Sec. 5-1] 



RA DIA TK) .V TliA SS hl r ( 'E US 



253 



order of 10 -14 amp), and i p the photocurrent at the cathode. The 

 signal-to-noise ratio is 



S 



N 



\2e &f(i t + i v )l 



(1) 



Engstrom 1 remarks that this equation does not include the noise 

 resulting from thermionic emission from the dynodes, which he esti- 

 mates to be 3 per cent of that expressed in Eq. ( 1 ), nor that associated 

 with random secondary emission, of the order of 15 per cent. 



90 

 80 

 70 

 60 

 50 



I 

 I 



Uo 



i 



30 

 20 

 10 



-10 



Volts per stage =100 

 _ Tube: Type 1P21 _ 



Db reference: db =1 microvolt across 1 meg load 

 Bandwidth of analyser = 1.8 cps 



' Cathode illumination : 

 Lumens = 1.5 x IO" [0 ~ 

 Color temperature = 2,870° K 



- Chopped, 90 cps 



-240 



10 



10"'° 3 



E a> 



a> o 



io-" £ s 



10" ,2 Z 



> CI 



— £ 



-120 -80 -40 

 Tube temperature, °C 



10 



20 



13 



Fig. (5-1)19. Variation of signal and noise with the tempera- 

 ture of the photomultiplier. The signal-to-noise ratio can 

 be found as the difference between both curves [from R. 11'. 

 Engstrom, Rev. Sci. Instr., 37, 420 (1947); by permission]. 



From Eq. (1) one would expect that the signal-to-noise ratio is 

 independent of the amplification and, therefore, independent of the 

 voltage applied per stage. However, this is not so in practice. 

 Experimentally obtained characteristics, shown in Fig. (5-1)18, indi- 

 cate that the signal-to-noise ratio decreases slightly with increasing 

 voltage per stage until, at a voltage where regenerative ionization 

 sets in, the S/N ratio declines sharply. 



The most effective means to reduce the noise level in photo multi- 

 pliers consists in reducing the thermionic emission by operating the 

 tube at low temperature. The effect of temperature on the signal 

 output and the noise output is illustrated in Fig. (5-1)19. Reduction 



1 Engstrom, loc. cit. 



