high-pass filter in feedback loop

[Frances Rhodes]

hey y'all

i'm trying to mod a big muff that i bought.
i looked online for common mods, found a few things, but when i tried kitrae's feedback loop mods, it didn't work for some reason, whatever signal coming in, the output was only buzzing, not specially a loud buzz, but a low frequency buzz.
so i figured i may as well try a feedback loop from output to input with a variable resistor in the loop (as seen on beavis audio [although, i still quite don't understand the need to wire the pot as a variable resistor rather than a volume-pot, is it because the signal that is fed back is a current and not a voltage?]), and i came up with the idea of not feeding the whole signal back to the input and put a simple passive RC filter in the loop with a pot as a variable resistor to adjust the cutoff frequency to taste.
but i get the feeling that i'll need to buffer the loop for it to work.
so, all that said, am i completely wrong or do i need to look for an active high-pass circuit not to feed all the signal back to the input?

regards
frances

[GibsonGM]

Can you be more specific, Frances?  Would you like to try to make the kitrae mod work, or do #2, just send some highs from output back to input?
The buzzing you described sounds like a bad/lifted ground or something, which can be fixed...

If you wanna go output back to input, be aware that your output will be NONINVERTED, or 'self- reinforcing', positive feedback...so you wouldn't be creating traditional feedback, which is a 'mellowing out' sort of thing. You'd be cranking the circuit, at risk of creating an oscillator, LOL.    So, to do this the 'traditional' way, you'd need to invert the signal you feed back, OR feed it back to say the 2nd stage, so you are adding negative rather than positive FB.   Each stage here is inverting...the even number of stages means the output is in phase with the input.

You certainly can trim some highs before doing FB, you wouldn't want the full signal level anyway, so a passive RC HPF could work...feel free to play around, but there is an art AND science to FB - closure rates and stuff...too much and you make an oscillator -not enough, you don't hear anything...a little goes a long way!

Keep in mind that local rather than global feedback might be easier - they ARE using local feedback (the small caps from collector to base...) - just wondering what outcome you are trying to get to here?

[antonis]

i still quite don't understand the need to wire the pot as a variable resistor rather than a volume-pot, is it because the signal that is fed back is a current and not a voltage?

Current & Voltage are bond by a very strong relationship which is called Resistance..
(some guys are fond of another "reversed" behavior lady who they call Conductance..)

The main diffence between a (variable) resistor and a voltage divider (i.e. volume pot) is that the first allowes (or restricts) an amount of current (which is proportional to a voltage drop across the resistor) with no loss of the "unwanted" quantity where a voltage divider simply "wastes" it..

A resistor in a feedback loop controls the "amount" of fed back signal (added or exctracted, depending on phase..) and is placed "in series" between the loop's edges with no unwanted effects (i.e. loading the respective circuit)

There is no way to conventionally use a pot for this purpose without sever loss of signal and additionally upseting bias during feedback adjustment..

[GibsonGM]

There is no way to conventionally use a pot for this purpose without sever loss of signal and additionally upseting bias during feedback adjustment..

I didn't discuss this, thanks Antonis.   When universal feedback is 'performed', the amount of feedback is calculated (or done by trial and error), BUT one must account for any disturbances you created during the process!  If you altered bias, you must correct it.
Thus, you can't do this 'on the fly' with a pot. 

You COULD build something like a presence control, where you CAN change things on the fly, so start reading up    I think what's already there is the best compromise for the pedal, IMO.

[TejfolvonDanone]

Can you post a schematic about what you are doing/trying to do?
I don't think it is really an effective way to talk without any schematics.

BUT one must account for any disturbances you created during the process!  If you altered bias, you must correct it.
Thus, you can't do this 'on the fly' with a pot.

If it has a cap in series with the pot you don't alter the bias. So without any schematics it's really hard to solve the issue.

[Frances Rhodes]

ok, sorry if i wasn't very clear, i'll try to explain myself.

i tried kitrae's loop mods, both of them (connect the emitter of Q3 with the emitter of Q2, and connect the collector of Q3 with the collector of Q1). one didn't change anything and the other one was making the pedal buzz, as you said, like if there was a ground loop. i tried it several times, several days in a row and the results were always the same so i thought of trying a different mod which is a different feedback loop, from the pedal output to its input. i'm looking for something unusual with this mod, to make it what "gearmanndude" on youtube would call a "crazy fuzz box from hell".

i used to study basic electronics in college 10 years ago but never really got into it. i still remember a few things so i know enough to build something, to understand what is said about one circuit when explained "slowly", but not enough to be able to study or even less design a circuit by myself. i tried to dig in my old notes and a few books that i found (recommendations on MFOS about electronic theory and active filters) but i'm still very very rusty.
some things i do know are that every stage is indeed an inverting amplifier, with multiple feedback loops (caps, resistors and diodes) with a gain greater than 1.
4 of them in a row make the output signal in phase with the input, that's why i thought of the loop from output to input, since i'm looking for something a little "over the edge", i don't want a negative feedback to smoothe the effect but something to push it beyond the limits.
since, like you said, i might create an oscillator by feeding too much signal back into the pedal and i don't want that, i thought of putting a filter in the FB signal path, but since it's a variable resistor that sets the amount of signal that is fed back, i had the feeling that it might interact with a capacitor of the filter that may be in its path thus making the "feedback amount" and "cutoff freq" controls not independent from eachother, which i also don't want.

i also maybe should have said what i modded and worked, which are:
i changed the diodes for 2 ultrabright LEDs in the first clipping stage, and for 2 red LEDs in the second stage.
i bypassed the emitter resistor of Q3.
i replaced the tone stack with jack orman's presence control version 2.

i thought of adding clipping diodes in series with a variable resistor to ground (or 2 different pairs connected to a blend pot to shape the clipped signal) somewhere but i don't know where to put them (maybe right before the sustain pot...) and i think i would need to breadboard the whole circuit to try different points.

i was looking for a more "open" sound, less compressed, more dynamics, which really worked for me so far. but now i want to push it further.
i built a mosfet booster, i'll try to add it in front of the circuit like in the blackout effektors' musket to see if it works well for me, but i'm still looking for that really crazy mod that will make it "wild" but still playable!

[Frances Rhodes]

Can you post a schematic about what you are doing/trying to do?

ok, i'll try to draw something and post it here.

[TejfolvonDanone]

connect the emitter of Q3 with the emitter of Q2
(...)
i bypassed the emitter resistor of Q3.

So you also bypassed the emitter resistor of Q2. So you alter not only the bias but the gain of the stage. It is maybe the cause you don't really get any effect out of this mod.

and connect the collector of Q3 with the collector of Q1(...) the other one was making the pedal buzz...

As Mark said with this mod you also alter the bias of these stages. Try it with a DC blocking cap.

4 of them in a row make the output signal in phase with the input, that's why i thought of the loop from output to input, since i'm looking for something a little "over the edge", i don't want a negative feedback to smoothe the effect but something to push it beyond the limits.

For linear closed loops and feedback you have to be sure that there isn't a frequency which goes around in the loop is getting an amplification greater than 1 (as the basic mathematics suggest if it's negative then it is less than 1). So if the feedback is positive (as you try it) the feedback loop has to attenuate the signal more than the gain of the circuit. Else it will go into oscillation.

If you do a simple calculation where an amp has a gain of G and you feed back b-times the output to the input you get a gain of G/(1-G*b) where G*b has to be smaller than 1 (or you get oscillation). If b is frequency dependent (HPF for example) it will vary for each frequency. So e.g. G*b = 0.1 for low frequencies and G*b = 0.5 for high frequencies the overall sound will have an emphasis on the high frequencies.

Because it isn't a linear amplifier (and all the four stages have more than 30000 gain and 9V supply it can't be for guitar signals) it isn't easy to determine what is the actual gain for the whole amplifier. So i suggest some breadboard time. I don't even try to predict the outcome.
If you have an existing circuit the feedback isn't that hard to implement on a breadboard: just solder a three short wires to the output, to the input and to ground and connect these wires to the breadboard.

First i would try a feedback with a simple voltage divider with a big attenuation: 100k from output to a 1k pot's first lug and the last lug of it to ground. The middle lug of the pot goes to the input. Be sure it's DC decoupled with a big cap. It'll give you a minimum attenuation of about 99%. If there's an oscillation signal increase the 100k resistor. If it hasn't got enough effect decrease the 100k. I'm certain that you will run into oscillation a lot of times.
If you can settle with a good sounding feedback for the second experiment you can put a cap between the resistor and the pot (you get high pass filter like that). It will only increase the attenuation for low frequencies and don't touch the highs. In this configuration the overall cut of the filter won't move you just adjust the signal fed back. The cut off frequency of the HPF will be 1/(2*pi*C*Rs) (Rs was the 100k resistor in the begining).
You can create a "poor man's bass cut" type variable HPF but it will interact with the feedback pot. In the middle position of the bass cut the overall attenuation provided will be greater than in one of the end positions.

The "poor man's bass cut" is like in this schem formed by C2 C4 and the blend pot:
http://4.bp.blogspot.com/-nKhAbo_Px2M/T9pBTmfptxI/AAAAAAAABbg/2ND-qOm-5xY/s1600/Lovepedal+Silicon+Fuzzmaster+Rev3.GIF

I think the best way to try these things is to have an audio probe and just experiment. Maybe the overall outcome will be just unbearable. That's a likely conclusion.

Have fun experimenting and keep us updated. I'm really curios.

[Frances Rhodes]

connect the emitter of Q3 with the emitter of Q2
(...)
i bypassed the emitter resistor of Q3.

So you also bypassed the emitter resistor of Q2. So you alter not only the bias but the gain of the stage. It is maybe the cause you don't really get any effect out of this mod.

yeah, now you mention it, i can't remember if i bypassed the resistor before or after trying the loop mod... but in the case where one of the emitter resistors is bypassed, wouldn't it just also bypass the second resistor as well?

[Frances Rhodes]

ok, so, if i get you right, you're saying that i should wire the pot as a voltage divider (with a much bigger resistor on its first end to pad the signal) and not as a variable resistor? and then connect a filter to the center lug?

if the result is really really unbearable (which is what i'm looking for, sort of),i have other ideas:
1- i could try to feed only some of the bass frequencies to the first clipping stage, so the feedback would be negative.
2- i could put an active filter in the loop that will invert the filtered signal, like a simple sallen-key with frequancy and Q controls.
3- i could add a baxandall active filter (with maybe just mids and treble controls) with a tone switch to choose between the two stacks

but first things first, i need to try to add a booster in front of it to see if it sounds as good as "they" say

[TejfolvonDanone]

yeah, now you mention it, i can't remember if i bypassed the resistor before or after trying the loop mod... but in the case where one of the emitter resistors is bypassed, wouldn't it just also bypass the second resistor as well?

That's what i'm trying to talk about. 

ok, so, if i get you right, you're saying that i should wire the pot as a voltage divider (with a much bigger resistor on its first end to pad the signal) and not as a variable resistor? and then connect a filter to the center lug?

Yeah i should have said that in the first place. 
The output impedance is quite low and the input impedance is high for a stage. So with a voltage divider (and a huge padding) you know how much output you feed back.

I thought more about this thing and i came to a conclusion that you will defenetly need a buffer. If you feed back the last transistors output to the BASE (or input) of the first stage you will get a shunt feedback which decreases the input impedance. So with the a regular guitar plugged in it you may have a situation that you don't even get an output signal. So a buffer with really small output impedance (emitter follower) will be needed before the first stage (and the feedback).

[Frances Rhodes]

ok, but bypassing the second transistor shouldn't make a buzz, it should just give the second clipping stage more gain... i have a "virgin" big muff 9 that i want to keep unmodded, i'll try the feedback loop again and see if it makes it buzz too.

an other question, i know about stargrounding and when you open a big muff you see everything but all grounds connected to the same point.
i'm trying to replace every wire and specially every ground wire to change this but i'm still wondering, if i add a buffer or a booster in front of the circuit or in a loop, where shall i connect the ground of the second PCB? to the "star" or to the DC jack, in order to prevent ground looping?