PHOTOELECTRIC DETECTORS 



the tube (unfortunately the anode pin is adjacent to the cathode pin). This 

 leakage can sometimes be reduced by cutting a groove in the base between 

 the pins and coating with a silicone preparation. Most of the dark current 

 is due to thermionic emission from the photocathode and dynodes. Reference 

 to Figure 28.31 will show that the dark current increases rapidly as the 

 interstage voltage approaches 110. At higher voltages the photocell will 

 become very 'noisy' and may show ionization. The maximum signal-to- 

 noise ratio is obtained at about 90 V per stage. Some trial and error experi- 

 ments are required for each individual tube to ascertain the optimum 

 conditions. 



The thermionic emission from the photocathode depends upon its tempera- 

 ture and a considerable improvement in performance can be achieved by 

 cooling the cell to — 78°C with solid carbon dioxide or — 198°C with liquid 

 nitrogen (Figure 28.33^^). 



To detect very low-intensity light signals it is of course an advantage to 

 chop the light beam at some fixed frequency and then pass the a.v. signal 

 from the photocell through a high-Q amplifier tuned to that frequency. If 

 dark current noise is displayed on a C.R.O. it appears as 'grass'; each 'blade' 

 differing in size from the next in a random fashion. Noise due to an electron 

 being liberated from one of the higher dynodes is of smaller ampHtude than 

 an electron liberated from the photocathode. By counting the impulses 

 greater than a certain amplitude appearing over a period of time, first with 

 the cell in darkness and then with the light signal on the photocathode, and 

 then comparing the counts, very great sensitivity can be achieved. This 

 counting technique yields very high signal to noise ratios, since the effective 

 frequency band-width is equal to the reciprocal of the counting time. 



The output of a photomultiplier is linearly related to the light intensity 

 over a wide range provided care is taken to ensure that the dynode voltages 

 do not change with dynode current. The current flowing in the bleeder 

 chain shown in Figure 28.30 should be at least ten times greater than the 

 maximum current drawn by the last dynode. At high light intensities 

 non-linearity will occur due to space charge effects in the last stages. 



The frequency response of these tubes is very wide and can extend to 

 several megacycles ; the transit time of the electrons in the tube limits the 

 response. 



Fatigue effects in the 931 A are small providing photocurrents not exceeding 

 10 /^ A are drawn and the overall voltage is kept down to 700: the fatigue tends 

 to occur during the first 30 minutes but variations are found from tube to tube. 



The IP21 is a specially selected cell of the 931A type and should be used 

 for the detection of signals of low intensity. Mullard 27M2 is similar to the 

 931 A. The 27MI is a specially selected and tested version of the 27M2 and 

 has a signal-to-noise ratio about 10 times better. These four tubes have a 

 type S4 spectral sensitivity. IP22 has a type S8 sensitivity. IP28 and 27M3 

 have an ultraviolet transmitting glass envelope which enables signals with 

 wavelengths down to 220 m/^ to be detected. Worthy of special mention are 

 the E.M.I, group of photomultipliers. These tubes use a semi-transparent 

 end window photocathode which absorbs some of the radiant energy striking 

 it. Photocathodes with areas ranging from 78 to 9,700 mm^ are manufac- 

 tured. The dynodes have a 'Venetian blind' type of structure consisting of 



365 



