Global NFB - Calculating resistor value

[diddymix]

Hello Folks,

I was hoping somebody could perhaps help me with a querie in regard to trying to install NFB loop in my amp which currently doesn't have one. I don't know how to exactly calculate the resistor value with my amp which is a Marshall 40W combo, push pull 4 x EL84... with a typical long tail pair PI. The amp at the moment just has the tail resistor and AC shunt cap just connected straight to ground.
I'm looking at some basic schematics of say a Dumble 102.. and the loop is connected to the 4ohm tap of the OT. It then has a resistor value of 4.7K, then it connects to the tail resistor and shunt cap.. followed by a 390R resistor (variable pot and cap in parallel)... to ground.

Just a few questions if I may would really appreciate any input thank you ! My amplifier doesnt have a 4 ohm tap... only 8 and 16 ohm. If I was to implement this circuit into my amp would I just double those resistor values?? Eg 4.7K resistor becomes 9.4K worth?
Also I would be trying this with just a fixed value and no presence pot. Would I just say eliminate the pot and the 0.1uf cap underneath and keep the other resistor values the same??
Lastly would I need to change these values at all considering my amp is 40W EL84 and not a 100W 6l6??
Thanks for reading

[pdf64]

The NFB resistors are a potential divider, they take a fractional voltage sample β of the loop output back to the loop input, the fraction in your example being 390/390+4700 = 0.0766 or 1/13 if I've understood your circuit description right (a link would be helpful).
A 1 watt 8ohm signal has 1.414 x the voltage of a 1 watt 4ohm signal, so to put the same voltage to the loop input, 1/1.414 x 1/13 = 1/18.4 .
Keeping the 390 bottom resistor of the potential divider the same, the top resistor can be factored up:- 390x1/390x18.4 = 390 / 7170.
7170 = Rtop+390, so Rtop = 6780 so use 6k8.
The actual degree of NFB depends on both the feedback ratio β and the open loop gain Aol
The open loop gain Aol will be mainly affected by the LTP gain, power tube used, its operating point and its load impedance.
See https://en.wikipedia.org/wiki/Negative- ... _amplifier

Bear in mind that a lot of time will have gone into making your amp sound right with the power amp operating open loop; the power amp gain, freq response and interaction with the cab will change a fair bit when NFB is applied.

[diddymix]

Thanks! Thats really useful information and thanks for working that out!.. I guess I can give it a try and if it sounds awful I can put it back to the way it was.. or perhaps try less feedback. The thing I am trying to accomplish really is less output tube distortion, so a cleaner output for the earlier stages to work into. Would I be right in saying the NFB 'tightens' the sound and flattens the frequency response? I ,might try putting it on a switch so I can compare..

[Roe]

adjustable nfb is the way to go. use a pot in series with a resistor, eg. 100k + 22k

[pdf64]

Yes, the open loop gain of an EL84 amp may may be higher than a 6L6 amp; if trying to replicate the NFB implementation of a different amp, what matters is the dB reduction provided by the NFB loop; so it's a good idea to make it adjustable/switchable, initially at least.
Yes, NFB should act to damp speaker resonance, so they amp's response should feel tighter; and (presence / resonance / presonance circuits excepted) its freq response should be flatter.
Bear in mind that applying NFB, whilst reducing (within the linear range of the amp) the level of low order harmonics generated, it may actually increase the level of higher order harmonics.
It may be that higher order harmonics are more musically objectionable, as they can be discordant to the fundamental.
When the power amp signal level is above its linear range, ie it's being overdriven, the NFB loop inevitably loses effectiveness.

[diddymix]

Interesting! I think I will try it with a pot then at first and then perhaps find the best spot and use a fixed value from then on if it sounds good.. Yes I don't know much about tube design but at least some techs ive worked with in studio gear with op amps/transistors say the same thing.. I think they tend to think of NFB as a 'necessary evil' in that they want to reduce distortion as much as possible for clean audio, but at the same time not so much as it robs the signal of its 'musicality' and natural tone.. some even try to not use it at all claiming they would rather live with more distortion but a nicer kind that just sounds better and more involving. Guitar amps are obvsiously a whole different thing though as we want distortion and character!
Really I would like to get a cleaner output so I dont hear so much of the EL84's upper mid 'crunchiness'.. and just have a purer low distortion drive for the preamps to dynamically do their thing.. nice in idea but I'm sure it wont be as simple as that!

[diddymix]

adjustable nfb is the way to go. use a pot in series with a resistor, eg. 100k + 22k

Ok thanks mate! Can I ask a couple of questions if I may.. Ive never done this before gulp. So from what the other gentleman said, about the top and bottom resistor.. in this case : 6.8k (top) and 390R (bottom to ground).... How would I implement this with a pot.. do you mean I would use a 100k pot (for the 6.8k) and then do I need the 22k in between that and the 390R fixed to ground?. Or is it as simple as just the 100k pot + 390R?

Also after finding the setting, do I have to unsolder the pot and measure the setting out of circuit?

Really appreciate any help on this!

[pdf64]

Yes, I can see that using a pot with a potential divider might get confusing.
Just wire the pot as a variable resistor, and put it in series with the top resistor.
You might want to make the top resistor value a bit smaller (eg so 3k3 or 4k7 rather than 6k8), so that you can trial the effect of a greater degree of NFB (pot resistance at min), as well as a lesser degree (pot resistance up a bit).
If 100k, I suggest to use a audio / log taper pot.

[diddymix]

Yes, I can see that using a pot with a potential divider might get confusing.
Just wire the pot as a variable resistor, and put it in series with the top resistor.
You might want to make the top resistor value a bit smaller (eg so 3k3 or 4k7 rather than 6k8), so that you can trial the effect of a greater degree of NFB (pot resistance at min), as well as a lesser degree (pot resistance up a bit).
If 100k, I suggest to use a audio / log taper pot.

Ok right... however do I need that top resistor? Could I not just use the 100k pot or would that be way too sensitive to find the best sounding sweet spot?? Also if I did use the top resistor like you say... would I just go top resistor, followed by pot, followed by ground resistor? Once I find the best pot setting in this scenario would I just get the correct resistor value and add that to the 6.8k top resistor for the correct total value??

Thanks so much man, I think im nearly there forgive me, im a right noob

[Roe]

think of it as a voltage divider. usually one resistor is 5k and the other is 27-100k. Just replace the latter with a 27k in series with a 100k pot

[diddymix]

think of it as a voltage divider. usually one resistor is 5k and the other is 27-100k. Just replace the latter with a 27k in series with a 100k pot

Ah right no sorry what I meant was I want to find the setting on the pot that I want then use a fixed resistor instead. So would I just add the 5k to the setting I measured on the pot and use that total?

[martin manning]

What model is this amp? Do you have a link to a schematic?

[diddymix]

What model is this amp? Do you have a link to a schematic?

Hi Martin, yes I'll upload it here. Its a 40W Marshall combo DSL401

[Roe]

think of it as a voltage divider. usually one resistor is 5k and the other is 27-100k. Just replace the latter with a 27k in series with a 100k pot

Ah right no sorry what I meant was I want to find the setting on the pot that I want then use a fixed resistor instead. So would I just add the 5k to the setting I measured on the pot and use that total?

no single setting will work for all applications. a pot is better imho

[martin manning]

I would do this, and then find a FB level that you like. Then if you don't want to keep the adjustable FB, measure the resistance across the 500k pot (disconnect the end going to the 27k first) and replace it and the 27k with a single resistor that has a value equal to the pot measurement plus 27k. Presence pots are linear taper, the 500k could be log or linear.
A couple of things to note:
- I chose 500k for the FB control so you can go from almost no FB (less than 1% of speaker voltage) to heavy feedback (15% of speaker voltage). You could connect the 27k to the 8-ohm tap and reduce the range to roughly 0.7% to 11%.
- You could replace the presence circuit shown with the modern system that uses a 4k7 resistor and a 25k pot, but old scratchy sounds better IMO.
- The OT primary-secondary phasing is not known. If you get oscillation, try reversing the primary leads (W21 and W22).