Then why did Gordon Holt continually describe "tube sound" as "bright and forward"?
Its origin is the same as most other nonsense in the audio world- evaluation without even a shred of basic controls.
With modern software, it's easy to test things like your hypothesis about 7th harmonic (funny, Curl always rants about 9th harmonic!). Putting aside that I can't find an example of a solid state amp where 7th or 9th is significant or even within 20dB of 2nd or 3rd, you could take this past anecdote and show that you can distinguish music unaltered vs added (say) 0.01% 7th. Are you willing to try that?
Also, since I don't have one to put on the bench, could you show the distortion spectra of an Atma-Sphere amp at bass, midband, and treble, perhaps at 1W and near the power limit?
Regarding Gordon, I've no idea! He never said anything like that to me- he did comment to me that he liked the presentation of our amps (that was back in the early 90s when our amps had about 3x more distortion...). FWIW I learned that bit about the 7th harmonic being in the Radiotron Designer's Handbook from one of Curl's talks posted online. I had a copy of the book and looked it up- and sure enough it was there. However the 9th is certainly rant-worthy (I can rant about that one if you like
); for that matter any of the higher harmonics are simply not helpful! And yes, with modern software you can see this stuff pretty easily. I had an SET guy challenge me about some of my rhetoric regarding the harmonics of SETs and he showed me an ancient text from Olson. It was an old chart looking like it was from the 60s or some such and I had to break it to him that there was a lot more going on than that chart was showing!
I doubt I'd be able to hear much with just one harmonic- but I'm talking about a whole mess of them- the 5th all the way up to at least the 17th or so.
The last time we did a measurement it did not include spectra but its the same at all frequencies on the OTLs (they have power bandwidth well past 100KHz; only -3dB at 200KHz). The primary distortion product (if the test equipment isn't grounding one of the output terminals, which is a common measurement error) tends to be the 3rd harmonic. The 2nd is suppressed by the fact that the amps are fully differential from input to output. The distortion falls off a bit faster than in an SET on this account, since less distortion is compounded from stage to stage (owing to even orders being cancelled) and there's only one stage of gain. If you ground one of the speaker terminals the drive to the power tubes is essentially unbalanced at that point and the 2nd harmonic shows up- at that point the distortion spectra looks like a typical SET (and the amp won't make the specified power either).
We spec the amps at 120V at the IEC but to really do that right you have to pull the cover off and actually measure the voltage at the IEC at full power; the amps have
considerable draw and there is often a voltage drop across the variac and AC cord. So you have to run the amp up to clipping, boost the AC power, run it up to clipping again and keep doing that until its actually 120V while the amp is at full power. That nuance has shot down some of the measurements made in the past.
Honestly I'm really happy to have our class D project finally bearing fruit. There can be so many variables with tube condition, tube matching, line voltage, speaker impedance and the like, all of which can be a real headache. With the class D amp you just turn it on and it works. And you don't have to worry about most of my rant material.