Direct Current Supply Line Ripple Rejection
Avoidance of the intrusion of AC ripple or other unwanted signal components from the DC supply rails can be helped in two ways: by the use of voltage regulator circuitry to maintain these rails at a constant voltage or by choosing the design of the amplifier circuitry that is used so that there is a measure of inherent supply line signal rejection. In a typical audio power amplifier, there will be very little signal intrusion from the +ve supply line through the constant current source, Q6 and Q7, because this has a very high output impedance in comparison with the emitter impedance of Q1 and Q2, so any AC ripple passing down this path would be very highly attenuated. However, there would be no attenuation of rubbish entering the signal line via R5, so that, in a real-life amplifier, R5 would invariably be replaced by another constant current source, such as that arranged around Q7 and Q8.
For the negative supply rail, the cascode connection of Q10 would give this device an exceedingly high output impedance, so any signal entering via this path would be very heavily attenuated by the inevitable load impedance of the amplifier. Similarly, the output impedance of the cascode-connected transistors Q3 and Q4 would be so high that the voltage developed across the current mirror (Q5 and Q6) would be virtually independent of any -ve rail ripple voltage. In general, the techniques employed to avoid supply line intrusion are to use circuits with high output impedances wherever a connection must be made to the supply line rails. In order of effectiveness, these would be a cascode- connected field-effect transistor or bipolar device, a constant current source, a current mirror, or a decoupled output, such as a bootstrapped load. HT line decoupling, by means of an LF choke or a resistor and a shunt-connected capacitor, such as R2 and C2, was widely used in valve amplifier circuitry, mainly because there were few other options available to the designer. Such an arrangement is still a useful possibility if the current flow is low enough for the value of R2 to be high and if the supply voltage is high enough for the voltage drop across this component to be unimportant. It still suffers from the snag that its effectiveness decreases at low frequencies where the shunt impedance of C2 begins to increase.