"Local" feed back

[trem]

I am going to adjust the resistor value myself and see what happens.....but I just want to ask.....in theory....can the value of this resistor have an effect on what sounds like "blocking" distortion.?
I am referring to the resistor in the Fender Bassman AB165. It is in that little feed back circuit in V2 where both channels mix. I am talking about the 470k between Pin 6 and Pin 7 on V2. Is it possible that resistor could cause distortion if it was not the right value.? What exactly is the purpose of that little circuit.?
Thank You

http://www.webphix.com/schematic%20h...b165_schem.pdf

[R.G.]

I am going to adjust the resistor value myself and see what happens.....but I just want to ask.....in theory....can the value of this resistor have an effect on what sounds like "blocking" distortion.?
I am referring to the resistor in the Fender Bassman AB165. It is in that little feed back circuit in V2 where both channels mix. I am talking about the 470k between Pin 6 and Pin 7 on V2. Is it possible that resistor could cause distortion if it was not the right value.? What exactly is the purpose of that little circuit.?

Let's do that in reverse order.

That resistor and the two 220K resistors that feed it make V2 into a negative feedback gain stage, much like an opamp fed by two resistors into its inverting input. The capacitor going to the grid blocks these resistors from having an effect on the DC conditions of the tube. V2 probably has an open loop gain (i.e.,with the 470K resistor made into an open circuit)of something like 25-35. The 470K and two input resistors make for a closed loop gain of about two for either input, and moreover, prevent the two signals from interacting because the junction of the resistors acts like a "virtual ground". In actuality, the non-infinite gain of V2 means that the effective gain for either signal is not really as big as 470K/220K, but it's good enough to get some gain and also the mixing and isolation.

That being said:
Can it cause distortion? Sure, but other things are much more likely to be the problem. It won't give you blocking distortion because blocking distortion pretty much requires a signal to push the grid into conduction. This circuit minimizes the actual voltage change on the grid and makes blocking distortion much more unlikely.

It's probably something else.

[trem]

Thank You.....
Bad choice of words on my part, blocking distortion was just a descriptive. The distortion occurs on both channels, when volume knob(s) is 4 or above, starts as the note peaks and remains as the note dies away, gets worse if you dig into the strings.
This was not the first consideration when looking for a solution to my problem. It is about the only avenue I have not pursued. It dawned on me that this is a big variance between the AA864 and AB165. So, it just made me wonder what the symptoms are of having the "wrong" value resistor in that circuit.
So.....I was wondering......in theory......is the above problem something that can be caused by that 470k being too big/too small.?
Thank You

[exclamationmark]

Distortion in a NFB looped stage can be very abrupt, as the loop practically collapses back to open-loop gain when tubes are in overdrive, making it more of an on-off type thing. Some don't like the sound of it. Too much NFB can also make the stage unstable and cause random oscillations.

[pdf64]

the (NFB) loop practically collapses back to open-loop gain when tubes are in overdrive

I've seen that stated before and probably re-iterated myself but haven't been comfortable with the rationale or what the practical effect / evidence is to back it up.
Can someone help?
Not so much with tubes, but I can't see that an op-amp flips to open loop gain whenever the output 'hits the rails'.
Sorry for minor hijack.
Thanks - Pete.

[R.G.]

I've seen that stated before and probably re-iterated myself but haven't been comfortable with the rationale or what the practical effect / evidence is to back it up.
Can someone help?
Not so much with tubes, but I can't see that an op-amp flips to open loop gain whenever the output 'hits the rails'.

Actually, it's mostly correct, but one term is wrong. The gain in a sense collapses, but not to the open loop gain. It falls from the closed loop gain towards zero, sometimes abruptly.

If you think about it, when any amplifier approaches the maximum output voltage it can produce, a bigger input can't drive the output any further. The incremental "gain" drops, as Vout/Vin cannot remain the same because Vout can't go up any more. In open loop devices, This is more or less a soft distortion entry. In tubes, gain rolls off slowly as the grid approaches cutoff because the shutdown of electron flow is not abrupt. If you have the tube biased so it goes into what would be saturation/max current in a transistor, it's also soft. (it can be razor sharp if it goes into grid conduction, but that's another story)

The feedback uses the open loop gain to "hide" the real underlying device operation. So as long as there is enough gain to let feedback cloak what's really going on, the closed loop gain appears to be unchanged. When the internal open loop gain reduces down to the point where there is no longer enough open loop gain for the feedback to hide what's going on inside, the external closed loop gain transitions abruptly to a much smaller value. The higher the internal open loop gain and more feedback there was, the more sharply the thing runs out of gain, and the harder the entry to clipping is. This is why opamps have generally flatline clipping when they clip.

In a tube with an open loop gain of a few tens, the transition to clipping is less abrupt than in an opamp with 60db of feedback, but it's not as smooth as in a no-feedback tube with only the inherent clipping as it bangs against "saturation" and cutoff.

All that being said, I'm still a bit confused by the original post describing the distortion as "blocking distortion". I do trust a tube amp guy to know blocking distortion and describe it right; Abrupt clipping of the tube isn't likely to cause real blocking distortion, as the grid drive point no longer swings as much as you would expect.

The feedback factor also divides the signal seen at the input node. If the tube has an open loop gain of 26db, and it's fed back to a closed loop gain of 6db (2x), then there is 20db of feedback, about 10x voltage gain used up in making it act externally like a gain of two. The tube doesn't know all this funny stuff is going on , it still amplifies whatever appears at its grid by 26db. The only way for its output to swing only 6db more than the input signal at the input resistors is if the signal appears at the real grid input is 26db less than the output. And that is what happens. The input resistors and feedback resistor "cancel" the output against the input, and only the amount of signal needed to drive the output to what is demanded by feedback appears at the grid.

This is -26db compared to the output, which is itself is +6db compared to the input, and so the grid signal is -20db compared to the input. The grid signal is *tiny* compared to the input signal or the output. This is why the inverting input of a feedback amplifier is said to be working as a virtual ground. The signal there is approximately zero if the open loop gain is high.

Sorry to blather on, but it's most unlikely that real blocking distortion is happening in that circuit. But it is correct that distortion is much more abrupt than an open loop tube stage.

[J M Fahey]

"Everything" is working with its full "open loop gain" if it's ... "open loop" ... (duh), meaning there is no feedback signal, that's what "open loop" means..
If you have a continuously varying input signal (say, a sinewave) feeding a linear amplifier, be it tube, transistor, Op Amp, whatever, and feed back, out of phase, a fraction (maybe even the full value) of the output signal, said amplifier will try to minimize the difference (which would be "the error/nonlinearity") lowering distortion.
But this implies the amplifier must still have *some* control on the output signal, or the very concept of feedback falls like a house of cards.
When the output hits the rails (or is clipped by back to back Zener diodes, it's the same), there is no correction possible, since the output is not "signal" any more (it's just the *fixed* rail value), there is no "feedback signal" to substract from the original varying one, and the amp gets back to open loop gain.
Other way to see it, is that the amp stage detects that the output is no longer following the input, and tries desperately to correct that, applying all its strength: "open loop" gain. (that's all it has).
Both explanations mean the same; no doubt there are other ways to say the same.
Transistor amps usually have heavy feedback (typically around 40dB) *because* they have *huge* open loop gain (over 60dB) , that explains their dreaded (for guitar) sound: sharp buzzy clipping, harmonics killing flat top squarewaves, high damping, etc.
Tube amps , besides other characteristics, have little (6/12dB) to no feedback.
When clipping, there's no great (or none at all) gain change, no buzz-enhancing effects.

EDIT: whoa !! simulposting again !!!
Anyway, we are agreeing, (as usual), although using different words to say so.

[trem]

No Problem.....
A little "hi-jack" is good. I have learned a lot from other peoples questions.

I've seen that stated before and probably re-iterated myself but haven't been comfortable with the rationale or what the practical effect / evidence is to back it up.
Can someone help?
Not so much with tubes, but I can't see that an op-amp flips to open loop gain whenever the output 'hits the rails'.
Sorry for minor hijack.
Thanks - Pete.

[Koreth]

Another hijack I suppose, but I believe it is at least partially related.

The cathode bias resistor seen in the ubiquitous common cathode stage provides a degree of local negative feedback, thus the cathode bypass cap to get the stage's gain back to its maximum. If I'm understanding the above statements correctly, a stage with negative feedback appears to clip very abruptly. So, if I were wanting design a stage to clip and break up gradually and gracefully, bias considerations aside, would I indeed want to partially or fully bypass the cathode on that stage to remove the negative feedback, or would I be working against myself by bringing in too much gain?

[J M Fahey]

Very short answer: bypass the cathode resistor for maximum gain=least feedback.
Since you are interested in (graceful) clipping, more gain works for you, not against.
It would be interesting to post a few scoped waveforms, with and without. (hint hint)