Yes Stanwal, I did mean the power amp - my dyslexia! Still, what was (is) meant by 'current-dumping'?
"Current-dumping" is Peter Walker's terminology, it refers to the power side of what is essentially a feedforward amplifier. In this topology, there are actually two amplifiers summed together in the output, one being a low-power Class A amp, and the other being a high power push-pull amp with virtually no bias (even less than Class B, sort of a symmetrical Class C). Here the high-power "current dumpers" send the vast majority of the current to the loudspeakers, and the Class A amp supplies current opposite to the inherent non-linearity in the current dumpers.
It's actually quite a brilliant circuit, typical of Peter Walker's elegant engineering approach . . . and it definately made sense with the power semiconductors of the time, which had linearity problems all across their range of operation. Modern power transistors however can be extremely linear in Class B right up to the crossover region, so it may be that the power-dissipation spent in a separate Class A amplifier in a feedforward design, may be better spent simply applying bias to the output stage itself, maybe in combination with lower-value emitter resistors.
Another "hybrid" Class A approach that holds promise is Class G (aka Class H), with the inside (main) amp biased into Class A. This might be thought of as a symmetrical Class C amp wrapped around a Class A amp (inside a dream within a dream, within a Taco Bell, inside a KFC!) Seriously, though, the problem with glitches in Class G/H amps in Walker's era are solved today with better switching diodes, which may also make the feed-forward approach moot in modern times.
