Variable Negative Feedback, good or bad? Opinions, please.

[dtbradio]

Hi All!

I'm currently working on my third amp project. Its basically a JCM800-type front preamp and tone stack coupled with a Fender Bandmaster57-style power section, with a few minor alterations. The Bandmaster has fixed NFB using a 56K into the cathode of the 1st PI tube, with a 5K-ohm and .1uf presence circuit. I'm using 3.3k on that cathode, and am going to try a variable NFB circuit using a 68K fixed resistor in series with a 50K variable resistor on a switch to allow either adjustment or elimination of the NFB. The presence control I've selected is using a 50K variable instead of the 5k used by Fender, basically for way more range of control.

I'd like any and all opinions of those changes to the basic Fender circuit. Thanks in advance for all replies!

[Esquirefreak]

Whenever I incorporate a presence control in a build, I use the B25k/4k7 divider with a 100nF cap (used in the later, big boy Fender brownface amps). I think it gives a really usable range all across the dial. 12 o clock for singlecoils and 3 o clock for humbuckers.

I've tried a B50k pot as well, but I think that the amp got a bit thin up the dial.

/Max

[HotBluePlates]

... a Fender Bandmaster57-style power section ... The Bandmaster has fixed NFB using a 56K into the cathode of the 1st PI tube, with a 5K-ohm and .1uf presence circuit. I'm using 3.3k on that cathode, and am going to try a variable NFB circuit using a 68K fixed resistor in series with a 50K variable resistor ...

The "feedback circuit" takes a voltage present across the speakers (really, at a specific OT secondary tap) and applies it to a voltage divider made of (at least) 2 resistors.

    In the Bandmaster this voltage divider is made of the 56kΩ and the 1.5kΩ cathode resistor.

    The Bandmaster sources feedback voltage from the "4Ω tap" (even though there is a 2.667Ω load present).


You have changed half of the voltage divider by increasing the cathode resistor to 3.3kΩ.

    All else the same, you would need to change the series feedback resistor to ~123kΩ (120kΩ is close enough).

    As planned, the 68kΩ + 50kΩ var resistor gives "more feedback" down to "a-bit-more-than-stock feedback."

    You might consider 120kΩ resistor + 250kΩ var resistor to give you a range of "less feedback" options.  Or 68kΩ + 250kΩ var resistor for more/less feedback.


What tap are you using to source your feedback voltage?

    The stock Bandmaster used a 4Ω tap; if you use an 8Ω tap on the OT, you increase feedback by √2.

    So using a different OT tap to source the feedback requires re-juggling the resistor values.

    If you have an impedance selector, connect the feedback wire to a tap on the non-speaker side of the selector switch.  Doing otherwise causes the amount of feedback to change with the impedance-change.

[tubeswell]

Just chipping in with my own worthless opinion - having tried this in numerous amps (which are admittedly mainly straightforward rock and blues type tube amps, but also a couple of medium-high gain amps).


When it comes to your typical guitar amp, there ain’t a heck of a lot of sonic difference between NFB and no NFB with everything else that’s typically going on when you’re playing guitar through it.


Yes - there’s a slight jump in output volume and decrease in bandwidth with no NFB, but it’s very slight, and after you dime the amp, it’s not like really like ‘night and day’ IMHO. A straight NFB switch has a more noticeable difference than a variable NFB control. And in a live band situation with all sorts of other noise things going on around, the difference is pretty non-existent.


Having said that, presence or resonance controls can make s subtle difference (in some amps), but you get ‘more’ out of a TS with a ‘raw’ control - if you really want another control. Just sayin’.


But try the variable NFB by all means. YMMV

[dtbradio]

Thank you very much for taking time to reply, and your insight!

Your thought on using a 250K POT makes sense, and would allow me to use either the 4 or 8 ohm tap for NFB. I will probably stay with the 50K switchable for now simply because I don't want to add another switch to a crowded panel for now; I don't have a 250K with a switch on hand. I was thinking of using the 8-ohm tap, but will likely stick with the 4-ohm for now. I may change the 68K up to an 82K to gain a bit more room in the less feedback area, as that will give me a resistance range of 82K to 132K.

I'll add that I'm planning on switching between 6V6 and 6K6 tubes, and using the 4-ohm side of the OTP when using 6K6 to reflect a higher (safer?) impedance to the 6K6 plate circuit. I will be using an 8-ohm speaker. I will also have a 10-ohm 20-watt loading resistor connected when a speaker is not plugged in for PA tube safety. My OPT is here: https://www.amazon.com/gp/product/B09PNVYX8Y

... a Fender Bandmaster57-style power section ... The Bandmaster has fixed NFB using a 56K into the cathode of the 1st PI tube, with a 5K-ohm and .1uf presence circuit. I'm using 3.3k on that cathode, and am going to try a variable NFB circuit using a 68K fixed resistor in series with a 50K variable resistor ...

The "feedback circuit" takes a voltage present across the speakers (really, at a specific OT secondary tap) and applies it to a voltage divider made of (at least) 2 resistors.

    In the Bandmaster this voltage divider is made of the 56kΩ and the 1.5kΩ cathode resistor.

    The Bandmaster sources feedback voltage from the "4Ω tap" (even though there is a 2.667Ω load present).


You have changed half of the voltage divider by increasing the cathode resistor to 3.3kΩ.

    All else the same, you would need to change the series feedback resistor to ~123kΩ (120kΩ is close enough).

    As planned, the 68kΩ + 50kΩ var resistor gives "more feedback" down to "a-bit-more-than-stock feedback."

    You might consider 120kΩ resistor + 250kΩ var resistor to give you a range of "less feedback" options.  Or 68kΩ + 250kΩ var resistor for more/less feedback.


What tap are you using to source your feedback voltage?

    The stock Bandmaster used a 4Ω tap; if you use an 8Ω tap on the OT, you increase feedback by √2.

    So using a different OT tap to source the feedback requires re-juggling the resistor values.

    If you have an impedance selector, connect the feedback wire to a tap on the non-speaker side of the selector switch.  Doing otherwise causes the amount of feedback to change with the impedance-change.

[dtbradio]

Thank you for your reply! I'll be curious to see just what the 50K does in this amp. This is all new to me, and I enjoy experimenting!

Whenever I incorporate a presence control in a build, I use the B25k/4k7 divider with a 100nF cap (used in the later, big boy Fender brownface amps). I think it gives a really usable range all across the dial. 12 o clock for singlecoils and 3 o clock for humbuckers.

I've tried a B50k pot as well, but I think that the amp got a bit thin up the dial.

/Max

[dtbradio]

Thank you for your time and reply!

I'm primarily interested in experimenting with the compression effect that NFB gives. No idea yet if I'll like more or less yet, hence the control to experiment. As for the Raw control you mentioned in the TS, this is the first I've heard of that. Is that basically a bypass of the tone stack? if so, my 2nd amp project (the first one that actually worked properly, lol) is raw.

Just chipping in with my own worthless opinion - having tried this in numerous amps (which are admittedly mainly straightforward rock and blues type tube amps, but also a couple of medium-high gain amps).


When it comes to your typical guitar amp, there ain’t a heck of a lot of sonic difference between NFB and no NFB with everything else that’s typically going on when you’re playing guitar through it.


Yes - there’s a slight jump in output volume and decrease in bandwidth with no NFB, but it’s very slight, and after you dime the amp, it’s not like really like ‘night and day’ IMHO. A straight NFB switch has a more noticeable difference than a variable NFB control. And in a live band situation with all sorts of other noise things going on around, the difference is pretty non-existent.


Having said that, presence or resonance controls can make s subtle difference (in some amps), but you get ‘more’ out of a TS with a ‘raw’ control - if you really want another control. Just sayin’.


But try the variable NFB by all means. YMMV

[pdf64]

I'm primarily interested in experimenting with the compression effect that NFB gives.

I dunno if that’s a thing?
An audible degree of signal compression in a valve guitar amp is typically due to HT sag / cathode bias squish.
eg a cranked AC30. No NFB.
https://www.aikenamps.com/index.php/what-is-sag

[dude]

I've been following this tread, great info. Not to cut in on dtbratio's tread, can I ask one question on NFB var. pot and resistors?  The Blues Junior lll, cream bd., I added a presence pot, (5kl, .1uf), exactly like the 6V6 plexi. I used a feedback R of 100k, the original was 75k. The presence works good but very fast change to brightness, only an 8 ohm tap. Since this PCB bd. is very difficult to work on, I"m wondering if more feedback to the 5kl would give me not such a quick brightness change when adjusting the 5kl? Perhaps adding another 100k in parallel to existing 100K, for 50K and more feed back or maybe I need less feedback and increase that resistor?

[dtbradio]

Thank you for your input! I've read through Aiken's sag write-up in the past, great info! Yes, there is some compression effect from sag. I'm using solid-state, so no sag there. I've thought about installing a small-value high-watt resistor in line with the main HT supply to simulate tube rectifier sag, but will implement that later. However, NFB DOES introduce gain reduction, and since the amount of NFB increases the harder you play (therefor more gain reduction), compression occurs. I just don't know how much of an audible effect it will actually have, since I have so little experience with tube amp builds at this point. Hence the experimentation.

I'm primarily interested in experimenting with the compression effect that NFB gives.

I dunno if that’s a thing?
An audible degree of signal compression in a valve guitar amp is typically due to HT sag / cathode bias squish.
eg a cranked AC30. No NFB.
https://www.aikenamps.com/index.php/what-is-sag

[dtbradio]

This is a thought based only on my thoughts and not experience, but have you considered changing the POT's taper? IE, try an audio taper if using linear and vice-versa. I do think your resistor idea would lead to less presence control effect, but then you may run into other effects that may not be good as a result of the added NFB. Again, not speaking from experience, just thinking through the circuit. Sounds like its time to experiment!

I've been following this tread, great info. Not to cut in on dtbratio's tread, can I ask one question on NFB var. pot and resistors?  The Blues Junior lll, cream bd., I added a presence pot, (5kl, .1uf), exactly like the 6V6 plexi. I used a feedback R of 100k, the original was 75k. The presence works good but very fast change to brightness, only an 8 ohm tap. Since this PCB bd. is very difficult to work on, I"m wondering if more feedback to the 5kl would give me not such a quick brightness change when adjusting the 5kl? Perhaps adding another 100k in parallel to existing 100K, for 50K and more feed back or maybe I need less feedback and increase that resistor?

[pdf64]

… since the amount of NFB increases the harder you play …

Could you explain your thinking on that?

[HotBluePlates]

... NFB DOES introduce gain reduction ...

Sure, but turning a Volume control to anything other than "full up" also imposes a "gain reduction."

... since the amount of NFB increases the harder you play (therefor more gain reduction) ...

Play Harder ---> More Speaker Volts

More Speaker Volts ---> More Feedback Source-Volts

More Feedback Source-Volts ---> Feedback Voltage Divider ---> More Feedback Volts


      However...

The "Feedback Voltage Divider" is a fixed pair of resistors.  They always deliver the Same-% of Speaker Volts.  So the %-of-Feedback is a constant.

There is no "change of feedback" going from playing very-quiet to very-loud.  That is, until the power section distorts which then eliminates the feedback (which throws away a portion of circuit-gain to use for feedback).

So the amp will compress... but only when the power section is distorting.  Which is where the power section will compress anyway, whether a NFB or No-NFB design.  And since you're using the JCM800 preamp, you're probably looking to make use of the Master Volume, which seeks to derive distortion from the preamp while keeping the power section clean (and delivering a non-changing amount of feedback).



You may be envisioning a feedback loop used within a compressor.  We don't really notice "gain reduction" until there are Attack/Release R-C circuits that change the "% of the output" delivered back to an earlier stage to control the gain reduction.

[dtbradio]

HotBluePlates covered that before I saw your reply, and his explanation was the same as I would have given. He offered some insight that cleared some things up for me, too, and exposed a flaw in my thinking. Which is why I posted in the first place, as I'm VERY new to tubes and looking for exactly that kind of info!

… since the amount of NFB increases the harder you play …

Could you explain your thinking on that?

[dtbradio]

Thank you for the extra info! I'd actually been toying with the idea of rectifying/filtering the NFB, then using it to drive a high-voltage MOSFET to bleed off varying amounts of the master volume's output to ground before the splitter circuit. I would put in a control to vary the amount of NFB drive to the mosfet for a varying amount of compression effect. No idea how that would work, but it would be an informative experiment!

... NFB DOES introduce gain reduction ...

Sure, but turning a Volume control to anything other than "full up" also imposes a "gain reduction."

... since the amount of NFB increases the harder you play (therefor more gain reduction) ...

Play Harder ---> More Speaker Volts

More Speaker Volts ---> More Feedback Source-Volts

More Feedback Source-Volts ---> Feedback Voltage Divider ---> More Feedback Volts


      However...

The "Feedback Voltage Divider" is a fixed pair of resistors.  They always deliver the Same-% of Speaker Volts.  So the %-of-Feedback is a constant.

There is no "change of feedback" going from playing very-quiet to very-loud.  That is, until the power section distorts which then eliminates the feedback (which throws away a portion of circuit-gain to use for feedback).

So the amp will compress... but only when the power section is distorting.  Which is where the power section will compress anyway, whether a NFB or No-NFB design.  And since you're using the JCM800 preamp, you're probably looking to make use of the Master Volume, which seeks to derive distortion from the preamp while keeping the power section clean (and delivering a non-changing amount of feedback).



You may be envisioning a feedback loop used within a compressor.  We don't really notice "gain reduction" until there are Attack/Release R-C circuits that change the "% of the output" delivered back to an earlier stage to control the gain reduction.