the system. There is plenty of room to maneuver gain and discriminator threshold for optimum performance.
The discriminator (Fig. 1b) triggers a one shot which opens a time gate for the counting of pulses in the digital measurement path. It also sets a Flip-Flop which opens the "before" switch. The end of the time gate sets a Flip-Flop which opens the "after" switch.
The digital measurement path consists of another buffer amplifier and discriminator which is connected to the output of the delay line (Fig. 1c). An AND gate is opened by the gate to allow counting the discriminated pulses in a scaler. The scaler is read out by the computer after which it is reset for another event.
The analog measurement path consists of an integrating amplifier (Fig. 1d). Its output is connected to two sample and hold circuits. One sample is measured by opening the "before" switch prior to the event exiting the 50 ns delay line. The second sample is taken at the end of the gate. A amplifier measures the difference between the two samples which represents charge that arrived during the gate open time. Note that it is possible to make high precision measurements with this technique even in the presence of "early" hits. It is also possible to achieve a dynamic range of 1E6 depending on the noise conditions. CDF calorimetry has this dynamic range in the calorimetry, though it has not been used. We have about 15 years experience with this "double correlated sampling" technique. It was used with several generations of liquid argonne systems where the signal was quite small, was used with proportional chambers in a spark chamber environment where the noise cancellation features were important, and it is also the basis of the CDF calorimetry. In all we have experience with of order 100,000 channels and 8 or so large systems. It is cheap, works with relatively slow components, the double correlation cancels most types of electronic error, and it is relatively easy to understand and to repair (important for large systems where technicians have minimal training). Again it is possible to use the scheme of Ref. [2] to allow small values of R in the integrating amplifier.
The following test list is intended to uncover electronic design deficiencies. We will welcome other test suggestions, particularly those which relate to physics questions.
What we plan to test:
[1] "Segmentation and Time Resolution Required for Surface Detctors in a Giant Air Shower Array", James W. Cronin, Unpublished, 1994
[2] "Design and Operating Experience with Electronic Systems for High Rate Liquid Argon Calorimeters." T. F. Droege et. al. IEEE Trans. Nucl. Sci. 27 (1980) 64-67