Determining [unity] volume for parallel fuzzes to be blended.

[Andon]

Howdy. I'm working on a dual fuzz pedal that will have a comparator fuzz and an op amp big muff-style fuzz in parallel, and rather than having two independent volume knobs I'd like to have a blend knob between them. I went through a myriad of blend circuits (ROG Splitter-Blend, panner/blend from R.G. Keen's "Panning For Fun", etc.) before settling on the "color" knob setup from the Boss OS-2, as it provided the most consistent results (no volume loss toward noon from either side, relative ease of build, and then some). Here's the schematic for what I've breadboarded so far:



Full size image available here: https://i.imgur.com/pRd8l0i.png

Also, for reference here's the OS-2 schematic sans input buffer and flip-flop/bypass, provided by Electric Safari.

In a nutshell, I designed a comparator fuzz (with some help from a recent thread on here - thanks to antonis and anotherjim!) and put a BMP tone stack afterward, and the other side is an op amp big muff-style circuit (that I'm still flexing with to adjust the overall sound, but the stock components are listed for the sake of showing what I have) with a RAT-style tone control (in reverse) after it. This gives the comparator fuzz a mid scoop, and the op amp big muff-style circuit a mid boost.

There's no input buffer on the schematic, though one is planned, and the top right and bottom left portions of unconnected circuits are an optional boost and the power section, respectively.

What I am trying to solve for here is how to set (and calculate) the gain for each fuzz. I know on the OS-2 that R8 and R20 (on my schematic, but R2 and R13 on the original schematic) set the gain level for the distortion and overdrive circuits, but I don't know how to properly calculate what value of resistors I would need for them.

I  think I understand how to calculate op amp gain, but the standard calculations I can find have certain assumptions for components like feedback resistors that I'm not using for the comparator fuzz (and subbing in a low number like 1 ohm doesn't tell me much), and when it comes to the op amp big muff-esque style fuzz I know how to calculate each individual op amp section (first section has a gain of roughly 6, second section has a gain of roughly 9, etc.) but I don't know how to put them together to get an overall idea of gain. Conventional knowledge from ElectroSmash and elsewhere tells us that a standard Big Muff has an overall gain of about 60dB, but again I don't know how to take that information and apply it to what exact resistor values I would need.

I've found that a 100k resistor at the end of the comparator fuzz circuit sounds roughly around unity gain, but plugging and chugging on an already noisy breadboard (lotta' wires to connect everything) is tedious and I would like to find an more concrete way of solving this dilemma. Really, if I could just get both sides of the blend close so that the level stays as close to the same as possible across the entire sweep that would be preferred since I can adjust the overall volume at the end.

The color/blend circuit from the OS-2 is the same as stock, save for replacing the color/blend knob with a 50k instead of 20k, and omitting a 1nF capacitor in the negative feedback loop of U4 (U4B) to increase bass response per Electric Safari.

So, I know some things, and I know what needs to be fixed, but I'm missing some steps to get there. Any thoughts?

[iainpunk]

you have the terms 'gain' and 'volume' mixed up a lot here.
but i think that you want the two fuzzes to have the same volume level so that when you mix you keep the level the same. right?
i'd say put a trim pot where either R8 or R20 is, (whichever is the loudest gets the trim pot) and attenuate to taste.
another problem is the difference in volume on different tone control settings, both have this but the rat control is more agressive at taking out volume level depending on setting, where the big muff has a higher static insertion loss (it loses about 8db independent of the control's setting)

cheers

edit: look at this thread, my posts there have some relevance here:
https://www.diystompboxes.com/smfforum/index.php?topic=127355.msg1221174#msg1221174

[Andon]

you have the terms 'gain' and 'volume' mixed up a lot here.
but i think that you want the two fuzzes to have the same volume level so that when you mix you keep the level the same. right?
i'd say put a trim pot where either R8 or R20 is, (whichever is the loudest gets the trim pot) and attenuate to taste.
another problem is the difference in volume on different tone control settings, both have this but the rat control is more agressive at taking out volume level depending on setting, where the big muff has a higher static insertion loss (it loses about 8db independent of the control's setting)

cheers

edit: look at this thread, my posts there have some relevance here:
https://www.diystompboxes.com/smfforum/index.php?topic=127355.msg1221174#msg1221174

Correct, I'm using "gain" here to refer to the signal going into the blend, not out of it (volume), so that the end result (output) will be uniform (I've always read that "gain" = input, "volume" = output, which is why I used it when referring to the blend pot, but I could also be wrong!) "Sustain" is what I have for controlling the level of distortion on both circuits.

Shucks. I was hoping to avoid a trimpot, but I suppose I'll have to give it a go if I can't figure this out.

[iainpunk]

i don't understand where you get this from, but you have your terms mixed up heavily.
gain is the amplification factor. a signal doesn't have gain, a circuit does have gain. "sustain" is the same as "gain" in most cases.
'volume' is 'signal level', independent of where in the circuit it is.
you want to match the 2 levels so the output level is the same-ish.

you might also want to take a look at the phase of the two circuits. i believe the bottom one keeps the signal in phase, but the comparator side flips it around, having signals out of phase might also cause things to sound thin/shrill

and you might want to add buffers between the tone stacks and the mixer.

cheers

[ElectricDruid]

+1 what Iain said:

The fundamental issue is that the levels out of your two fuzz circuits aren't constant. They'll alter depending on tone control settings, and (to a lesser extent) the Sustain settings. The phase is also important, and you need to make sure that the two outputs are the same phase when they get to the mixer. Out of phase signals will cancel the fundamental in the mixer and leave you with just the "fizz".

Since it's going to be pretty much impossible to make a Blend control where the point at which the levels are 50%/50% is always in the centre (sometimes it'll be off to the left, sometimes off to the right), I'd instead focus on making sure the control has a good range so that you can always easily find the mix you need, without focussing too much on exactly what that is or whereabouts on the dial it is found.

[antonis]

gain is the amplification factor.
"sustain" is the same as "gain" in most cases.

Hmmmmm...
(you like to confuse terms unfamiliar people, don't you..??)

"Sustain" refers on a given gain circuit incoming signal level..!!

[antonis]

Since it's going to be pretty much impossible to make a Blend control where the point at which the levels are 50%/50% is always in the centre (sometimes it'll be off to the left, sometimes off to the right)...

A servo Constant Volume FET amplifier for each effect output could turn impossible into obtainable..

[Andon]

I appreciate the insight, y'all. I actually had JFET buffers (a la what you'd find at the end of the RAT schematic before the volume knob) on each circuit for the exact aforementioned reasons but took them out on this latest iteration when I saw that they weren't being employed on the OS-2 build (there's a transistor on the top path of the original schematic but I couldn't tell it's exact function). I'll add them back in and see what's up.

EDIT:

Since it's going to be pretty much impossible to make a Blend control where the point at which the levels are 50%/50% is always in the centre (sometimes it'll be off to the left, sometimes off to the right)...

A servo Constant Volume FET amplifier for each effect output could turn impossible into obtainable..

You're referring to something like this, correct?

[ElectricDruid]

Since it's going to be pretty much impossible to make a Blend control where the point at which the levels are 50%/50% is always in the centre (sometimes it'll be off to the left, sometimes off to the right)...

A servo Constant Volume FET amplifier for each effect output could turn impossible into obtainable..

Actually, that's quite a nice touch. It might work rather well. Adds a few parts, but it's not outrageous.

[Andon]

Alrighty, made some revisions:



Full size image available here: https://i.imgur.com/VYv8v7j.png

Updates are annotated with blue text. I added the aforementioned input buffer and a phase inverter for the comparator fuzz side, as well as a constant volume amplifier that I'll plug into each side after I get some confirmation on the design. Here's the schematic I found for the constant volume amplifier, taken from the Making Circuits link I provided in my last post:



And here's my recreation of it in KiCAD:



I'm using a BF244A in the schematic because my library doesn't have a 2N3819, though I have a few on hand, and I'm using a dual op amp because if this design works then I'll just duplicate it for the other side of the amp so it can pull double duty.

The original drawing from Making Circuits isn't super clear on a couple of points, namely my assumption that all of the grounds won't need to connect to the input or output along the bottom track of the original, and the suspect node pointed out with the red arrow. Is the original drawing saying that there is a negative feedback loop, and that there are two ouputs each going separately to R2 (10K) and R3 (330K), or is there supposed to be a junction as I placed? I'm assuming the former, but the original drawing is a little unclear with what looks like up to three tracks joining at the output of the op amp.

Also, is 1M an appropriate guess for the value of R1 (on the original Making Circuits schematic) for both sides of the fuzz? The article mentions the value is dependent upon the "optimum expected amplitude" of the input signal, but perhaps for the comparator side it should be 100K since I found that to be comparable to unity per my first post. 

[ElectricDruid]

That joint highlighted with the red arrow shouldn't be a joint. In the original schematic the wires cross, but don't join (no dot on the junction). It goes straight from the output of the op-amp up to the top of the 330K. Doesn't pass go, doesn't collect $200.

[Andon]

Right, that's what I figured since there wasn't a junction (like I said I figured it wasn't the case but I had it that way in KiCAD until I could discern the connections) but how does the 10K connect? To both the inverting input and the output as well as the bottom of the 330K, or just from the output to the bottom of the 330K since the negative feedback loop wire is a thinner line?

[ElectricDruid]

how does the 10K connect? To both the inverting input and the output as well as the bottom of the 330K, or just from the output to the bottom of the 330K since the negative feedback loop wire is a thinner line?

I think it goes from the inverting input to the junction of the 330K and the 1u. The thin/thick line is just a screw-up in the drawing.

Also I think the electrolytic caps are all drawn the wrong way up in the original schematic. Someone's used the black-white symbol instead of the two-lines-with-plus-sign symbol, and got the colours the wrong way around (white is positive).

[Andon]

how does the 10K connect? To both the inverting input and the output as well as the bottom of the 330K, or just from the output to the bottom of the 330K since the negative feedback loop wire is a thinner line?

I think it goes from the inverting input to the junction of the 330K and the 1u. The thin/thick line is just a screw-up in the drawing.

Also I think the electrolytic caps are all drawn the wrong way up in the original schematic. Someone's used the black-white symbol instead of the two-lines-with-plus-sign symbol, and got the colours the wrong way around (white is positive).

Thanks! Alrighty, so something like this?



I had to flip everything after the output along the horizontal axis so that I could make the connections on the schematic, so apologies that half of it is now "upside down" compared to what it was.

Also, I sort of understand what the FET is doing from reading the article (slightly boosting the volume when it's lower than the input signal, and attenuating it when it's higher), but is there any way that this could be done with the other half of the op amp or is it dependent upon the given arrangement? I get what it's doing insofar as what I've read about it, but the "how" is sort of lost to me at the moment.

[Andon]

Hm, no sound from the constant volume amplified circuit, but I did find where the Making Circuits article referenced their drawing from:

Marston - 110 Operational Amplifier Projects For The Home Constructor

Pages 36-37 of the PDF (30-31 of the document itself).



Looks like the polarized capacitor orientation was right, though the first one appears to have been added in (input capacitor, I'm guessing), and the capacitor between the 10K and 330K resistors is non-polarized in the original drawing.

The new original drawing also shows the input, output, non-inverting input, Q1 source, R5, and C1 all go to 0V, but pin 4 of the op amp goes to -9V (battery powered, I assume) - should I be tying these all somewhere else besides ground, or biasing the input? I'm also a bit lost on where the output connects (the op amp output pin obviously, but there's so much going on after it).

[ElectricDruid]

Good find! Yes, looks like I was wrong about the polarised cap, sorry. I hadn't noticed that the original circuit is on a bipolar supply, which explains how the cap can be upside down.

This is getting complicated. If we're to make this work, we have to convert the circuit from bipolar power to unipolar, which is typically going to mean changing the circuit from +/-9V to +9V. That means that things are getting half as much juice as previously, and that could well affect the choice of FET and its biasing.

The basic gist of "how it works" is that it's an inverting amplifier. The op-amp will try to keep its -ve input (pin2) at the same voltage as the other input, which is tied to ground. Hence if an input comes down R1, an equal and opposite input needs to come down R2 to cancel it and maintain the pin at ground. Usually that equal and opposite signal comes direct from the output, and if R2 is larger than R1 then the signal at the output has to be larger too. That's how the basic inverting amp works.
In this case, we've got a voltage divider on the output made by R3 and the FET. This *reduces* the amount of signal that is fed back by R2, which means that the op-amp has to produce *more* output to get enough signal back to pin 2 to keep it at ground - e.g. it changes the gain.
Now, that FET's resistance will vary depending on the voltage on its gate, and that is set by a very roughly rectified (D1) and filtered (R4/C1) "envelope follower" signal which gives a indication of the overall level of the input. So as the output goes up, the FET's gate gets turned off more, increasing its resistance and increasing the amount of feedback, so decreasing the gain. As output level goes down, the reverse happens; FET gets turned on more, feedback level goes down, op-amp increases gain to compensate.

(It's a compressor circuit, in fact, and of the feedback type, rather than feedforward.)

[Andon]

Excellent explanation of what I was very loosely grasping - thank you! The literature even goes on to mention that you can put two of these circuits back-to-back for even more compression. And yeah, this is getting pretty complicated ha ha. I also found another constant volume amplifier - also designed by Ray Marston - that uses a couple of 2N3904s in lieu of the op amp configuration:



Nuts & Volts Magazine (June, 2000)
FET Principles And Circuits — Part 2 By Ray Marston

You'll find it on Figure 18 under "Miscellaneous JFET Circuits" This design appears unipolar, though the suggested power supply is "+12 to +20V". It's also got almost a dozen more components than the op amp-based constant volume amplifier.

Digging more I also found this from 1995:



Electronics Now (March, 1995) Active Audio Filters for Singal Processing and Control - Ray Marston

The drawing on the last page actually almost perfectly matches the initially posted constant volume amplifier design, including the input capacitor. So, in '75 they suggest an op amp configuration, in '95 the same but modified, and in '00 it's switched to a transistor-based design (though perhaps only for the sake of topic).

As much as I appreciate all that I've learned about this in such a short amount of time in this thread, and as much as I want the circuit to work how I had imagined it with two separate fuzzes with separate "sustains" and "tones" for each side, maybe Occam's Razor should play into this and I should just eschew that idea to follow the conventions established by the OS-2 and put the "sustain" and "tone" after the blend? 

[Andon]

No sound from the transistor-based drawing, either. I put a Big Muff in front of it and cranked it to see if I'd get anything and it was barely audible at best. Further reading suggests that you need a decently loud signal coming through the input for it to really be effective, which I think would explain the op amp in the original design. I also found this:

https://www.qsl.net/w2aew/youtube/audioleveler_sch.pdf

This design just uses 2N3904s for the transistors which I have plenty of, though I don't have any 1N34As diodes on hand at the moment. Is 12v really necessary here? I'd rather not have to go through the trouble of building a charge pump only to then have to regulate it back down to 12v.

[ElectricDruid]

If you're going for Occam's razor, I'd drop any attempt at the constant-volume malarky and just implement a basic blend control that can give you enough range to cover any variations. The constant-volume circuits are going to have other side-effects that we haven't discussed like squashing any remaining dynamics out of your signal (and fuzzes are bad enough for that already).

I can see how you can put the tone control after the blend (e.g. alter the tone of the mixed signal, instead of the two fuzzes separately) but I don't see how you can put the sustain control after (that's the gain of the fuzz stages, isn't it?).

I'd go: Input buffer, split to two fuzzes, each with gain+tone controls, Blend control, out.

[Andon]

Sorry, you're right, the "sustain" was shared on a dual 270k pot on the original OS-2. Definitely leaning toward my initial idea again, I just wanted to exhaust any other options I had (and it was fun learning about these constant volume circuits, too) before going with that.

Would increasing the blend pot give more control over the variance between both sides? I was thinking 50K over  the 20K on the OS-2 (I've got 100K on the schematic I posted above, which can be disregarded). For example, I know typically increasing the value of the "sustain"/gain pot will give you more distortion but less overall volume, if I'm not mistaken.