limits to amount of NFB

[pjd3]

Hello,

I was experimenting with various amounts of negative feedback in one of the amp channels in my current stereo plexi/20 watt project.

Reason I became interested in doing so was that the volume on this amp seemed to be much greater than the single channel plexi/20 watt that I use for gigging - same circuit, same tube set. The tone was also more hairy and aggressive so, I had to assume it could be a NFB amount thing.

Currently, it has a 22K ohm resistor that splits with the 5K presence pot. (From Sluckeys 6V6 plexi schematic).

And while I could live with that, I found that going as low as 10K ohms on that resistor could yield a very clean result, almost "Fendery" kind of presentation that could be good for a more transparent portal for some of my pedals.

My pondering is if there is any issue with having that much NFB (a 2 to 1 resistor ratio).

Thank you,
Best,
Phil D

[mhuss]

As long as it's stable (no persistent or intermittent oscillation), no worries.

[Helmholtz]

The more NFB you use, the more PI input signal is needed for full output.
More NFB means a higher damping factor which makes the speaker's frequency response flatter.

[pjd3]

Thanks guys for the insight and assurance.

Best,

Phil

[R.G.]

Stability and amount of NFB is a deep rabbit hole.

The quick and dirty of it is that the more you lower gain with NFB the closer it gets to oscillation for a given open loop gain.
The higher the open loop gain, the harder it is to keep it stable, no matter what. Controls guys are fond of saying that anything will oscillate if you give it enough gain.

An open loop amplifier is the least likely to oscillate. As you add negative feedback, it gets closer to the tipping point in oscillation. Some source followers and gain-of-1 opamp circuits want to oscillate at unity gain under some kinds of load.

So yeah, there is a threshold where using more NFB (that is, making the whole amp+NFB have a lower gain) can cause oscillation. Most guitar-style tube amps have enough native bandwidth so they're not balanced on a knife's edge, which is another way of saying what mhuss said - try it and as long as it doesn't oscillate, you're fine.

Keep in the back of your head that increasing the basic amplifier open loop gain can get you to oscillation at the same NFB amount; and for the same open loop gain, using more NFB to lower the amplifier+NFB overall gain can also inch you closer to the edge.

Tube power amps have a built-in limit on the total gain they can stand if they use NFB. The output transformer contains two of the three "poles" or open loop frequency rolloffs needed to get to oscillation-enabling -180 phase shift at some high frequency, so the amount of NFB and/or the open loop gain has limits.

[nuke]

Yeah, that's the key, phase shift.

An ideal op-amp has "infinite gain" and can handle 100% of negative feedback. At least the ones in the Unobtainable Amplifier product line from Miracle Silicon Inc. That creates a unity gain buffer amplifier with infinite bandwidth with no distortion at all.

We get all kinds of phase shift in tube power amps, so, there are real life limits to how much negative feedback you can get.

One can slog through all kinds of math to calculate it, and still get it wrong (at least, that's what happens when I try).

Or empirically look to see what others have done before us in similar amplifiers to whatever you're working on.

Or just set it up and change component values and see what happens.

[pjd3]

Thank you guys,

As usual, I am left with a handful of relevant things to assimilate. But I don't think I will apply so much NFB that I'll propose a problem.

Just for the heck of it, I turned the "parallel piggy back potentiometer" almost all the way to the end of its throw which for a very brief moment surely went near zero resistance. I expected to force the amp into oscillation but, didn't think about what the frequency of that oscillation would be. But I didn't hear anything, just the sound of my guitar becoming almost faint nearing the end of the pots throw. I just had to see if I could create any momentary calamity but, not sure if I did. Did nor hear anything abnormal.

So, I'm feeling confident I can find a good level for the application without getting the amp in trouble.

Question: when you speak of the amp going into oscillation, can the frequency of that oscillation occur over a very wide range? As in anywhere from 20Hz to 200Khz or more? Reason I ask is that I've mentally tried to make it a point to scope out the output of any amp I build over a wide range of frequency just in case there was a super-sonic oscillation that I wasn't pickup up with my ears.

Thank you for all your great input.
best,
Phil D pjd3

[R.G.]

Question: when you speak of the amp going into oscillation, can the frequency of that oscillation occur over a very wide range? As in anywhere from 20Hz to 200Khz or more? Reason I ask is that I've mentally tried to make it a point to scope out the output of any amp I build over a wide range of frequency just in case there was a super-sonic oscillation that I wasn't pickup up with my ears.

Self oscillation from phase shift with NFB can happen any time there is a combination of phase shifts that equal 180 degrees on top of the nominal 180 degrees of an inverting amp. In general, this only happens two places - at the very bottom of the frequency range, where be call it "motorboating" from the distinctive put-put-put-put sound, and at the high end. The higher frequencies are where every capacitor or inductor frequency cutting reaches its eventually ever-so-close-to-90-degrees shift. It can be anywhere above whatever we call the midrange, audio band, ultrasonic, or low RF even. One early problem with MOSFETs was that they have such a high frequency response that they often didn't oscillate until hundreds of MHz, often above the bandwidth of 20MHz o'scopes.

Guitar amps usually can't get too high. The limits of their output transformers keep them from it. But you can get up into 100's of kHz. Usually a modest 20MHz scope (like mine!!) will see it if it's singing. So if you can't hear it put-put-putting and you scope doesn't show anything, you're probably fine.

Mysterious situations are often a tip-off when high ultrasonics and low-RF is happening. The high frequencies eat a lot of power, which makes for more hum in whatever audio that can get through. Sometimes no audio can get through because the amp is busy singing so hard. Causes a lot of output stuff to get hot too. Big power supply ripple when there is "no audio". Things like that.

[Roe]

Normally, the limit is around 15-18 % NFB