Paul Klipsch thought that harmonic distortion was a minor issue while intermodulation distortion was a major problem because most music contains harmonics but intermodulation introduces nonmusical frequencies.
Paul Klipsch was right, it's one of the main reasons why Jack E - spends so much time eliminating IMD from his designs often using transfomers instead of active components.
An auto former for phase inversion is an absolutely briliant idea.
All the stuff on his site had now gone, so luckily I archived it while it was still there.
We had many interesting conversations one of course was related to doing IMD tests with a DAW, then real time FFT to show the IMD waveforms, fed back via a voltage divider directly from dummy load or speakers..... which for various reasons he couldn't accept was valid, who knows why!
ie. these methods...
(The smart thing about it, being setting bias and anode/screen load/NFB to null out IMD - and as he suggested avoiding Beam tetrodes at all costs).
The other great thing about DAW (pro sound card digital audio workstations) is the low noise/distortion - with resistive voltage dividing. eg. I routinely use a digigram card in a notebook (AKM DAC+ADC) with a program called "scope".
U can compare input waveform into the output valves with the output from the transformer.
I used this method on an ancient cathode follower/choke driven AB2 amp by substituting a Bogen 6V6 output transformer for the choke then reading the input waveform off the secondary.... against the output waveform from the OPT. (superposing the 2 traces, recording them on the DAW and doing FFT on each channel to determine which bit of the chain was responsible for which bit of the IMD....)
This enabled aboslute optimisation of NFB, valve matching and balance.
It was nothing close to recommended values, or the manufacturers figures...