A Signal Sampling Photomultiplier Circuit

Shigeo Minami and Katsuhiko Nishikawa Department of Applied Physics, Faculty of Engineering, Osaka University, Osaka, Japan Received 13 August 1965.

In the time-resolved photoelectric photometry for periodically emitted faint radiation, the photomultiplier sampling methods are sometimes utilized.1-4 The sampling gate pulses, which are synchronized with periodically controlled light sources are applied to photomultiplier dynodes in order to sample the photo-current inside the tube. If, on the other hand, a sampling de­vice is connected at the high-impedance output of the photo­multiplier, then this device will also disturb the shape of the original periodic signal which is to be sampled. This report is about a pulse injection circuit for the photomultiplier sampling system. The new circuit gave efficient sampling operation from low-voltage pulses with photomultipliers of the side-window type which are now widely used for photometric purposes.

Two typical methods which have been used for the above operation are the following: first, the gate pulses are applied directly to the supply of the voltage divider chain for all of the dynodes; and second, the pulses are applied to only a few of the dynodes. In the first method, high-voltage pulses are necessary to permit sufficient control of the output current; furthermore, it is well known that the pulse height needs to be strictly stabilized. In the second method, the shut-off of the output current during the closed-gate period, in which the gate pulses are not applied, is not complete. This background output, which appears during the closed-gate period, limits the lowest level of the measure-able signal, in case the duty ratio of the gate pulse train is ex­tremely small.

For the purpose of making an efficient control of the output current, the authors examined the behavior of the side window type photomultipliers such as 931A, 1P22, 1P28, etc., with the variation of each dynode voltage. For the earlier dynode stages near the cathode, the output current controlled with the supply voltage is considerably affected by the location of the area of illumination on the photocathode. Also, the shut-off charac­teristic at the biased dynode voltage is not complete, since some of the incident photons still hit the successive dynodes. For the later dynode stages near the anode, the shut-off behavior is also poor because of the high-electron density in the signal current. Moreover, the control signal applied to the later dynode stages induces an intolerable amount of spurious signal a t the anode, which is present even without illumination. In contrast to the above, fairly efficient control is provided in the middle stages; however, difficulty still remains in the completeness of the shut-off operation.

After the further examination of the characteristics with combinations of dynodes, the authors finally chose the connec­tion shown in Fig. 1. The even-numbered dynode stages are deeply biased from their normal rating, while the others are held at normal voltages. The gate pulses are applied to those biased dynodes in parallel through the load resistances. In the

January 1966 / Vol. 5, No. 1 / APPLIED OPTICS 173

Fig. 1. Diagram of the signal sampling photomultiplier circuit.

Fig. 2. Typical waveform of sampled output for constant light flux.

Fig. 3. Relationship between gate pulse height and integrated signal output for constant light flux.

system shown here, the output of a blocking oscillator is fed through a 75-Ω coaxial cable. The potentiometers are individu­ally adjusted so as to minimize the output signal of the illuminated tube when the gate pulses are not supplied, and to maximize the signal obtained with fairly small gate pulses. The above ad­justment makes it possible to obtain a ratio of open-gate desired signal to closed-gate undesired signal as large as 104. Figure 2 shows a typical output signal waveform from the anode under the sampling operation, as the gate pulses of 60-V height and 100-nsec duration are applied to the tube being exposed by the

174 APPLIED OPTICS / Vol. 5, No. 1 / January 1966

constant light flux. From the figure it is seen that a time resolu­tion of 80 nsec could be attained under this condition.

The characteristic of the circuit for constant incident flux is shown in Fig. 3. The sampled output at the anode, with a repetition frequency of 10 kc/sec, is integrated with a simple RC network and fed into a high-impedance dc vacuum-tube volt­meter. It is desirable that the output signal should not be sensitive to variations in the pulse voltage. Fig. 3 shows the dependence of the output on the pulse height voltage.

The authors wish to thank S. Fujita for his valuable dis­cussions.

References 1. H. M. Crosswhite, D. W. Steinhaus, and G. H. Dieke,

J. Opt. Soc. Am. 41, 299 (1951). 2. C. F. Hendee and W. B. Brown, Philips Tech. Rev. 19, 50

(1957). 3. K. B. Keller and B. M. K. Nefkens, Rev. Sci. Instr. 35, 1359

(1964). 4. M. L. Bhaumik, G. L. Clark, J. Snell, and L. Ferder, Rev.

Sci. Instr. 36, 37 (1965).