POWER PACKS 



but the principle of them all is the same. The output voltage, or a fixed 

 fraction of it, is compared with a reference voltage derived from a battery or, 

 more commonly, a voltage reference tube. Any difference between them is 

 amplified and used to correct the output voltage by altering the d.c. resistance 

 of a valve either in series with the output terminals (series stabilization) or 

 in parallel with them (shunt stabilization or 'absorber valve'). Series 

 stabilization is commoner and will be dealt with in greater detail. 



In its simplest form the electronic series voltage stabilizer is shown in 

 Figure 37.5. V^ is a high /j, triode, and V^ is the series control valve. VR is 



+ 



Unstabilized 

 Stabilized J> \ yf^ supply in 



supply out 



Set output 

 voltage 



Figure 37.5 



the voltage reference tube. i?i, i?2 and R^^ form a potential divider across 

 the output, from which a pre-set fraction of the output voltage is applied 

 to the grid of V^. R^ is the anode load for V^, and current flows down R^, 

 through Fi and VR, igniting VR and maintaining V^ cathode at a fixed 

 potential above earth. When the circuit is in equilibrium the setting of R^ is 

 such that Fi grid is slightly negative to its cathode. If, for some reason, 

 the output voltage tends to rise, F^ grid goes positive, increasing Fj anode 

 current and, causing F^ anode (and therefore Fg grid) to fall in potential. 

 This increases the resistance of Fg and so opposes the original tendency — 

 and vice versa. 



Clearly this device not only opposes slow changes in the output, but also 

 opposes ripple. Electronically stabilized power packs are characterized by 

 very low ripple in the output despite quite low value chokes and smoothing 

 capacitors. Ripple is further reduced by connecting a capacitor (shown 

 dotted) as in Figure 37.5, since by this means nearly all the ripple signal is 

 applied to F^ grid to produce corrective resistance changes in Fj, instead of 

 just the fraction determined by the potential divider R^R^Rn^. 



This simple electronic stabilizer works quite well, reducing the output 

 impedance to a few ohms, but, operated as it is by its own error, it can never 

 give perfect stabilization. An improvement due to Lindenhovius and 

 Rinia^ is to supplement the signals applied to Fj grid with one derived from 

 the unstabilized input, to help correct for mains fluctuations, and one 

 derived from the output current, which reduces the output impedance. 

 The first signal is applied via R^, and the second is developed across R^, to 

 give the circuit of Figure 37.6. In this manner the performance can be 

 improved but only over a rather restricted range of output current. 



588 



