Why do we want to distortions?

A very interesting topic. I recently read a paper from Cheever (1989) that's quite interesting. Our aural distortion seem to play a role here. Both from a masking perspective, but also from a musicality and holographic soundstage perspective.

I got the opportunity to borrow a Benchmark AHB2 over a weekend some weeks ago. For listening and measurements. I have an audio measurement system that I use for development so it would be interesting to measure some on the very high performance AHB2 amp. And indeed, it has a very very low distortion. Its a real masterpiece in that sense. And by using forward error correction instead of traditional negative feedback they have reached some incredibly good numbers regarding "traditional" amplifier measurements. That's one part of the coin. The other task was to listen to it. And to me it was not musical or engaging to listen to. But of course transparent, good dynamics and with high resolution.

Papers I have read, own experiments with harmonic distortion and discussions have got me into an idea around distortion. If we add low order harmonics inline with our hearing harmonics we get a masking effect but we also, for us as humans, a natural way to increase harmonics from instruments and voices. Without affecting timbre (I think). This could maybe also explain improved holographic imaging. A least to how I think.

Sorry have issues with postings. Very sorry.  I get page error when posting. I will try to sort it out

I don't agree, it's impedance that is important for tube amps. From what I've experienced, tube amps don't like low impedance. Yes newer speakers typically have lower impedance than older ones, but tubes play well with 8 ohms or more. Are there speakers after 1965 that have 8 ohm or higher impedance, yes.

Interesting. A few years ago I made a prototype mosfet amp with adjustable output impedance (or damping factor if you will). By using adjustable current feedback. Damping factor of 20 was preferred with my speakers. I tested with DF 30 down to negative output impedance. Just to see how the sound was affected by this. But it comes down to what the speaker was design together with. I also did a spice simulation model of my speakers to be able to simulate the behaviour with respect to source impedance. Used this for some simulations of a current amplifier together with the speaker simulation model. However, my experiments using the MOSFET (class A voltage gain and push-pull output stage and some global negative feedback) did not show a strong correlation with tube sound. There where some to changes in the base as the base driver gets a bit more freedom. But this was not a scientific analysis, and it was only tested on my setup. I could not increase the DF over 30 in the design.

Both Cheever, Pass and several others agree on that to much negative feedback create high order harmonics that we are very sensitive to. Cheever also discuss aural masking and the effect by adding such distortion to the playback system. But does this masking effect also apply to to some extent to intermodulation distortion?

Intermodulations do not get masked. I agree with Cheever and Pass on this- but only to a certain extent. If the feedback is creating higher ordered harmonics there simply isn't enough of it. That is why I say you need in excess of 35dB. With that amount the amp will have the ability to clean up the distortion created by the feedback itself.