I'm in WAY over my head...

[mooreamps]

So that should give it a wider/higher/flatter frequency response ????
Less growl in the midrange ??? If so, for a guitar amp, better ? worse ?????
or is something like this too much "Hi-Fi" for a guitar amp ???

-g

[J M Fahey]

I guess the answer is "all of the above"
Besides, the higher the NFB, the sharper will be the square wave edges at clipping, making it closer to an SS amp than the typical "rough" tube amp we are used to.
Of which many respected examples just use no NFB *at all* , go figure.

[Chuck H]

I guess, just like OT undersized/oversized, plate voltage high/low, sag lots/less, It's part of a fomula that each amp uses. There is no right or wrong. Only the combinations that seem to work together. For example, two notorious classic designs for overdriving the power tubes would be the 100W Marshalls and the Vox AC30's. Many, if not most 100W Marshalls use considerably MORE feedback than most other guitar amps. The Vox AC30's use none. And I'll wager that an AC30 with large amounts of feedback would sound crummy. As would a 100W Marshall with none. It's just one ingredient in a bigger design.

[Enzo]

Good wager.

Sort of a corollary to my "it ain't the parts it's the circuit" philosophy.

[mooreamps]

I guess the answer is "all of the above"
Besides, the higher the NFB, the sharper will be the square wave edges at clipping, making it closer to an SS amp than the typical "rough" tube amp we are used to.
Of which many respected examples just use no NFB *at all* , go figure.

Fine.. so just put a knob on the NFB, and then call it a day.... "Let the player figure it out"..

-g

[Steve Conner]

That's more or less what "presence" and "resonance" controls are.

Another thing about the Williamson is, it has large interstage coupling caps for a hi-fi frequency response. Large caps + lots of global NFB = really dreadful blocking distortion when overdriven, for reasons explained by Morgan Jones in his hi-fi books.

Generally, the more NFB, the worse amplifiers behave when overdriven. This isn't necessarily true for solid-state amps, but it seems to be hard to avoid it in practice.

[J M Fahey]

@ mooreamps
Not a bad idea at all
The amp can go from tight, sharp and glassy to woofy and loose.
Obviously adjustable for very different styles of playing.
@steve conner.
Yes, Lin type SS amps (99.99% of those built today) can´t avoid being high NFB type, because of the monster "Class A" gain stage they include, but the NFB can be configured with two nets: the "normal" voltage one, an extra current one, and a pot to smoothly go from one to the other.
I have seen Hi Fi amps with a "damping" control which does exactly that.
And typical guitar amps usually have both, only they are mixed in a fixed percentage, not user adjustable.
Although Valvestate rack amps and some Polytones had a switch to add/bypass current feedback at will.

[Chuck H]

Other than presence controls "I" (personally) am not a fan of variable feedback. To me it seems like there's an amount that just seems right for a given amp design. Reduce it and the bottom end goes south. Increase it and the amp gets bland. But it's sort of like seasoning your food isn't it? There is an amount of salt that the cook will use that's right for most diners. But they still put salt on the table. I think what puts "me" off about a variable feedback control is that they're generally very wide in adjustment giving the control a whole lotta useless range. Back to the salt analogy, it would be unforgivable to be served a meal where the chef used NO salt. The variable feedback control allows for that sort of thing.

[R.G.]

The amp can go from tight, sharp and glassy to woofy and loose.
Obviously adjustable for very different styles of playing.

There is a secondary issue that needs addressed if you turn NFB down.

The way NFB works is that it uses 'excess' gain to correct errors. The 'excess' gain is the difference between the open loop gain and the gain with NFB. So if you turn down NFB, you effectively release more of the open loop gain. An amp with, for instance an open loop gain of 40db might be fed back to a NFB gain of 23db, the extra 17db being used for the internal error correction while NFB is active.

If you then reduce the NFB to zero, then the full 40db of gain happens, and the amp sound not only changes from more or less 'accurate' in the technical sense, it also has a much higher actual gain. This side effect may or may not be well understood by the user, and may or may not be thought of as good. Different users will think of it different ways.

One possible way around this is to attenuate the input signal in some way to match the reduction in NFB and increase in actual gain of the amp. The apparent effect of this combined with reduced NFB if done well is just a change in the character of the amp, not its apparent gain.

Lin type SS amps (99.99% of those built today) can´t avoid being high NFB type, because of the monster "Class A" gain stage they include,

Actually, they can. It's easy enough to dial down the gain of the voltage amplifier section or input diffamp to reduce open loop gain to whatever value you want. They're just amplifiers. Getting lower gain is easy. Getting higher gain is sometimes harder.

[Alan0354]

That's exactly true, if you reduce NFB, the gain of the stage goes up. If you get rid of the NFB, you'll be running a open loop gain. One thing to add in your example of 40dB open loop gain and 23dB NFB gain. The 17dB is called loop gain, it is the overhead that you talked about. The higher the loop gain, the less distortion.

For tube amp, I think the open loop gain is so low even if you have NFB, there are still quite a bit of distortion. Anyone actually try open and close the NFB resistor and see how much gain change?

[Steve Conner]

For a typical guitar amp, it changes about a factor of 3-10 depending on the design. This modest amount of NFB probably doesn't help much with distortion, as it's in the range where Baxandall's "reentrant distortion" argument applies: it just shifts the distortion to higher orders rather than reducing the amount.

It's really more to damp the speaker by lowering the output impedance of pentode or beam tetrode power stages.

For the Williamson, the feedback factor is nearer 100, at midband at least. It has to be tapered off at low and high frequencies to avoid instability.

For comparison, a typical solid-state hi-fi amp has a feedback factor of about 20-50 at 20kHz, increasing 6dB/octave from there downwards.

@JM Fahey: Irrespective of whether the feedback is voltage or current, the more there is, the deeper the transistors get driven into saturation when the amp clips, and the longer they take to recover.

For instance, I had a lot of trouble with the voltage amp stage in my Doug Self Blameless. Self used two cascaded transistors in this stage for extra loop gain, but the second transistor gets massively overdriven on negative clipping and takes a long time to recover from saturation. Not a problem really as it's a hi-fi amp. Newer SS guitar amp circuits use things like Baker clamps or Peavey's DDT to prevent saturation altogether, but I don't think the problem was understood until recently.

[R.G.]

@JM Fahey: Irrespective of whether the feedback is voltage or current, the more there is, the deeper the transistors get driven into saturation when the amp clips, and the longer they take to recover.

...Newer SS guitar amp circuits use things like Baker clamps or Peavey's DDT to prevent saturation altogether, but I don't think the problem was understood until recently.

Yep. I was introduced to saturation limiters as a green-eared engineerling when we were trying to get 25kHz power switching out of BJTs (power MOSFETs hadn't been invented at the time!) at fractional KW levels. It was an ugly process, and letting a power switch go into deep saturation was pretty much a guarantee of a LET (Light Emitting Transistor) being generated.

Antisaturation techniques are really a Good Thing when you're handling fast signals that can saturate the outputs.

Another thing many people don't think about is not letting the VAS stage be driven into insanity when output protection, saturation, or anti-saturation circuits make the outputs quit responding.

[Steve Conner]

Another thing many people don't think about is not letting the VAS stage be driven into insanity when output protection, saturation, or anti-saturation circuits make the outputs quit responding.

I didn't think about it either until it was happening on my bench! A VAS current limiter will stop the VAS transistor from actually blowing up and hitting you in the eye, but it can still saturate and hang up for a long time.

Amazing how things change, we've gone from BJTs that struggle to switch an amp or two at 25kHz, to IGBT driven Tesla coils on the tabletop.

[mooreamps]

That's more or less what "presence" and "resonance" controls are.

for most g amps perhaps... I have a different approach for introducing "presence".

-g

[J M Fahey]

Mmhhhh , interesting.
Can you elaborate?
Thanks.