[Radio Wrangler]

Taking a careful look at that amplifier, you'll see that it's a peculiar design. Instead of the power devices being used as emitter-followers, or source followers with unity voltage gain after a voltage amplifier with large swing, this amplifier is a Blomley type, originated by Peter Blomley in Wireless world in the 70's. Peter Blomley only had plain old bipolars to play with in his output stage, but this follows his principle of using q4 and Q1 to phase split and switch the signal between positive and negative halves of the amp. The output transistors are used as common-source amplifiers with appreciable voltage gain.

This circuit has a number of attractions, but it has a nasty disadvantage. The output transistors try to control the CURRENT into the load, not the voltage across it. The voltage gain is strongly influenced by the impedance the speaker presents and so the speaker impedance strongly influences the stability of the feedback loop.

The thing can only be designed to have reasonable stability over a limited range of speaker impedance. With wider ranging impedances, the amplifiers dynamics change a lot and you either have to have less feedback than you'd like, or your amplifier can oscillate into some speaker. R28/C17 is an attenmpt to provide loading at high frequencies.

In the late 70's I built an amplifier to this scheme and got to fall into all its pitfalls. When things were favourable, it was enticingly good.

I think this circuit has deliberately little feedback to keep it tame.

One advantage is that they ar less prone to the firecracker sequence of destroying transistors that the tradition (Lin/Tobey/Dinsdale) architecture suffers from.

David