Recalculating resistor values for mixed mode feedback

[Mick Bailey]

Considering Rod Elliott's diagram regarding mixed-mode feedback: When applying this to an existing amp, what's the calculation in order to maintain roughly the same nominal gain? For example, If without current feedback R1 is 47k and R4 1.2k, using Rod's values for R2 and R3, how would I calculate the new value for R4?

When I've previously applied current feedback I've substituted a trimmer for R4, set it by ear, Then replaced the trimmer with the nearest value fixed resistor. This seems an involved way to do it when a calculation would be more straightforward.

[Helmholtz]

That's a bit tricky, as both R2 and R4 affect voltage NFB as well as current NFB.
Calling the effective shunt part of the voltage NFB divider R4', we have 1/R4' = 1/R4 + 1/R2 as they are essentially in parallel.
OTOH, the voltage divider for the current NFB is formed by R2 and R4.
So if you increase R4, also R2 must be increased by the same percentage for the same amount of current NFB, which in turn changes R4'. ..

Can you post a link to the Elliot article?

[Mick Bailey]

I wondered if it would be more complicated than simply considering R4 to be in parallel with R2 (ignoring R3). Fender does this differently (for example in the London Reverb) by eliminating R4 and increasing the value of R2 to what R4 would have been. That appears to be a simpler way to do this though I don't know what the downside would be.

[Helmholtz]

Not really sure what your context/application is, but teemuk's great book on Solid State Guitar Amps might help:
https://www.thatraymond.com/download...ttala_v1.0.pdf
See pages 87ff.

[Mick Bailey]

The application is in improving the sound of boxy sounding SS combos. Changing to mixed-mode feedback has a drastic improvement in bass and treble response where previously voltage only feedback was used and a sub-optimal cabinet size had been chosen by the manufacturer.

Getting the right balance between voltage and current feedback for any particular amp design can be a bit of trial and error. When applied incorrectly the power amp can end up with too much or too little gain, excessive bass, unwanted noise, or little or no improvement. The main consideration is choosing the most appropriate values for R2/R4 for a given amp.

Maybe it's not as straightforward as just calculating the values, as there's a good deal of subjectivity in how an amp sounds. In particular, the speaker has a bearing - for example, a Celestion G12K responds quite differently to a G12H. I find that mixed mode feedback gives a much better representation of different speaker characteristics.

[Helmholtz]

The application is in improving the sound of boxy sounding SS combos.

I see, so guitar amp. Another application could be a reverb driver.

Yes, guitar speakers don't "like" low output impedance/high damping factor.
Tube amps using NBF typically have an output impedance between 2 Ohm (e.g. JTM 45) and 20 Ohm (Fender BF amps) at the 8 Ohm output.
The low impedance amps benefit most from a presence control.

Tube amps not using NFB might have an output impedance around 100 Ohm (e.g. Vox AC30: 75 Ohm, Marshall 18W: 90 Ohm, Tweed Deluxe: 80 Ohm).
Data from Zollner's book.

Could Teemu's book help?

[Mick Bailey]

Maybe it could, but I'm not seeing past a particular aspect with regards to the gain calculation. A practical example is perhaps the schematic in fig. 2 here: https://sound-au.com/project27.htm

Ignoring the current feedback the nominal gain of the power amp is 23, but when R23/R26 are included, why don't these behave as if they are in parallel to R6? If they did the gain would be roughly 10x the nominal gain of the voltage-only feedback amp, but this can't be the case. What I'm looking for is a more simplified explanation.

[Helmholtz]

Ignoring the current feedback the nominal gain of the power amp is 23, but when R23/R26 are included, why don't these behave as if they are in parallel to R6?

I think they do, but the current NFB lowers the gain again.
You should see high gain with speaker disconnected.

[Mick Bailey]

I did some practical tests with a donor amp this afternoon and for my purposes the most predictable way of adding current feedback to a voltage-only NFB amp is to disconnect the ground side of the existing NFB divider resistor, place a 0R1 5W in series with it to ground and connect the speaker ground to the junction of the resistors. It works well without having to alter any of the resistor values, with just a slight gain boost.

It woud be interesting to know the proportion of current feedback at different frequencies.

[Helmholtz]

Maybe not exactly what you're after, but I would measure gain with speaker connected at different frequencies. Gain should vary with the speaker impedance.
Then I would disconnect the speaker to disable the current NFB and measure gain again. Should not depend on frequency now.

I would be more interested in measuring output impedance.

[Mick Bailey]

After some more thought the circuit in post #1 suddenly made sense and I re-read Teemuk's section on mixed mode feedback with better appreciation of that particular circuit. In the diagram comparing two different topologies on page 89 and looking at practical applications of both in commercial amps, the question arises over the no-load gain of both types, given that the main divider resistors often have comparable values, but the second type has no parallel resistance.

I did an A-B comparison yesterday with a single amp and whilst the first circuit type has a much greater gain boost, when adjusted for roughly equal volume I didn't notice any audible difference in frequency response between the two. Fortunately I have a pair of Sessionette 75 combos to compare side by side and later today If I have the time I'll implement each configuration for a direct comparison.

[Helmholtz]

I did an A-B comparison yesterday with a single amp and whilst the first circuit type has a much greater gain boost, when adjusted for roughly equal volume I didn't notice any audible difference in frequency response between the two.

Are you referring to the circuits of Figure 3.35?

So "first circuit" means voltage NFB only and second circuit only current NFB?

[Mick Bailey]

Figure 3.37, page 89.