BIOLOGICAL AMPLIFIERS 



taps on the HT battery; then to the nearest 1^ V by taps on the grid bias 

 battery. Final adjustment to within a few milUvolts is performed on the 

 2-5 k ohms potentiometer, which is suppHed from a large unit cell (U2). In 

 this manner the output impedance of the supply is never greater than about 

 1-25 k ohms if the batteries are in good condition. The price paid for the 

 use of a triode cathode follower, rather than pentode, is an input capacitance 

 of about 4 pF instead of some 1| pF. 



In the complete amplifier (Figure 39.7) the interstage and output coupling 

 networks are seen to be of the potential-divider type and about | of the signal 

 at zero and low frequencies is lost in each. However, the upper resistors of 

 the potential dividers are shunted by capacitors so that fast alternating com- 

 ponents are passed without attenuation. Neglecting for the moment the 

 network associated with the first amplifier stage cathodes, the frequency 



c 



(0 



en 



a> 

 > 



i_ 



oi 

 o 



/ 



\ 



log frequency 

 Figure 39.13 



response of the amplifier is clearly of the form of Figure 39.13; the full 

 gain is available for the detection of action potentials which may be distant 

 from the microelectrode, and rather less than half for measuring membrane 

 potentials and slow changes such as root potentials. In practice this is 

 usually precisely what one wants, since the latter two are large, and in 

 addition the reduced gain at zero frequency gives less trouble from drift. 

 However, for workers who find such a frequency response confusing, the 

 'high frequency lift' may be approximately cancelled by introducing a series 

 CR circuit with the switch S4; this is simpler than switching out the 4 

 capacitors in the coupling circuits. 



The arrangement of the 'a.c. gain' and 'd.c. gain' controls is due to 

 Matthews^. These terms are convenient but not quite accurate, since the 

 d.c. gain control actually operates at all frequencies, therefore the d.c. gain 

 cannot be more than the a.c. gain; in practice this presents no difficulty 

 because one never wants it to be. The network enables one to reduce d.c. 

 gain steadily until the amplifier behaves for practical purposes as if it were 

 a.c. coupled. This is a much more flexible arrangement than one which is 

 switched between 'pure a.c. coupled' and 'pure direct coupled'. 



Meters are provided to facilitate balancing of the amplifier and to check 

 the state of the heater battery. In addition a simple calibrating device 

 enables one to: 



(a) check the gain (600 max on d.c, 1,300 max on a.c, single-sided output) 



(1,200 max on d.c, 2,600 max on a.c double-sided output) 



(b) estimate the microelectrode resistance or resistances ; 



(c) set the input cathode follower grid current to a low value and measure it. 

 The output terminals are fed in push-pull and both may be used to feed a 



626 



