Hi Scott,
Very good question, and I am impressed that Maty, being a Spaniard, is being a mild Inquisitor, and I'm grateful...... I have been preaching for twenty years that distortion should never be called so; it is harmonic profile, and accepting that 'distortion' is inevitable - impedance is futile after all - we need to work with it. Actually, Maty is onto me; he understands all my tricks now. I take my hat off to you, Maty!
I have spent a long time profiling harmonics from SS amps and this one has one of the closest to monotonic decrease that I have ever seen, short of a single ended triode.
So YES, and I love your photography analogy, it is apt and I use it too with my photographs of my amps! Saturation is indeed H2! There are ways of changing the ratios of H2/H3/H4 and they are perverse. The singleton input stage is one of the best, and I used this in the NAKSA, though in the SAKSA I used this still first and second stage from the ALPHA. Reason? It gives better listening sessions with BAD recordings; the NAKSA was somewhat unforgiving. Normally a LTP followed by a Lender connected VAS has H2 pretty much equal with H3, and low, but if you put a diode like the ALPHA left LTP leg you throw more current drive onto the emitter, whilst giving less voltage drive on the base. The effect is to drive the npn VAS in a hybrid of common emitter and common base; this speeds it considerably and brings up the high pole, making compensation benign and easy to achieve. The result of this is much less H5 and beyond, which improves the listening experience with bad recordings, and of course we all know that about 80% of our recordings are BAD, not just M. Jackson, which is actually recorded very nicely with lots of H2 from Quincy Jones.
So, to answer your question, yes. We can put in a pot in place of R17, the resistor in series with the 220uF on the CCS. At the 560R level, the dynamic current swings of the upper and the lower devices are close to IDENTICAL, which gives you H2 at -70dB and H3 at -79dB for +20dB output (14.14Vp into 8R; H5 is at -111dB). If you reduce the R17 to 330, the currents swings are LESS on the pmos (1.2A-1.85A) than on the nmos (0.1A-3A), and H2 is -69dB and H3 is -83dB, H5 is still at -111dB. This change of R17 does change the ratio of H2 and H3 but I prefer to set the current swings equal so you can maximise the power output.
So, you can change H2 but changing the Pass CCS; by changing the input and VAS topology; and by using asymmetrical source resistors on the output stage of a PP Class AB amp. A 'niceness' knob is quite difficult to arrange however, and if you did, many purists would say that it's not designed correctly because it increases the distortion.
In decades in the future, people will come to understand this is an acceptable way to make a sound better than the 0.0005% monsters we used to buy in the eighties.
HD