Re the Quad 50D/E “industrial” amplifier, well one could say that it was an
honorary valve amplifier, in that it had a valve amplifier topology – an input stage with the main feedback loop returned to its emitter, a phase splitter, a two-transistor driver for each side (Williamson-like, with drivers after the phase-splitter), identical output stages, with the transformer primary doing the recombination. A separate secondary winding (a Baxandall idea from the late 1940s) was used for feedback, but that looks to have been because an isolated output was required for driving 100-volt lines, etc.
In the Quad case, the transformer was “integral”, whereas in the McIntosh case it was an impedance and power matching device outside of the amplifier proper, which [the amplifier] would have been capable of directly driving a load without the (auto)transformer.
There is some more on the McIntosh transistor amplifier output transformer here:
http://www.roger-russell.com/mcintosh1.htm; scroll down to “Autoformer”.
It is curious that both Quad and McIntosh built transistor amplifiers with output transformers, and that both had embraced partial cathode loading in the valve era.
Where output transformers were used in small transistor amplifiers (non-hi fi) in radio receivers and record players, my understanding is that it was not normal to take feedback (if used at all) from the output transformer secondary, but rather from one or other of the output transistors. So transistor amplifiers with output transformers within the feedback loop might have been quite scarce overall.
The original Lin quasi-complementary circuit (Electronics, 1956 September, p.156ff.) did not use an output transformer. The earliest reference I can find to the fully complementary circuit is in Electronics, 1953 September, p.140ff, and the audio version of that did not have an output transformer.
(Those Electronics magazine articles are available here:
https://www.americanradiohistory.com...aster_Page.htm.
Cheers,