and copper and they are unblemished by shame), keep the coupling capacitor. In fact, the more complex self-biased constant current source of the earlier design offers much better power supply noise rejection. One PSRR-enhancing technique is shown below.

     Much simpler. Yet, we can further reduce the parts count by eliminating the internal coupling capacitor and its biasing resistor.

     The circuit above uses an injection of power supply noise to lower the noise at the output.  The theory behind this move is that if both triodes in right side of the circuit see an identical noise signal, identical in voltage and phase, between the their cathodes and grids, then that noise will cancel at the output.
     The ratio between capacitor values sets the amount of noise injected at the bottom triode's grid. This ratio should equal the ratio of power supply noise present at the top triode's grid. At least that is what should happen with identical triodes that are identically operated.
    In this circuit, however, the bottom triode's cathode sees an unbypassed resistance, which must be unbypassed to allow the output impedance to be greatly increased, but this has the effect of unbalancing the operation of the two triodes. In other words,  we must take into account the loss of transconductance that the unbypassed cathode resistor entails in our calculations. In this example, the bottom triode's transconductance undergoes an 80% decrease.

     The cross-coupling of the two stages has come at a price, however. The cathode follower's output impedance has risen greatly and the first stage's constant current source has lost some impedance. Furthermore, the first stage is no longer AC isolated from the second stage's load. So the internal coupling capacitor is beginning to seem better after all. If your taste does not run after $200 coupling capacitors (I am not joking, they are made from the purest oil

< PREVIOUS

NEXT >