Project Chameleon: A very flexible od / dist / fuzz project

[Fancy Lime]

Hi there,

some time ago, when looking at a bunch of DOD overdrive and distortion circuits from the 80's and 90's, I thought to myself that what they did, namely making fairly generic PCBs that could be used for many related but distinct circuits, would be even more useful for a DIY project than it is for production line pedals. Not that no-one has thought of that before but I have never had a crack at developing a "generic multi purpose drive". So I went and tried to find out what topology seems to me to be the most flexible for a variety of different distortion flavors. The fruit of my labors are tentatively called "Project Chameleon":




Explanations and mods

Opamp
The circuit relies heavily on opamp distortion. Opamp choice therefore matters. With the NE5532 shown in the schematic, you get a nice bluesy overdrive at low gains, solid distortion in the mid gain range and an increasingly compressed, pretty angry fuzz sound at the highest gains. With a NJM2068, the low gain overdrive is even better (absolute Blues Driver killer in my opinion). At medium to higher gains you get more of a modern distortion sound without the fuzziness of the NE5532. For bass I prefer the NE5532 (NJM2068 is a bit harsh here). For guitar I would go with the NJM2068 for blues, classic or hard rock, and most metal sub-generes, but with the NE5532 for alternative or stoner rock, doom/sludge/drone metal, and so on. You could call the NJM2068 the "tight" option and the NE5532 the "loose" one. I did not test other opamps but hope to do so in the future.

Diodes
All diodes are optional. I love me some crossover distortion (D1, D2) as a switchable option but many people will probably leave those diodes out. Try them, though, if you are unsure. Very nice touch at higher gains and with bass use, in my opinion. Plus, these also act as a noise gate, which can come in handy.
The diodes to ground (D3, D4, D5) increase the distortion massively. Especially with the NE5532 and all three of those diodes connected to ground (symmetric clipping), you get a very thick modern fuzz. The asymmetric option (only D3 and D4 connected to ground.), you retain more dynamics and volume but still not nearly as much as without any diodes. I like to use a Z-switch here (so I can have the order: 1 no diodes; 2 asymmetric; 3 symmetric) but a DPDT on-off-on works, too.
The cap before the diodes (C6) keeps the diodes from messing up the bias of the second opamp half and also limits the frequency range that is clipped at all. R6 and C6 form a low pass filter that only lets frequencies above 72Hz clip fully. Anything below that is clipped progressively less. That makes no audible difference for a standard tuning guitar but means that the lowest octave of a bass comes through with much less clipping and more volume, which sounds very nice indeed. This effect is much more obvious with symmetrical clipping than with asymmetrical. Lower tunings on guitar also get a bit of a bottom boost from this, which to me seems a good thing for the genres, where one would tone ones guitar lower.
I experimented with clipping diodes in the feedback loop of the first opamp stage, including all variants of MOSFETs as clipping diodes. While several combinations sounded pretty good, they pale in comparison to the sound of the opamp clipping. But that is just my taste, go nuts with all the diodes if you like.

R2
R2 is a current limiting resistor that is not necessary with BJT opamps (and can even be detrimental to the noise performance) but may be a good idea with JFET input. It is mostly in here as a place holder, you can leave it out.

Controls
The Gain and Volume controls should be fairly self explanatory. Presence is effectively an additional gain control for just the highs, boosting at around 3400Hz when fully open. The Tone control is a Rat style high cut filter, simple but effective. So basically, we have a variable high boost before the clipping and a variable high cut after the clipping. This allows very effective control of the overall frequency response as well as the "flavor" of the distortion.


Cheers,
Andy


P.s.: This is work in progress. There will be some changes over time and hopefully a PCB soon. If I get around to it...

[Mark Hammer]

1)  Nice work.  Kind of a "greatest hits".  Made me recall the contributions of our former member, the late Ricky Don Vance.

2) T'wer I, I'd opt for Schottky diodes for D1/D2.  You'll lose less level, and retain more body.

3) What made you choose the components for D3-5?  They seem to have Vf that is conspicuously higher than what a 9V supply might be capable of.

4) Got me to thinking about the Bluesbreaker/KOT architecture.  As you may recall, the gain pot simultaneously alters the gain of stage 1 and 2 in that circuit, however all the clipping is done in stage 2.  I wonder if the same dual-gain-adjustment trick might be used to affect gain in an earlier vs later stage.  In other words, stage 1 might have some potentially clipping diodes in the feedback loop, and stage 2 would have them going to ground at the stage 2 output.  The gain could be continuously applied to stage 1 for "softer" clipping, or stage 2 for "harder" clipping.  I'm going to have to mull that one over.

[Fancy Lime]

Hi Mark,

thanks! Yes, it is indeed a kind of mashup between the most popular opamp distortions in terms of topology and voicing. I'd say it gets most of it's DNA from the Rat, Zendrive, and OCD, but a dozen other circuits are not much farther off either.

I forgot to discuss the diode choice, dammit.

2) Normally I indeed use BAT41 Schottkys in the D1, D2 position. However, with no diode clipping before these, the level drop is not really a issue and I kind of digged the harsher clipping of the 1N4148's in this particular circuit. But this is definitely a place for extensive experimentation. BAT41's are usually my favorite, but I have used BAT42, BAT46, 1N5817, and even 1N4001 for this. The resulting sounds are very different.

3) The Zener voltage does not really matter. As long as it is larger than the forward voltage of the blue LED (3-ish V), the Zeners will never reverse open. So the Zeners act as normal silicon diodes here. However, the sound seems subtly different. A bit harsher. I may be imagining things though. 1N4148 or 1N4001 will sound very similar here but not completely identical. Experimentation highly recommended.

4) Having 2 gain pots for 2 different clipping stages is great. It's what I did here:
https://www.diystompboxes.com/smfforum/index.php?topic=118391.msg1144853#msg1144853
"Gain" controls the feedback clipping and "volume" controls the MOSFET clipping (the actual volume control is called "master"). Very flexible. Probably even more so when done to something where the clipping stages sound more obviously different, like in your example. Kind of like a Tube Screamer pushing a Rat... Nice idea for a 2-channel drive pedal, no?

Cheers,
Andy

[paul.creedy]

Your project sounds way more sensible than my recent idea.

On discovering the 1590DD enclosure I decided it might be fun to see how many fuzzes I could cram into it.

I won't mess up your thread with my nonsense though, I'll start my own

[Mark Hammer]

Yes, sometimes there CAN be "too much of a good thing".  One of the things I've found over years of building is that I am more easily persuaded that I can easily forfeit this or that feature because, after all, I can simply make them another pedal.  No mo' FO-MO. 

[Fancy Lime]

I pondered the curiously large difference in sound between the NE5532 and NJM2068 in this circuit and thought I remembered that the NE5532 had a largish bias current. So I went and replace the bias resistor R5 with a 330k resistor and the result was very similar to the sound with the NJM2068. A lot less fuzzy, smoother, more like the typical classic rock stack-in-a-box sound. I also went the other way, obviously, trying a 10M. Indeed the sound gets more fuzzy and sickly at maximum gain. Seems like the opamp hitting the rails really messes with the bias in an interesting and pleasant sounding way. The variation is large enough and musically useful enough to me to warrent a switch that lets you select the value for R5. If a buffered bypass is used (which my builds of this will), then one can easily go as low as 100k. So I would install a 3-way switch with 100k, 1M and 10M.
Only after this experimentation did I check the data sheets to find that the bias current of the NE5532 is only slightly higher than that of the NJM2068 (200nA vs 150nA typical). Curious. Anyway, I have yet to test the influence of R5 on the sound with an NJM2068.

I also checked out variation of supply voltage from 6V to 18V. Nothing unexpected here. The opamp clipping starts earlier at lower voltages, whereas higher supply voltage gives you more headroom (well, duh!). An interesting side effect is that the "apparent bass boost" created by R6 when the clipping diodes are engaged, gets more pronounced when the opamp itself clips less. So at high gain settings with the diodes in, you get more bass at higher supply voltages than at lower ones. Nice extra option, that.

Andy

[idy]

I'm still not getting how the second opamp stage is biased... It's DC coupled to stage one, except for those x-over diodes... The first stage you will usually see VB at the output, isn't one of the diodes conducting then? Or that high impedence input to stage two says "no dice" (no current draw?)  and the diodes shut the signal off...?

I see how it would gate if both sides are biased the same way, only switching on when the difference between the output of stage one sways a diode drop above or below VB...

Thanks for putting this out there!

[Fancy Lime]

I'm still not getting how the second opamp stage is biased... It's DC coupled to stage one, except for those x-over diodes... The first stage you will usually see VB at the output, isn't one of the diodes conducting then? Or that high impedence input to stage two says "no dice" (no current draw?)  and the diodes shut the signal off...?

I see how it would gate if both sides are biased the same way, only switching on when the difference between the output of stage one sways a diode drop above or below VB...

Thanks for putting this out there!

Hi idy,

thanks for the question! I wanted this to be a bit of a demonstration project for several techniques I find useful, so I'm glad you asked because it gives me the opportunity to explain one of them.

The short answer: Diodes are diodes, except when they are not.

The long answer: If D1, D2 are bypassed, the situation is simple. There are 4.5V at the output of the first opamp stage; this is DC coupled to the input of the second stage; which is therefore biased at 4.5V. With D1, D2 in the signal path and not bypassed, things get a bit more complicated but not too much. Let's assume that at some point in time, the voltage on the input of the second stage were far away from 4.5V, for example 0V, and that the AC signal suddenly stops. The output of the first stage would return to 4.5V. That means there is a voltage of 4.5V across D1 and D2. That would cause D2 to open until the input of the second stage is one diode drop (say, 0.7V) below 4.5V, so about 3.8V.
If the AC signal were to start again now immediately, the second stage would be a bit mis-biased but not enough to keep it from functioning properly. Remember, the maximum signal swing that the second opamp can ever see when D1, D2 are engaged, is the maximum swing of the first stage minus the diode drops of D1 and D2 on both sides. So mis-biasing by exactly those diode drops is exactly the maximum of mis-bias we can allow without having problems such as additional distortion in the second opamp. Neat, huh?
Of course that is not 100% true because opamp distortion is not 100% "sharp". There is a little bit of a knee to it but not enough to matter in practice. So far, our diodes have behaved like an ideal diode should.
But wait, there is more. If the AC signal does not start exactly when the input of the second stages reaches 3.8V but later, then the input of the second stage will "slowly" (meaning over a matter of milliseconds*) return to 4.5V because the diodes have a thing called "leakage current". There is forward and reverse leakage, the latter of which is larger for many (most?) common diode types. The fact that diodes leak means nothing other than that they act as resistors when they are reverse biased or below their conducting threshold. Resistors with pretty large resistance (and non-ohmic ones at that), but resistors nonetheless. A theoretical ideal diode has no leakage but real devices do. And 1N4148's have a pretty substantial leakage. There are low-leakage diodes, and I would not really recommend them for the D1, D2 position. The inherent diodes of silicon bipolar and JFET transistors can have pretty low leakage, Ge-diodes usually have very high leakage. Schottky and low-voltage Zeners (<5V) tend to have relatively high leakage as well. Your standard garden-variety Si-diodes are kind of in the middle.
The combination of (fast) diode conduction as long as the voltage difference is >0.7V and (slower) leakage conduction when the difference is <0.7V makes the input of the second stage always return to the same voltage as the output of the first when there is no AC signal present.


Hope that helps. Don't hesitate to ask again if my wordy sermon was unclear (that happens a lot in my explanations, I'm afraid...) or if you have other questions.

Cheers,
Andy



*I'm too lazy to calculate the exact time constant because I would have to measure the leakage current of the diodes I used first. My be faster, may be slower, does not really matter in practice. The point is: It's still fairly fast.

[phasetrans]

So I went and replace the bias resistor R5 with a 330k resistor and the result was very similar to the sound with the NJM2068. A lot less fuzzy, smoother, more like the typical classic rock stack-in-a-box sound. I also went the other way, obviously, trying a 10M. Indeed the sound gets more fuzzy and sickly at maximum gain. Seems like the opamp hitting the rails really messes with the bias in an interesting and pleasant sounding way. The variation is large enough and musically useful enough to me to warrent a switch that lets you select the value for R5. If a buffered bypass is used (which my builds of this will), then one can easily go as low as 100k.

Any idea what increasing the base current at the input is doing here to the opamp sound? This is an effect I would like to understand conceptually for myself.

[Fancy Lime]

So I went and replace the bias resistor R5 with a 330k resistor and the result was very similar to the sound with the NJM2068. A lot less fuzzy, smoother, more like the typical classic rock stack-in-a-box sound. I also went the other way, obviously, trying a 10M. Indeed the sound gets more fuzzy and sickly at maximum gain. Seems like the opamp hitting the rails really messes with the bias in an interesting and pleasant sounding way. The variation is large enough and musically useful enough to me to warrent a switch that lets you select the value for R5. If a buffered bypass is used (which my builds of this will), then one can easily go as low as 100k.

Any idea what increasing the base current at the input is doing here to the opamp sound? This is an effect I would like to understand conceptually for myself.

No idea, really. With high values for R5, there seems to be a gating effect during the loudest notes when the signal slams into the rails. I therefore suspect that hitting the rails upsets the bias voltage at the inputs of the opamp internally. if R5 is small, then the bias current is high enough to force the bias back to the right point. But if R5 is large, than that takes long enough to be audible on audio frequencies. This is, however, no more than a half-educated guess. I'd love if someone could explain it properly.

Interestingly, changing the input cap C4 has the opposite effect from what I would have expected at first based on the above speculations. A smaller cap (but still large enough to keep the C4-R5 high pass cutoff below the audio band) makes the "mis-biasing effect" more pronounced with large values for R5. With a 470n cap at C4 it is almost gone, even with a 10M R5. Seems like the large cap stores enough charge to keep the input from going mis-biased in the first place. With small values for R5 (say 100k), where there is no mis-biasing effect anyway, the influence of C4 is just what one would expect, so using small cap values here can be used to tighten the sound by cutting bass before the clipping section. C4 and R5 should be experimented with on the breadboard before building. I am considering making one or both of those switchable on the PCB.

Cheers,
Andy

[Gus]

Have you seen this?

http://www.gmarts.org/index.php?go=217

A web page that has been around for some time

[rankot]

How could I've missed this???

[Fancy Lime]

Have you seen this?

http://www.gmarts.org/index.php?go=217

A web page that has been around for some time

I was unaware of that one. Great resource, very educational and written in a way that should be easy to understand even for the uninitiated. A similar article that I enjoyed reading several times when starting this hobby, is this one on GGG:
http://www.generalguitargadgets.com/how-to-build-it/technical-help/articles/design-distortion/
Although I seem to remember that it was called "Cook your own distortion" back then. Looking back at it now, that is certainly not too different from the Chameleon.

I hope I'll get a few sound samples done and implement the latest changes into the schematic over the holidays.

Cheers,
Andy

[Fancy Lime]

Due to some recent insights into noise sources, the input stage will get a substantial overhaul.

Andy

[ThermionicScott]

Explanations and mods

The cap before the diodes (C6) keeps the diodes from messing up the bias of the second opamp half and also limits the frequency range that is clipped at all. R6 and C6 form a low pass filter that only lets frequencies above 72Hz clip fully. Anything below that is clipped progressively less. That makes no audible difference for a standard tuning guitar but means that the lowest octave of a bass comes through with much less clipping and more volume, which sounds very nice indeed. This effect is much more obvious with symmetrical clipping than with asymmetrical. Lower tunings on guitar also get a bit of a bottom boost from this, which to me seems a good thing for the genres, where one would tone ones guitar lower.

Sorry (not sorry) for the necro-bump, but I happened along this thread and this part piqued my interest.

With overdrives/distortions, a common theme is to cut bass before the clipping in order to prevent intermodulation distortion and other unpleasantness.  But if we can just selectively clip the midrange and higher frequencies, it would seem that we could avoid a mid-humped sound.  Is there a catch to this?  Why don't we see this used more often?   

[iainpunk]

Explanations and mods

The cap before the diodes (C6) keeps the diodes from messing up the bias of the second opamp half and also limits the frequency range that is clipped at all. R6 and C6 form a low pass filter that only lets frequencies above 72Hz clip fully. Anything below that is clipped progressively less. That makes no audible difference for a standard tuning guitar but means that the lowest octave of a bass comes through with much less clipping and more volume, which sounds very nice indeed. This effect is much more obvious with symmetrical clipping than with asymmetrical. Lower tunings on guitar also get a bit of a bottom boost from this, which to me seems a good thing for the genres, where one would tone ones guitar lower.

Sorry (not sorry) for the necro-bump, but I happened along this thread and this part piqued my interest.

With overdrives/distortions, a common theme is to cut bass before the clipping in order to prevent intermodulation distortion and other unpleasantness.  But if we can just selectively clip the midrange and higher frequencies, it would seem that we could avoid a mid-humped sound.  Is there a catch to this?  Why don't we see this used more often?   

because it can get even muddier when stacking this into another distortion, due to the huge low end still being available. and it only works as long as the opamp itself doesn't clip.

a rule of thumb i like for opamp clipping is that for most common guitars, at A=10 (20dB) the opamp clips the sustain slightly, at A=4 (12dB) the opamp clips the attack but leaves most of the sustain clean.
this dictates that this will work best with low gain settings on the opamp, limiting the effective range of this function

cheers

[Fancy Lime]

Well, the most famous pedal that uses the same basic trick is the Big Muff. That's why lowering the value of the feedback caps in a BMP makes it such a bass monster. But as Iain said, theis can get out of hand fast, so for best results it needs to be balanced with some bass cut. If done right, you can design something that does not get muddy of fuzzy from intermodulation distortion but maintains a fat low end. The way I adjusted it for the chameleon means it is only reall effective at sub-guitar frequencies, to make the whole thing usable with bass without needing a clean blend. The cap is effectively a low frequency clean blend.

For the noise reduction overhaul I promised way too long ago (sorry!): Just use a TL072 instead of a NE5532. Should be quieter in this situation.

Andy

[iainpunk]

i personally don't mind a bit of intermodulation muddiness, but i won't miss it either.

reminds me, i should experiment with bridge-T based selective clippers.

cheers

[ThermionicScott]

Well, the most famous pedal that uses the same basic trick is the Big Muff. That's why lowering the value of the feedback caps in a BMP makes it such a bass monster. But as Iain said, theis can get out of hand fast, so for best results it needs to be balanced with some bass cut. If done right, you can design something that does not get muddy of fuzzy from intermodulation distortion but maintains a fat low end. The way I adjusted it for the chameleon means it is only reall effective at sub-guitar frequencies, to make the whole thing usable with bass without needing a clean blend. The cap is effectively a low frequency clean blend.

For the noise reduction overhaul I promised way too long ago (sorry!): Just use a TL072 instead of a NE5532. Should be quieter in this situation.

Andy

Fascinating!  I knew the caps were needed to block DC, but it didn't dawn on me before that they would also affect which frequencies get passed through. 

I agree that some bass cut might still be desired.  Gotta try this on the breadboard, thanks!