To add my $0.02 to Peter’s excellent explanation, my understanding is that there are two schools of amplifier power-supply design:
(1) Use a power supply with an excellent regulation to start with, so that the power-rail voltage won’t “sag” too much even when all channels are fully driven. If the power supply maintains, say, 90% of the voltage under full load, then there will be less benefits of using a bulkier, higher-capacitance power capacitors.
(2) Don’t care too much about the power-supply regulation, and instead, supplement the power supply voltage under the full-load transients during music playback by the discharge from big power capacitors. This is so-called the “crest factor” school of amplifier design.
For example, the earlier generations of NAD amps (don’t know about the current models) and the current Sony ES receivers belong to (2) -- they can deliver a much higher transient power for a short time period (1 second or so) than what they can deliver steady-state. In contrast, according to the HiFi News (UK) measurements, the Pioneer Elite 45TX clearly belong to (1) above -- its power supply does not give in too much even when fully loaded steady-state.
Both schools of design are valid, and in reality, most practical designs belong to somewhere in-between. But my understanding is that the high-regulation power transformers have come a long way, so that today it is often easier and less costly (both in terms of money and space) to go with (1). That is why we don’t see anymore, even in the high-end separate amps, those HUGE capacitors that used to be found often in older amps.
