That is well and good, but pointless. People use such a circuit to get a heavier distortion with the same input signal, not to get light distortion with a smaller signal.

Let's put it this way: if you have high gain metal distortion circuit gained up to a 5mV clipping threshold, and then you roll down your guitar volume knob to play the blues, you're kinda using it wrong.

It's pointless in the case that you mention, where you're going for a hard clipped metal distortion tone. I was envisioning more of an overdrive sort of application where hitting the rails is NOT the desired effect. The 5mV hard clipper was an extreme example, but you could conceivably use it for soft overdrive/compression with a small enough input signal which was what I had in mind.

If hitting the rails is not desired, then don't configure the effect to have so much gain.

The point still stands that you can get the distortion effect, at the desired amount, at various places in the signal chain.

The problem you're talking about is about using too weak a signal and then using make-up gain. That can also happen anywhere in the chain.

I can reproduce the problem without distortion at all, using a clean signal chain.

Repro 1: turn down the guitar volume to simulate an extremely weak pickup. Then turn up the input gain to compensate. Result: noticeable increase in noise.

Repro 2: turn down pre-amp volume, and crank up power amplifier volume to compensate. Result: noticeable increase in noise.

Gains/levels have to be evenly distributed in the chain.

Sure, you will lose headroom if your intent is to have some soft distortion from the back-end of the amplifier, but you put a heavier distortion earlier in the signal chain. The heavy distortion masks the later one; it evens out the level so the headroom from the back is not available. There is a fix for that: you can punch that distortion in and out with your foot as needed.

are SNR issues like that and the headroom an issue for a guitar amp? I mean why couldn't the solid state part be designed conservatively with extra headroom? Or do the desirable distortion products make that difficult even with a conservative design in later stages?

Generally you want to avoid noise early on in the gain chain just because it gets amplified later on. Headroom is entirely relative, you just need to design your gain stages such that the power amp isn't railing when you turn your guitar up halfway.

Note that your two points are contradictory:

- avoid noise early on in the change

- turn your guitar up halfway

If you want to avoid noise early in the gain chain, do not rob your guitar of signal by attenuating the output of the pickups with the guitar's volume. Now though that is done, by some players (and works for them), it isn't a way to get headroom or minimize noise.

You get the least amount of noise by keeping the volume of the guitar at maximum, and switching tones later in the chain. A clean pre-amp channel has much better noise characteristics than a dirty channel, and rolled down guitar volume control.

The point I was making about turning the guitar up halfway was that you need to design amplifier gain stages so that there is not so much gain that the amplifier hits the rails with a reasonable input signal. E.g. If your preamp stage has a voltage gain of 100 and your expected input signal is 1v p-p but you only have +/-20v rails for the power amp stage, you objectively don't have enough headroom.

I see. Intersting. How do you think about the desirable distortion characteristics vs bad "noise"? Is there an industry standard definition for guitar amps?

Seems like the bad noise in the case we're discussing would be what the tube circuit generated when it was not being intentionally driven to distortion... is that a correct interpretation?

Noise is anything that is still present when you take away the desired input signal, excluding deliberate time-based effects like reverbs and delays.

Frequency components that are not in the original signal, but are only present when the signal is there are usually distortion.

(Why "usually": radio-frequency interference is noise. In audio, you may not notice RF when there is no signal. But when signal is present, it may mix with RF in a nonlinear way, causing additional distortion products attributable to the RF. I.e. the RF noise is "descends" into the signal's frequency range via intermodulation.)

Noise would be thermal noise generated in circuit components, or mains hum, anything that is there when you aren't playing any notes. I wasn't spraking exclusively about guitar amps, all this applies to analog electronics in general.