In all the push-pull topologies, the goal is the same: to provide an equal drive signal for the output devices. This goal is important, as it ensures a low distortion output signal by forcing each output device to work equally into the load impedance, a task required for low distortion operation. But even when we have labored to ensure an equal drive signal, we may still find the output signal tainted with noise from the power supply.
Don't all push-pull output stages reject the power supply noise from their output? Some come close, but not all push-pull output stages see a balanced load impedance or an equal share of power supply noise or are run in true Class-A mode. This last point is almost never mentioned, but should be, as the push-pull's two output devices must be conducting current in order for the noise to drop out of the output signal. While at idle, both output devices of a Class-AB amplifier certainly do draw some current, so the cancellation of power supply noise is great. But once any one device stops conducting, it can no longer work to cancel the noise. Unfortunately, as amplifiers are only tested for noise at idle, the real dynamic noise rejection figure is never measured.
Therefore do not conclude that if an amplifier is push-pull, it must already be noise rejection optimized. Thus, our added goal is to understand the transmission mechanisms used by power supply noise to make its way to the output.
The techniques outlined in this article are also applicable to pure tube OTL amplifiers or hybrid amplifiers. As a consequence, last issue's use of a schematic symbol made up of a box with a capital "N" in it will be retained to illustrate the universality of the techniques. Once again, this symbol denotes either a triode, or a pentode, or a N-channel MOSFET, or a NPN transistor. All these devices are functionally identical in that they can be used as output devices in the six basic output topologies.