Adding NFB control to AB763 Deluxe Reverb build

[Stratz]

Hi all.
I would like to add a 50k pot on the front of my Deluxe Reverb build and I have a couple of questions.
This was my first build so I'm not an amp builder or tech by any stretch but I've been servicing, biasing and have done some on modding my tube amps for nearly 40 years so you dont need to worry about my safety around high voltages.

First, can someone kindly provide somewhat detailed instructions on adding the pot to the NFB loop.
If I understand correctly it must go in series with the 820 ohm resistor (I plan on leaving the resistor in place) to the wiper lug of the pot. Is this simply taking the NFB wire off of the speaker jack (the Fender DR layout shows it going to the EXT speaker jack first) and adding the pot in series? If not could someone please explain what I need to do. Also what type of wire to use and the best path to run it since it will be running quite a ways from the back of the chassis to the front of the amp.

Secondly, I'd like to put the pot on the front of the amp in the CH1 low impedance input jack hole since it never gets any use.
My question here is which resistors on the input jack need to stay and which can go.
Theres a 68k resistor on each jack and a 1Meg resistor on the jack that I wont be touching.

Thanks.
Rob.

[teemuk]

Magnitude of negative feedback is generally set by -ratio- of impedances in the "feedback loop". The ratio, in this case, is set by 820-ohm resistor that connects secondary side output to "inverting" input of the phase splitter (through its tail), and 47-ohm "shunt" resistor that forms a voltage divider with it and effectively attenuates the feedback signal.

The amount of this attenuation establishes magnitude of applied negative feedback: No attenuation = 100% negative feedback, full attenuation = NFB disabled. You usually wish to limit this range to somekind of margin (e.g. magnitude of NFB can not increase above certain %). Some amplifiers may oscillate at either extreme, negative feedback decreased too much (or disabled) or negative feedback increased too much. You need to experiment.

So, between the lines you may have read that you have two options: You can modify resistance that substitutes the 820-ohm resistor in the NFB loop, or you can modify resistance that substitutes the 47-ohm resistor. What you are after is in essence just a variable -ratio- of those two. If you wish to use a potentiometer for adjustment (I'll get back to that later) component of higher impedance is most likely more readily available. A certain taper may be required as tweaking magnitude of NFB (among several other things) also alters amplifier's gain and linear adjustment may seem insensitive in that regard.

Yes, you can simply wire a potentiometer as "variable resistor" in series with the 820-ohm resistor. That's one way. Do note that effectively this limits amount of max. NFB to "stock" feedback settings of the design. Replacing the 820-ohm with lower value may result to increased "tightness", "punch" and "cleanness" with slightly increased headroom and "crispier" overdrive. By leaving 820-ohm in place you can exploit "looser" characteristics of decreased NFB but never the other extremes, which also exist.

But anyway, you have many ways to "skin the cat", and this probably also makes clearer how negative feedback -based "tone controls" (e.g. presence, resonance, etc.) work: They simply employ frequency-selective negative feedback. This is why they also interact with any "master" feedback loop control.

In my experience, feedback control is not something you "fine tune". Such precision adjustment is hardly ever needed and that renders continuous adjustment an unneccessary feature. Overall you may benefit from a simpler control, like 3 -or 4-way switch with "damping" adjustments along lines of "High" (Most NFB you wish to apply), "Medium" and "Low" (no NFB or least amount of NFB you wish to apply). You probably notice some difference between each setting but overall our sense of the "intermediates" in between would be somewhat insensitive. So I would just use a switch. If precision adjustment is required I would employ both: Switch to set the overall "range" of loop impedance ratio, and potentiometer to "fine tune" within that range. But, IMO, unneccessarily complex.

Physical wiring concerned, I'm not experienced with AB763's but there are few guidelines that apply to most situations:
- Keep the feedback wire loop physically as short and tidy as you possibly can
- Keep the loop away from potential "noise" and signal sources (e.g. circuit segments and wires carrying high currents, transformers and magnetic fields, etc.)
- Shielded wiring may be beneficial

[Chuck H]

Along the lines of what teemuk said (if I interpret correctly), I agree that feedback circuits can be idealized and left alone. My own experience with variable feedback is that for any given amp an ideal will present itself and I would rarely even use the control afterward. Even for different tones. But that's just my own preferences.

teemuk also mentioned (correctly) that the 50k pot in series with the circuit doesn't offer intuitive, linear adjustment. That is a rough circuit used by less technical amp modders who just needed a way that worked, but it's far from idealized. Better to use a smaller pot in place of the series or shunt resistor that comprise the feedback circuit. The series resistor is the 820R and the shunt resistor is the 47R. The "ratio" function of this circuit was explained by teemuk. The 820/47 ratio of the stock loop is roughly 17:1 You can duplicate the ratio with higher resistance values (within reason) that allow the use of readily available pots to make the variable circuit and the "stock" setting of the control will behave the same as the unmodded Fender circuit.

An aside on stability. The feedback loop is connected to the amplifiers output. It's generally considered "bad" to have the amps output near it's input. In fact you want it as far away as practical. So a wire from the output jack to the board and/or a front panel pot is a bad idea. teemuk mentioned, and I've indicated above that as long as the ratio of the loop in maintained it works the same in the amp. So, best practice for stability would be to have a feedback series resistor of a somewhat higher resistance soldered right on the output jack. This way you're sending an already attenuated output to the board or pot rather than the full amps output. In this same light it's also a good idea to locate your panel control as the last knob, as far from the input as possible. And route any leads for the feedback circuit out and away from the preamp as much as practical without excessive lead length.

Ok, so we need a ratio of 17:1 you can use a 1k pot wired as a variable resistor in place of the stock 47R and a 18k resistor soldered to the output jack in place of the 820R. This gives a variable ratio of infinity:1 to 18:1. And with a much more intuitive feel and finer adjustment to the control than simply kludging in a 50k pot on the stock circuit and less risk to the amplifiers stability. If you wanted to be able to dial in a notable amount MORE feedback than stock you could use the 1k pot as above and instead of an 18k resistor, use a 10k. This would allow a ratio from infinity:1 to 10:1 which is more feedback than stock (the lower numerical ratio indicated more feedback).

Hope this helps.

[Stratz]

Thanks very much for the detailed explanations guys.
I did install a 50k pot in the back (ground switch hole) using the attached diagram that I found on this forum and left the 820 ohm resistor in place. I dialed in what I feel is the sweet spot so there it will stay.
I hear no ill effects so I hope it will be alright.
Much appreciated.
Rob.

[Dave H]

I did install a 50k pot

It says 5k on the sketch, much better.

[Chuck H]

Also, a common suggestion here is to use a pot to "dial in the sweet spot", measure it, replace the pot with a fixed resistor of the measured value and just skip having the added control on the amp.

[mozz]

As was said, a switch is great for this situation. I usually have the stock resistor, one lower, one higher. You can have stock, clean , cleaner or stock, dirt, dirtier.

[Randall]

I just did this with a 5K pot on a Deluxe Reverb clone. Wired per the above drawing keeping the 47R the same. It works fine for me.

[Stratz]

Same here but I removed the resistor to get the most swing.
I also have a midrange control in the vibrato LO input. I use it a LOT.

[Chuck H]

Same here but I removed the resistor to get the most swing.
I also have a midrange control in the vibrato LO input. I use it a LOT.

What "midrange control".?. Do tell. Is it the variable mid resistor, ala Kendrick "Texas tea" or is it a variable bigger pf cap bypassing the treble cap? (I like that one)

[Stratz]

It's just a pot (dont recall the value at the moment) coming from the bass control pot where the mid resistor was.
You can see it on the CH2 low input.

[robrob]

The AB763 circuit has quite a bit of NFB so I like a 3-way switch with stock, light, no feedback:

https://robrobinette.com/AB763_Modif...ative_Feedback

[SoulFetish]

Teemuk hit it on the head. Because of the divider rule, a series pot is okay if you want to switch up values quickly. But, I wouldn't expect it to have a comfortable taper. In order to control the feedback factor in any meaningful way both values need to change in proportion to their combined resistances. That's why a series pot may get you where you want to be, but it makes for an awkward "control".

The first amp I built used feedback. But, after a rebuild, I was getting massive oscillations and instability. This was a fortunate and hard lesson on respecting feedback, which I needed to learn. At first I couldn't understand why my amp was a howling mess, and gradually worked my way back to having no feedback at all. Then I learned (and am learning) about feedback theory, and went through the exercises of designing an appropriate level of feedback for my amp.
I found that I really liked the sound and feel of my amp with and without feedback. I found both to be quite different and ultimately settled on a simple "on-off" mode.
In my experience (other than bandwidth), the biggest difference I've found when using global feedback in a particular amp is in the way it interacts with me as a player. I find they play much differently in the way they transition into clipping, and respond to picking dynamics. It's weird, but in my amp, when feedback is engaged–I find that it is "harder" to play articulately and is less forgiving. Where as, when feedback is disengaged, the amp plays much "smoother" or "easier" to me; particularly when playing fast.

[SoulFetish]

Any thought, as to why Fender used such low impedance values in their feedback resistors? Was it to ultimately keep the tail resistances to a minimum and maximize headroom? I'm not sure I see the sense of using such low values, but maybe there's something I'm missing.

[Chuck H]

Probably just an effort on the part of the engineers to minimize the effect of the feedback loop shunt resistance on phase inverter gain calculations. Though anything under 1k for a shunt would seem ignorable, 47R leaves no doubt. Just guessing though. Another thought is that they already had the 820R value in use as a shared cathode resistor in most BF amps they simply decided to use that value as the NFB series and calculated the 47R value for the ratio. That way they only had to add one resistance value to the inventory. There must be more reasons they might have done it, truly valid or not, but the actual reason is likely lost to undocumented history.