Thumb Sucker Compressor distorts instead

[matopotato]

Hi,
This is a separation from @Wook22's thread https://www.diystompboxes.com/smfforum/index.php?topic=129191.0 that I felt I was hogging too much for my own problem.
The symptoms are quite similar, and the builds are quite similar.
My original problem was that when I turn up Ratio towards 3 o'clock and Threshold around 9, there is a distorted sound. Not quite like a good distortion pedal but imho worse.
I started a thread for help here https://forum.pedalpcb.com/threads/thumb-sucker-compressor-scrapy-sound.10516/ back in early February and I got some good hints that I tried out. But the problem remains. The link also have photos and a sound sample of what I mean by distorted sound.
There is also a sense that Attack and Release does very little, but that could be a secondary fault to the first one.

Then it went in to hiatus until I found @Wook22 had a similar issue, so I piggy backed that thread for a while, but now it's time to make a separate thread.
The schematics are here

And if you want the pcb and zoom in the schematics the build pdf is here https://www.pedalpcb.com/docs/ThumbSucker-PedalPCB.pdf
And not to worry, although there are pictures in the linked other forum I will produce new ones.

I have recently re-heated all solder joints.
I have traced the connectivity on legs as much as possible trying to avoid direct solders when possible.
I tried to measure the components on the board, which checked out except for:
Threshold B1M
Ratio A1M
R101 and R102
D100 hard to get some sensible reading.
I swapped out C5 and C6 from Electrolytes to Film MKT. Forgot I had and swapped back to Electrolytes despite I had then learned it should not matter. And it didn't. Wish I hadn't swapped back. (And had better memory.)
I installed a trim for R10, and it is set to ~1k
The VREF and VCC plus ground all measure ok.
All reading further down.

What I have done:

Swapping TL072s and LM13700
Since there are two in this build I can swap them. No change. Later on I measured pin 7 on IC2 and it was zero, swapped again and now pin7 on the new IC2 got a reading. And so did the other pin7. Probably irrelevant or not pushed hard enough in the socket. Just mentioning.
I also got an opportunity to check another LM13700 from my friends Engineer's Thumb build and no change in my build regardless of LM, and his build sounds just fine.

Swapping the 2N3906
I tried with other PNPs and with an old Germanium one the distortion is much less. But not gone. The hfe of this Ge was quite low, so I doubt it is a "fix", but more of reducing symptoms.

Trying two JRC 4562
Supposed to be less noisy, I tried with these and it reduced the distortion. In combo with the PNP. Plus using 18V I felt that the problem was almost enough "smoothed out", but turning some knobs the distortion appeared again but with other settings. Not as loud, but still there. So this also feels like a way to lessen the symptoms without finding or fixing the problem.

Probing
It is clean until I hit IC1. Pin 3 is fine, but pin 2 and 1 (more loud) are distorted, as well as the C2 after. So stopped there.
The Threshold pot Pin 3 is fine, pin1 to VREF is quite silent and pin 2 sort of ok. IC2 pin 3 fine, pin2 slight sparkle, pin 1 bad, but nowdays very bassy. D1 Kathode is fine but slightly sparkly, R3 too and leg 1 on Release pot. But the other legs on Release are bassy distorted. The D1 anode and D2 are bad.

Small caps parallel with D1
On recommendation I tried 47pF, 82pF and 120pF, but no real change. Feels more like the category of trimming a circuit than trying to find the root cause for the distortion.

Scope trials
Got a tone generator and a 1-channel scope to probe around, and it was a clear flat top of the signal wave. But not the bottom of it. Might explain why this distortion sounds different than a pedal intended for that purpose.

3mm LEDs
For a while I thought they were causing problems since one of them measured really flimsy values but the other firm.
And they both shone quite bright. Until I swapped the ICs again, then back to "quiet" and the dodgy one stopped being that.

with looper/signal         only power no signal     
feed   VCC    Vref          feed   VCC    Vref
9.10   8.91   4.45          9.10   8.90   4.45

IC1                                IC1     
4.8-5.06  8.91                4.55   8.90   
4.46        4.45                4.46   4.45   
4.07        4.36 rising       4.08   4.43   
0.0          4.35 dropping  0.00   4.43   


IC2                           IC2     
4.50   8.91                4.66   8.89   
4.46   3.9-4.7            4.44   3.88   
4.27   4.43                4.26   4.43   
0.0     4.25                0.0     4.25   


IC3                         IC3           
1.10   0.0                1.35   0.0   
0.0     0.0                0.16   1.69   
4.46   0.0                4.43   1.45 jumpy 0.25   
4.45   0.0                4.44   0.36   
4.47   0.0                4.45   0.31   
0.0     8.91              0.0     8.89   
0.0     0.0                0.0     0.0   
0.0     0.0                0.0     0.0   

     PNP                   PNP
C   1.16                  1.20     
B   3.9-4.7              3.85     
E   4.45                  4.38   

Next steps
Will probably put back the C5 C6 as Film Caps MKTs again. Feel it got worse with electrolytes.
During my testing I have never really gotten much change out of Attack and Release. And sometimes the way the compression is released when the tone fades it come on very obvious and not smooth at all. So this together makes me wonder about if this circuit will ever contribute to the pedal chain,
But during the swap-fest yesterday I had some moment of clarity when all remedies were working toward a greater common good. Then it felt just fine and worth all these efforts. I touched a knob and the scraping came back in some places and others not. I fiddled mostly with Threshold and Release when this happened.
I've just about suspected everything, re-heated/soldered all and some twice and maybe more.
I have a hard time accepting the design should be faulty as such considering it is a popular pedal and others did not have my issue.
The problem solver in me wants to find out what the problem is. Would be a boost of morale if nothing else.
And I have had great help in this forum before.
But I am also nearing the point where I would order another kit and start fresh. I have knobs and drilled box, so that is worth a try.
Or just give up as the first one that failed on me. (OK I re-bought re-built an American 60's Fuzz Pro (?) (a Fuzzrite clone), but that was me destroying the board until it was just useless. 2nd build ever. I like to think I have evolved since...)
And either buy a good compressor from a shop or settle with the Zirconia (Diamond Comp clone) opto-comp that is quite nice.

[matopotato]

Some pictures:

[eh la bas ma]

I built this version :

https://www.musikding.de/Engineers-Thumb-Compressor-kit

I have also some non-musical distortion appearing when the Treshold pot is going CCW, depending on how Ratio and Attack are set.

I have a fancy expensive compressor from Walrus, the Deep Six. There is also some sort of distortion when Sustain , Attack and Blend are fully CW.

I have the Meridian too, but it is kind of special, not like the usual compressor. But extreme settings will do extreme things to the signal...

I think it is normal, maybe depending on the pickups...

Both my guitars have two active hambuckers and I never tried my compressors with a passive single-coil pick up.

Are we sure this distortion issue is actually an issue ?  Many circuits need to be carefully adjusted, or they won't be very useful, compressors would be among them in my experience.

On mine, I just turn the Treshold control until the noises disappear.

[matopotato]

I built this version :

https://www.musikding.de/Engineers-Thumb-Compressor-kit

I have also some non-musical distortion appearing when the Treshold pot is going CCW, depending on how Ratio and Attack are set.

I have a fancy expensive compressor from Walrus, the Deep Six. There is also some sort of distortion when Sustain , Attack and Blend are fully CW.

I have the Meridian too, but it is kind of special, not like the usual compressor. But extreme settings will do extreme things to the signal...

I think it is normal, maybe depending on the pickups...

Both my guitars have two active hambuckers and I never tried my compressors with a passive single-coil pick up.

Are we sure this distortion issue is actually an issue ?  Many circuits need to be carefully adjusted, or they won't be very useful, compressors would be among them in my experience.

On mine, I just turn the Treshold control until the noises disappear.

Thanks,
All good points.
In my case I am convinced. The distortion happens early and on moderate settings. If I dial it "out" then there is basically no compression. So the "useful" area does not match the available area very much.
I compared with Eng Thumb build. Same settings, and Eng is fine, but mine distorts.
Also had replies to soundfiles in the thread on pedalpcb forum that tended to agree. With my settings. Low Attack and Release. Volume to stay at unity. Ratio noon to 3, Threshold 9 to noon. But now more all over.
If this was something I had bought, and the seller said it was as designed, I would have returned it.
But I am also aware of the extremes of this and other compressors can cause "bad" output, just wish there was any "good" in mine.

[eh la bas ma]

I would say the strength of the Engineer's Thumb is to get a more balanced sound when you play full chords, allowing to hear the lows and the highs more evenly. Other than that, besides the volume swell with the Release control, I think it's quite unoticeable.

If I want to really increase the attack when I do some fingerpicking for exemple, I use an other compressor able to do that well.

[ElectricDruid]

Are we sure this distortion issue is actually an issue ?  Many circuits need to be carefully adjusted, or they won't be very useful, compressors would be among them in my experience.

You should be able to use a circuit with the controls in any position without something sounding terrible. Ok, certain settings are better than others, and extreme settings might not be that "musical", but the circuit should continue to function correctly. If  a compressor only works with the knobs within a certain range, either there's something wrong with it or the design isn't much good, in my view. So, yes, I think there's an issue.

[Mark Hammer]

We guitar players tend to forget that guitars do not put out smooth signals.  We rely on compressors to do that for us.  But the unevenness of the guitar signal means that there are unexpected peaks as a note seems to die out.  Of course, people will pay special attention to the decay phase of strummed guitar strings, because that is where we expect compressors to create "sustain", by electronically maintaining a steady level.  Paradoxically, it is during that decay phase where the signal level poses the most challenges.

Fifty years ago, I was studying for an exam in the student residence lounge and lying back on a bench seat.  As I looked out at students entering and leaving the building, from an upside down perspective, I couldn't help but notice that their gait was a LOT more uneven than I was accustomed to.  Normally, seeing people walk "rightside up" every day, one imposes perceptual smoothing and continuity in their movement, as if the actual bobbing up and down with each step (remember, that as you extend one leg/foot well in front of the other, you have actually shortened your height by a few inches, compared to standing straight) simply didn't happen.

The same is true of guitars (and probably many other instruments).  We PERCEIVE the decay as very smooth when it is actually very bumpy and glitchy.  Notice this when it comes to analog octave dividers using a 4013, where the signal falls unpredictably above and below the triggering/tracking threshold, and we notice this with noise gates whose gate-on threshold is set wrong.  When it comes to compressors that use a half-wave rectifier, the momentary fluctuations in compensating gain will show up if the decay of the rectifier is set too fast or too slow.  Indeed, the audible consequences of half-wave rectification in noise-gates, compressors, and auto-wahs is regularly reported by people as "distortion".

The Thumb-Sucker is not a bad design, but its use of an OTA driven by a half-wave rectifier makes the instantaneous response of the OTA susceptible to envelope ripple.  That's one of the good things about using LDRs in compressors and auto-wahs: they "smooth out the glitches" arising from half-wave rectifiers...at least if you let them.

The solution is usually either shortening the gain recovery time, so that one leaves ripple behind quickly, or lengthening gain-recovery time so that the circuit behaves like a slow-to-respond LDR.

[matopotato]

We guitar players tend to forget that guitars do not put out smooth signals.  We rely on compressors to do that for us.  But the unevenness of the guitar signal means that there are unexpected peaks as a note seems to die out.  Of course, people will pay special attention to the decay phase of strummed guitar strings, because that is where we expect compressors to create "sustain", by electronically maintaining a steady level.  Paradoxically, it is during that decay phase where the signal level poses the most challenges.

Fifty years ago, I was studying for an exam in the student residence lounge and lying back on a bench seat.  As I looked out at students entering and leaving the building, from an upside down perspective, I couldn't help but notice that their gait was a LOT more uneven than I was accustomed to.  Normally, seeing people walk "rightside up" every day, one imposes perceptual smoothing and continuity in their movement, as if the actual bobbing up and down with each step (remember, that as you extend one leg/foot well in front of the other, you have actually shortened your height by a few inches, compared to standing straight) simply didn't happen.

The same is true of guitars (and probably many other instruments).  We PERCEIVE the decay as very smooth when it is actually very bumpy and glitchy.  Notice this when it comes to analog octave dividers using a 4013, where the signal falls unpredictably above and below the triggering/tracking threshold, and we notice this with noise gates whose gate-on threshold is set wrong.  When it comes to compressors that use a half-wave rectifier, the momentary fluctuations in compensating gain will show up if the decay of the rectifier is set too fast or too slow.  Indeed, the audible consequences of half-wave rectification in noise-gates, compressors, and auto-wahs is regularly reported by people as "distortion".

The Thumb-Sucker is not a bad design, but its use of an OTA driven by a half-wave rectifier makes the instantaneous response of the OTA susceptible to envelope ripple.  That's one of the good things about using LDRs in compressors and auto-wahs: they "smooth out the glitches" arising from half-wave rectifiers...at least if you let them.

The solution is usually either shortening the gain recovery time, so that one leaves ripple behind quickly, or lengthening gain-recovery time so that the circuit behaves like a slow-to-respond LDR.

Thanks. Most interesting indeed. Will probably have to digest it a bit more to let it sink in.
And yes compressors seem to me more and more as an area where "it comes together" in many aspects. So I am learning and digesting all I can.

Still, it sounds really crappy compared to others I've heard. Considering the many buyers and users, it just has to be something in my unit that is different.
I fell I have just about suspected everything. Possibly the pots themselves remain.
Take them off each one and test their reaction to turn to look for anything out of the expected perhaps.
I was contemplating removing the LEDs at the beginning of the signal path, but there is no distortion there yet, so not sure.

[matopotato]

Are we sure this distortion issue is actually an issue ?  Many circuits need to be carefully adjusted, or they won't be very useful, compressors would be among them in my experience.

You should be able to use a circuit with the controls in any position without something sounding terrible. Ok, certain settings are better than others, and extreme settings might not be that "musical", but the circuit should continue to function correctly. If  a compressor only works with the knobs within a certain range, either there's something wrong with it or the design isn't much good, in my view. So, yes, I think there's an issue.

Thanks and yes, that is my thinking as well. And the design should hold as far as others have witnessed.
Unless I get some good hints on specifics to try out, I will have to decide if I should get another same kit and hope to get it right a second time.
As I just wrote above, I could try the pots, but I feel I have tried most things.
And can't see anything that screams out from the readings.
I could measure the diodes A and K as well, but no idea what I should expect there.

[Mark Hammer]

Try a larger value for C6, like 2u2 or even 4u7, and see if that improves anything.

[matopotato]

Try a larger value for C6, like 2u2 or even 4u7, and see if that improves anything.

Thanks.
OK, will do. However it is a bit along the lines of tweaking or workarounding IMHO. And here I also have 18V, other more low-noise ICs, Films for C5 and C6 and different PNPs to work with as well.
And even if I get it to a good and useful level, I'd be happy of course, but still want to learn what I had done wrong in the first place. It is bordering to being "personal" with the pedal almost...
If I do not find the problem I am tempted to getting another kit, breaboard it (yes, should have done that in the first place) and build it and then run them "half/half" if possible until I find the culprit.

[puretube]

Tried a 100n cap (ceramic or film) or 2 or even 3 across the power-pins of the opamps/OTA yet? (In fact in parallel to C100, but closer to the IC-pins). And/or maybe 1 in parallel to C101?

[idiot savant]

Just as a point of reference, both the PedalPCB, and Musikding schematics in this thread omit the 1k resistor from the OTA output to the inverting input of the gain opamp relative to the OG version:

http://www.valvewizard.co.uk/engineersthumb.html

[matopotato]

Tried a 100n cap (ceramic or film) or 2 or even 3 across the power-pins of the opamps/OTA yet? (In fact in parallel to C100, but closer to the IC-pins). And/or maybe 1 in parallel to C101?

Thanks
No not yet. Plan to try @Mark Hammer's suggestion for C6. Then I can try this.
Still, even if it helps, it would not find the root cause?
Unless of course the design is flawed.

[matopotato]

Just as a point of reference, both the PedalPCB, and Musikding schematics in this thread omit the 1k resistor from the OTA output to the inverting input of the gain opamp relative to the OG version:

http://www.valvewizard.co.uk/engineersthumb.html

The one between Ratio and Treble Boost mod?
Do you think not having it it could be the reason for my distortion problem?

[matopotato]

Current plan is to try the various suggestions here and in another thread.
Perhaps with all the swaps and mods as well.
But I think I will breadboard with spare parts as much as possible and take from this build the rest to see if it flies at all. Might lead to a new pcb though, or perf board co sidering my limited finesse and high "development potential" when it comes to soldering...

[Mark Hammer]

Still, even if it helps, it would not find the root cause?
Unless of course the design is flawed.

From my perspective, sometimes the root cause is the guitar signal itself.  We THINK it is smooth and steady, but it isn't.  It's very erratic.  The challenge, when it comes to compressor design, is to turn the erratic INTO "the smooth" with as few parts and stages as possible.  If one skimps a bit and thinks "There, THAT should be enough to do the job", that's not a design "flaw"

[matopotato]

Still, even if it helps, it would not find the root cause?
Unless of course the design is flawed.

From my perspective, sometimes the root cause is the guitar signal itself.  We THINK it is smooth and steady, but it isn't.  It's very erratic.  The challenge, when it comes to compressor design, is to turn the erratic INTO "the smooth" with as few parts and stages as possible.  If one skimps a bit and thinks "There, THAT should be enough to do the job", that's not a design "flaw"

Thanks!
FWIW, I have the issue with both humbuckers and single coils.
Perhaps my thinking that "if it is a kit with good reviews, then there must be a findable root cause with a fix by changing faulty component" is the flaw, at least for compressors.
I tend to rationalize something I thought was very binary, but as you are making me see, it is not. I just have to get that "paradigm shift" into my mind.

[puretube]

Just as a point of reference, both the PedalPCB, and Musikding schematics in this thread omit the 1k resistor from the OTA output to the inverting input of the gain opamp relative to the OG version:

http://www.valvewizard.co.uk/engineersthumb.html

Good find! But then,  in his version 2 (2016) the Valve Wizard omitted it there, too ...
http://www.valvewizard.co.uk/engineersthumb2.html

[matopotato]

Tried a 100n cap (ceramic or film) or 2 or even 3 across the power-pins of the opamps/OTA yet? (In fact in parallel to C100, but closer to the IC-pins). And/or maybe 1 in parallel to C101?

Ok, so I tried 100n and also 220n across pin4 amd 8. No change or effect either way.
Also on C101 but no effect.
Was the expected result that the dist would go away, or be reduced?
Thanks for suggesting though, appreciate it.