Questions re:  GaNfet technology vs other designs.

Read my words carefully. "MAINLY" comes from a non-flat frequency response. That is especially true of most tube amplifiers such as the high output impedance of your own OTL amplifiers.

@theaudioamp Our OTLs are intentionally designed to work without feedback and so do not operate on the principle of voltage driven loudspeakers. Like SETs, you have to be careful of the loudspeaker used to get proper response that isn’t colored. See http://www.atma-sphere.com/en/resources-paradigms-in-amplifier-design.html for more information. This is done with intention since its impossible to apply the kind of feedback you need to really get rid of distortion that is otherwise generated by the application of feedback itself (this occurs because the feedback node is always non-linear, which means that the feedback signal itself gets distorted before it can do its job). As Norman Crowhurst pointed out decades ago this causes generation of higher ordered harmonics. You need +30dB to get around that problem and since tubes won’t have the gain bandwidth product to allow for 30dB at all audio frequencies its a Sisyphean task. So zero feedback is how the harshness of feedback is avoided since in a tube amp you still have pretty good linearity if you’re careful.

So the OTLs were not a good example. If the amp has enough feedback or else a low enough output impedance to act as a voltage source then you really have to ask yourself why one can sound bright while another does not- because if its acting as a voltage source there is inherently no frequency response variation.

If you’ve not found any information about why the higher ordered harmonics are so audible when in such small amounts, its easy to demonstrate with simple test equipment. IOW you’ve not looked all that hard. The ear uses the higher orders to sense sound pressure. If there are higher ordered harmonics added to a signal, they will be perceived by the ear in two ways: louder, and also harsher. Because of ’louder’ that also means brighter (and Mr. Fletcher and Mr. Munson don’t help). If you need to know the procedure to demonstrate this to yourself let me know.

The ear/brain system has a variety of tipping points in its perception of sound. Since it assigns a tonality to all forms of distortion (the ’warmth’ of tubes for example being caused by the 2nd and 3rd harmonic) it will pay more attention to this sort of tonality than it will actual FR errors. You might be looking in the wrong place if you go to ASR and Stereophile; as best I can make out they don’t seem to have made the connection (yet) between measurements and audibility.

OTLs are an excellent example as they will have a totally colored response compared to probably virtually any other amplifier type with any given speaker. You can choose one that works best, but no matter what you do, it will sound different with other speakers. A quick review of your posts shows you pairing it with speakers that were not intended to be used as such.

Your interpretation of how the ear perceives harmonics is too simplistic to be useful as a rule. Here is a link to help: https://www.researchgate.net/publication/323914860_THE_EFFECTS_OF_DISTORTION_ON_THE_PERCEPTION_OF_LOUDNESS_IN_LIVE_SOUND Low feedback will result in mainly compressive distortion. Insufficient feedback will result in expansive distortion using their defined parlance.

W.r.t being easy to demonstrate with simple equipment, I agree. I low distortion amp, low distortion amplifier (across 20KHz), and low distortion headphones. Then use a distortion simulation program like Distorter (VST plug in) to simulate various level (and orders) of distortion, or Pkanes Distort. Sure if you did 0.1% distortion of purely higher order for a 500Hz tone, you will hear it before lower order distortion but you and I know that is not a realistic occurrence. And this is detecting with a pure sine tone at optimal frequencies. With real music, no way.

Take a perfectly digital source and low distortion SS amp. Then add in the distortion profile of a typical SET amp especially the 2nd/3rd distortion, at levels one could expect outside clipping. Then blind A/B them with music. It won’t sound warm or cold. On general music you will notice the distortion before you notice any warming ... and yes, I encourage everyone to repeat this experiment themselves if they have a low distortion DAC/AMP and preferably headphones as lower distortion but you could use speakers. Please don’t take my word for it. That whole "warming" due to typical distortion components of tubes is just Philelore.

Yes the brain does have a variety of tipping points, but it also has the ability to extract the underlying information out till things get really bad. It will not pay attention to this distortion tonality more than it will the raw frequency response.

So where should I be looking for this connection between measurements and audibility? Feel free to send a link with some sort of objective listening test of distortion (and reasonable levels) versus tonality, where the only variable changed is distortion. I don't know how you cannot find information explaining why higher order distortion is more audible. This is a readily known thing. Higher order harmonics will not be masked by the primary tone. That is for tones. They will be masked in music.  For pure tones, Fletcher-Munson comes in again, because at low frequencies, the higher order harmonics will be areas of higher sensitivity.  But SS amps typically don't have much distortion at those frequencies. Keep in mind with pure tones, the masked harmonics will themselves mask the higher orders for typical distortion profiles (to a point). That coupled with music again limits the practical audibility of higher order harmonics of real equipment. Philelore.

Regarding the issue of how distortion spectra affect listening preferences, there's a great interview with Nelson Pass by Steve Guttenberg on YT from 2019. It's called "Nelson Pass on his favorite amps, playing with distortion, and his 21 inch subwoofer." Especially relevant discussion starts at about the 10:17 mark in the video. 

Nelson emphasizes that even at very low overall distortion levels in a solid-state amp, listeners tend to prefer more 2nd or 3rd order distortion compared to higher order distortion. He also describes how the sound changes as phase and distortion are varied.

These issues have relevance for the design of high-quality GaN FET amps, as @atmasphere discussed earlier in this thread.   

@sdl4, no one is discounting that higher order distortion is more audible than higher order harmonics. This is not in question and the mechanism is very well understood. It is due to masking functions and the shape of our sensitivity to different frequencies.

Nelson's ideas of very low distortion is probably not what we think of as very low distortion, as well, Nelson discusses feedback in SS state in other papers and how that translates to distortion over frequency and this relates to the claim, I consider false, by Atmasphere that many SS amplifiers with negative feedback suffer from too little feedback at high frequencies. This is a flawed premise on several fronts:

1) This is just not factually true based on an easily done review of tests by Stereophile and Audioscience where even relatively inexpensive amplifiers have low distortion out to 20KHz, and this appears to have been the case for many years.

2) Because of the upward curve of hearing sensitivity, which reduces after about 3.5Khz, distortion at high frequencies becomes harder and harder to hear. If you think about it, at 3.5KHz, the 4th harmonic is 14KHz. Many people here can't even hear that. At 5KHz, the 4th harmonic is 20KHz, inaudible. Where higher harmonics are most audible is at low frequencies where higher harmonics are where our hearing is even more sensitive than the fundamental. 100-1Khz is a prime area. However, all those SS amplifiers with claimed insufficient feedback have tons of feedback at <1KHz (if they have Negative feedback).

This whole amplifiers that measure well and sound poor is nothing but Philelore, started by amplifier companies that make amps that measure poor and then chanted by their customers. If you have an amplifier that measures well, with today's ready suite of measurements, and sounds bad, it is not the amplifier that is the issue, it is your system. If a poor measuring amplifier sounds better, then it is just its poor transfer function complimenting the rest of the flaws in your system. It is one way to achieve a desired end result, but it is 2022 now. There are better ways if audiophiles would just get out of their own way.

In terms of phase from an amplifier, there are readily available metrics for what is audible in phase change. Most modern amplifiers are so far below audible limits w.r.t. phase changes that any discussion of it is a total waste of time. However, go back to those bad measuring amps that sound good, and perhaps you are complimenting a flaw in your system (that should not be there).

Last, keep in mind, that many of these claims of distortion, at the levels of the amps in question causing tonal changes, has never, what 50 years later, be shown to be fully factual even though experiments are readily done. I think most would be rather surprise how hard it is to detect distortion when listening to music. Far far harder than with tones, even when you use higher order distortion products. What is needed is far higher than one would ever expect to see in an amplifier today.

Nelson Pass obviously has a huge following in audio, mainly due to his strong involvement in DIY audio. He is a far more public figure than almost any other audio amplifier designer. However, I suspect few professional amplifier designers (and not just hifi) have spent much time studying his work, where they have spent a ton of time studying Doug Self, or a whole host of people people who listen to audio have never heard of.

I'm thinking that "philelore" is the new "fetch."