Anal. Chem. zyxwvuts 1980, 52, zyxwvu 1273-1278 1273 High Speed Pulse Amplifier/Discriminator and Counter for Photon Counting Richard A. Borders and John W. Birks" Department of Chemistry and Cooperative Institute for Research in Environmental Sciences (CIRES), University of Colorado, Boulder, Colorado 80309 John A. Borders Naval Ship Weapon Systems Engineering Station, Port Hueneme, California 93043 The design for a Pulse Ampllfier/Discrimlnator (PAD) and high speed counter is given. The PAD is completely self-contained, has complementary Emitter Coupled Logic (ECL) outputs which can drive a 504 load (coaxial cable) to a remote counter, and costs less than $150 to construct. The PAD is typically capable of detecting a single pulse with an amplitude of I 100 zyxwvutsrqp yV, typically has a Pulse Pair Resolution (PPR) of 10 ns for 350-pV to I zyx 40-mV pulses, and can operate to >250 MHz with periodic Input pulses of I8 mV. The front end of the counter circuit uses high speed ECL flipflops. The output data from the front end Is converted to TTL, and the remainder of the counting Is done using slower, cascaded l T L counters. The counter is capable of counting at frequencies >250 MHr, and the maximum count is 2= - 1. The counter is interfaced to a 8080-based microcomputer. The PAD and counter were tested using a matched pair of EM1 9658RA Photomultiplier Tubes (PMTs), and the relationship between current and photon counting was found to be linear within 5% to >2 MHz. Pulse (photon) counting has been widely compared with other methods of measuring the output signal of electron multipliers and found to be the method of choice for systems with low-to-moderate signal fluxes zyxwvutsrq (1-1 I). The advantages of photon counting over analog measuring systems are (1) discrimination against noise which does not originate at the photocathode, thereby increasing the system signal-to-noise ratio; (2) digital processing of inherently discrete spectral information; (3) elimination of the errors associated with data domain conversions; (4) sensitivity to very low light levels; zyxwvuts (5) accurate long term integration; and (6) reduced sensitivity to l/f noise, long term drift, voltage changes, and temperature changes. The disadvantage of photon counting is that pulse pileup occurs at moderate-to-high signal fluxes causing the signal to become nonlinear (3-7, 12-15). Pulse pileup is caused by either the arrival zyxwvutsr of two or more photons a c h e photocathode within the pulse width of the first photon or by ~~~~~ the arrival of photons at a rate which the amplifier/counting system cannot resolve. The measurement of high signal fluxes re- quires an analog measurement system or the use of dead time and/or count loss correction techniques (3,12-15) to extend the range of photon counting. The use of faster counting systems applies the advantages of photon counting to higher signal fluxes provided that the resolution of pulses is limited by the counting system, as is often the case. A block diagram of the pulse counting system is shown in Figure 1. The electron multiplier may be either the electron multiplier associated with a mass spectrometer or a PMT. In our system, the amplifier and discriminator are combined into a small (12 inch X 2 1/4 inch X 2 inch), self-contained module (the PAD). The output from the PAD is counted for a period of time which is controlled by a microcomputer. The microcomputer collects the data from the counters (and other sensors), processes the data, and prints the result. The PAD and counter described in this article perform as well as, if not better than, commercially available systems at a fraction of their cost. EXPERIMENTAL Definition of Terms. Previous articles which describe am- plifiers or amplifier/discriminators have not rigorously defined the terms which they have used to describe circuit behavior. This makes it extremely difficult to decide whether the circuits de- scribed are suitable for a particular application. To avoid this problem we define the following terms. False Alarm Rate. The false alarm rate is the number of false counts per second from the PAD when the input is terminated with its characteristic impedance (50 Q). The false alarm rate is a measure of the internal noise of the PA11 and is strongly dependent on the discriminator level. The peak value of the false alarm rate occurs at the voltage where the comparator changes logic states, the threshold voltage. Normalized False Alarm Rate. The normalized false alarm rate is defined as the ratio of the false alarm rate at a particular discriminator setting to the maximum rate to which the system can respond. For example, if this maximum rate is 100 MHz, and the sensitivity is specified for a normalized false alarm rate of lo4, then the sensitivity is measured at the discriminator threshold voltage which gives a false alarm rate of 100 Hz. Sensitiuity. The sensitivity is the pulse amplitude for which the output frequency of the PAD is equal to one half the frequency of the input signal (Le., the level of the signal a t which one half the pulses are detected and one half are not detected). If no normalized false alarm rate is given with a sensitivity, then it is implied to be lo4. The combination of sensitivity and normalized false alarm rate provides a uniform method of comparing PADS that have different internal noise levels and response rates. Pulse Pair Resolution (PPR). We define the PPR as the period of time between the trailing edge of the leading pulse to the leading edge of the trailing pulse (both at the half-maximum) at the point where the frequency of the output signal is 1% less than the frequency of the input signal. This term is defined so as to minimize the effect the pulse width has on the PPR; however, the PPR will still be affected to some extent by the pulse width. Design Goals. The PAD described in this article is intended to be useful with a variety of analytical instruments which utilize photomultipliers and electron multipliers. In order for the PAD to be useful for most applications, the primary design criteria were chosen to be sensitivity and pulse pair resolution. Sensitiuity. It is possible to estimate the approximate sen- sitivity needed for a particular PMT, once its gain and pulse width are known, if one assumes a triangular shape for the pulse (10). The gain varies widely from lo4 to los, with 5 X lo6 being typical, and the pulse widths vary from 2 to 30 ns, with 10 ns being typical (16,17). The approximate peak amplitude of the pulse is A,, G is the gain of the electron multiplier, Wp is the Full Width of the pulse at Half-Maximum (FWHM) amplitude, e is the charge on an electron in Coulombs, and RL is the anode load resistor (typically 50 a): fiG,? RL (1) A, = -- WP Substituting the typical values into this equation gives the result 0003-2700/80/0352-1273$01 .OO/O t2 1980 American Chemical Society