BIOLOGICAL AMPLIFIERS 



receipt at the input of a brief pulse of excessive size, and occurs as follows: 

 Suppose the polarity of the pulse is such as to drive the anode of the penulti- 

 mate amplifier valve positive. The grid of the final valve tries to follow it, 

 but grid current flows to charge the coupling capacitor and increase the P.D. 

 across it. At the end of the pulse the penultimate anode returns to its proper 

 mean potential, carrying the final stage grid far more negative than its proper 

 bias and cutting off" the final valve. The amplifier is then paralysed until the 

 P.D. across the coupling capacitor returns to normal, a period which is 

 often about ten times the time constant for the coupling, and may therefore 

 exceed half a minute. In a double-sided circuit a large input pulse of 

 either polarity will block the amplifier. The magnitude of blocking effects 

 depends on the amplifier gain, so we have here another factor to limit this 

 quantity. 



The base line produced by direct-coupled amplifiers is subject to flicker 

 and carbon resistor noise, to the effects of small valve electrode movements 

 and to drift. 100 [xW per in. may be regarded as a reasonable upper limit 

 to the deflection sensitivity in direct-coupled systems. 



It should be noted parenthetically that if a record contains a small event, 

 which might be signal and might be noise, a certain time after the delivery 

 to the preparation of a stimulus, and if a similar event appears at a corre- 

 sponding time upon repeating the stimulus, then the probabihty that the 

 event is a signal and not noise is increased. This gives us a clue to a technique 

 which allows us to measure signals actually smaller than the noise, and which 

 has been exploited by Dawson^. If corresponding segments of the traces from 

 repeated experiments are added, then the signals sum to produce a total 

 which is directly proportional to the number of experiments, whereas the 

 noise produces a total which rises as the square root. Dawson used a bank 

 of capacitors which were charged, in succession, from the amplifier output 

 via a rotating switch synchronized with the time-base generator. In this 

 manner the voltage across each capacitor corresponds to the average of a 

 large number of values from a particular segment of the record, and a voltage 

 distribution eventually emerges from the whole bank which may be scanned 

 to produce a relatively noise-free record. The degree of freedom from noise 

 depends upon the number of times the experiment has been repeated. 



When special techniques such as this can be employed the usable amplifier 

 gain can be increased. The difficulty is, of course, to be sure that each 

 repeated stimulus is producing similar physiological events. If one can be 

 certain that this is so, then theoretically an indefinitely small effect can be 

 detected if one pursues the experiment long enough. 



The division of the recording chain into pre- and main-amplifiers is usual 

 and convenient, for these reasons : 



(1) Flexibility. The nature of the pre-amplifier is determined by what is to 

 be measured. Similarly the type of main amplifier depends on the display 

 device — cathode ray tube, penwriter, etc. More can therefore be done with 

 fewer units. 



(2) Electrical convenience. Whilst the main amplifier can usually be fed 

 from mains-driven power supplies of conventional type, pre-amplifiers are 

 generally supplied from batteries, or from the mains via special, very highly 

 smoothed and stabilized, power packs. 



617 



