What exactly happens in tube/transformer with speaker load?

[Steben]

I mean: In a speakers imp graph there is a 75-90Hz impedance hump and rising impedance towards treble (1-10kHz).
Thus at the resonance and towards treble we have higher impedance than the amp is suited for. But what happens exactly?
Less gain or lower headroom?

[Steben]

EDIT:

As far as I know now, it seems the amp+transf boost the frequencies with higher impedance. So there is a "natural" tone control - IF the amp has low or no negative feedback! -
As it seems, some of the sweet overdrive classic tube amps offer comes a bit from this equalization, since the treble and the low bass resonance are more easy pushed into clipping instead of the mids.

[R.G.]

It's a matter of impedance - that of the amp versus the speaker. This is where the term "damping ratio" comes from.

Modern speakers tend to respond to the voltage across them. The current drawn for a given voltage is then whatever the impedance curve lets flow. The amplifier also has an output impedance, which looks like an R-L-C network in series with a perfect voltage source inside the amp. Modern amps are designed to match modern speakers and have a very low impedance; the ideal would be a zero impedance.

The damping ratio is the ratio of the speaker impedance to the amplifier impedance. Modern SS amps have internal impedances so low that the wire resistance on the way to the speaker may lower the damping from 400-800 down to "only" 200-300. The higher the damping ratio, the lower the impedance of the amp and the tighter the "grip" of the amplifier on the speaker. Modern SS amps are so good at this that it's irrelevant; damping ratio was invented back in tube days, where the difference in a damping ratio of 1 and 5 meant something.

(bear with me, we're getting there)

Tube amps have higher output impedances. The output impedance of a pentode is quite high indeed. It approaches a constant current source, which would put out a constant current at any voltage necessary. The output transformer acts like a block an tackle to convert this high impedance down to something lower. From there, negative feedback lowers it even more to give the amp a better "grip" on the speaker. A speaker fed a current sets its own voltage by its internal impedance, so the speaker does what it wants to, un-damped by the amplifier, including ringing on transients. Low impedance amplifiers control the speaker better so it follows the audio signal better.

If you reduce damping, the speaker audio output tends to follow the speaker's impedance curve, with a flabby bass hump at the speaker resonance and a raised low-treble response up to the point that the speaker mechanics can't follow the treble motion of the signal very well any more. Above that the speaker has the ragged treble response humps you see in a speaker response curve, but exaggerated, as the mechanical resonances on the way down cut in.

It's not so much a natural tone control as it is listening to the speaker in the sense that one listens to an acoustic guitar - good or bad - because the amplifier plays much less a roll in the sound. The bass hump in the speaker is from speaker mechanical resonance, so it does not depend so much on the amplifier output voltage to drive it. I'd have to look carefully at the amp to see whether it's clipping on these high voltages, or whether the speaker resonance was driving the higher voltages back onto the plates and the plates simply ignoring them.

I think that over all, the result is primarily that any distortion the amp happens to have is not increased or reduced by the speaker loading, but that the distortion is filtered through the speaker's now-exaggerated natural response, including the enclosure's effect on the speaker's acoustics. So I don't think that the question of less/more gain/headroom has any real meaning. As a more constant current source, the amp ignores the speaker more.