re: NFB in guitar amplifiers

[Chuck H]

I've noticed that many popular designs and modern clone based amps seem to use ubiquitous ratios regardless of other circuits aspects. Examples:

Fender Deluxe Reverb 820/47 at 8 ohms 20W
Fender Vibrolux 820/47 at 8 ohms 50W
Fender Bandmaster 820/100 at 4 ohms 50W
Fender Twin Reverb 820/100 at 4 ohms 100W

What I see is that the higher wattage amps are using more NFB with these relative ratios. And among the two power tube models the move from 820/47 for 8 ohms to 820/100 for 4 ohms is a straight x2 difference when I think it should be closer to x1.5, right?

And this all over the BF and SF era amps by Fender. It always seems to be 820/47 or 820/100 and the amps listed above all use the same AB763 circuit otherwise.

And then there's Marshall's classic designs that use either 100k/4.7k or 47k/4.7k on different taps depending on some gain differences in preamp circuits. But they seem to have totally missed the 50W to 100W difference. For equivalent era and amp type (like a 50W '70 "Super Lead" compared to a 100W 70' "Super Lead") they usually went with the same secondary tap for both models but the 50W had the 100k series resistor and the 100W had the 47K ??? So that would be A LOT more NFB for the 100W models.

Not being very technical I have to ask about this. Is this arbitrary? Or maybe even technical error? Or is there a reason for the higher NFB ratio for the higher wattage amps?

[loudthud]

To properly analyze differences in feedback ratio, you need to account for output transformer ratio... and you can't really compare different output tubes. Back in the old days, you would measure the open loop gain in dB, then the gain with feedback connected, subtract the two and call that number the amount of feedback in dB. One reason for this was that since open loop gain was not sky high like it is with opamps, the gain closed loop gain would be several dB less that the feedback ratio would indicate in an opamp circuit.

The old Fender circuits were not perfect, but close enough for Rock and Roll. The biggest error was probably the Super Reverb with it's 2 Ohm OT and 820/100. Note the 5F6A was probably correct with 27K/5K and it's 2 Ohm OT.

The big effect of all this is the actual impedance that the speaker sees driving it. Look at this attachment for an easy way to measure output impedance.

[Helmholtz]

Open loop gain and NFB voltage divider determine closed loop gain and power amp sensitivity via the feedback factor (aka feedback ratio).

Available amp power does not enter the feedback formula.

But as loudthud mentioned open loop gain depends on OT step-down ratio.
An OT for 4 power tubes typically has half the primary impedance of a 2 tube OT (assuming same tube type).
So the voltage step-down ratio for the 4 tube OT is 41% lower and open loop gain will be 41% higher (at same secondary tap).
To compensate for same power amp sensitivity as with the 2 tube version one could either connect the NFB to the next lower impedance OT tap or lower the NFB series resistor.
The latter method would mean more NFB and lower output impedance.

[Chuck H]

Thank you LT and Helmholtz. I was looking at it as a purely voltage dependent circuit. ie: how much voltage from a given output tap and the NFB circuit is reaching the PI tail shunt. I had not considered loop gain differences due to transformer ratio. This helps even though I haven't fully digested it yet.

[loudthud]

Bottom line is there will be slight differences in tone between the 820/100 and the 820/47. Some may be due to the difference in gain and THD, some because a slightly different impedance is driving the speaker. For my money, the easiest way to tell is to install a pot to go between the two and give it a twist while playing. Something to plug up that Ground switch hole.

[Chuck H]

Bottom line is there will be slight differences in tone between the 820/100 and the 820/47. Some may be due to the difference in gain and THD, some because a slightly different impedance is driving the speaker. For my money, the easiest way to tell is to install a pot to go between the two and give it a twist while playing. Something to plug up that Ground switch hole.

Well this is how I've always done it. I'm not tech savvy but I do my best and have made made several custom designs using the "dial it in" method for the NFB ratio. WRT guitar amps whatever sounds right IS right. Not so for reference amps or any other circuit intended for minimum distortion.

As I mentioned, I was looking at the circuit as strictly voltage dependent. So, for example, with a 50W "Super Lead" with a 100k off the 8 ohm secondary for the NFB circuit you have about 1.9V of NFB fed to the 4.7k shunt at the PI tail. But with a 100W unit from the same year using a 47k from the 8 ohm secondary for the NFB circuit you have about 2.8V at the 4.7k PI tail shunt (using 50W and 100W as reference points regardless of their real accuracy). In my mind that is simply MORE NFB voltage at the PI tail for the 100W model. How that ultimately affects gain relative to OT ratio is still foggy. I'm working on it. But at the very least I would think that it should strongly affect the PI wave form independently from the loop gain strictly from a difference in the applied antiphase AC voltage. But I'm really over my head on this so any layman's explanations about how/if this evens out would really help.

[Helmholtz]

Feedback is based on gain, i.e. a ratio between 2 voltages.
It does not depend on absolute voltage level.

More exactly the effective amount of NFB is given by the open loop gain times the NFB divider attenuation factor.This product is called loop gain.
It is the gain the NFB signal experiences when running around the full loop, i.e. from output to PI and through amp back to output.

So when the open loop (forward) gain is larger, less NFB attenuation could be used to compensate. But then closed loop gain will be higher.

NFB affects a number of amp properties like gain, output impedance, frequency response and distortion (especially power amp clipping characteristics).
Also presence will be more effective with larger NFB.

There is no "correct" amount of NFB, but if you use too much the amp will oscillate.