Thumb Sucker Compressor distorts instead

[matopotato]

I re-breadboarded the circuit on a new breadboard. Looked up the neatness and got luckier. Tried to adjust accordingly.
Same distortion. It is more on the Breadboard version than the PCB still. If I touch with finger on some of the pots, there is a change to the noise, but the distortion mainly remains the same. It is less on treble strings, and more on bass. And quite full on chord strumming. Which is consistent with my findings since day 1.
Tried two wall warts and battery. No differences I could detect. Stayed with the 9.0 stamped wart in the end.

Readings BBD V2:



Breadboard, txt:

Power in 9.17 (Got a b
VCC 8.92

IC1
4.57   8.91
4.48   4.46
4.03   4.47
0.00   4.29

IC2
4.59   8.91
4.41   4.77
4.37   4.46
0.00   4.37

PNP
C 1.10
B 4.76
E 4.46

IC3
1.10   0.00
0.18   0.66-0.06*
4.47   0.50-0.01*
4.46   0.67-0.06*
4.48   0.70-0.06*
0.00   8.91
0.00   0.00
0.00   0.00

* These were falling values. And that side is not even used, so probably non-interesting. But I added them just in case.

Will try the trimmers as @puretube suggested later on.
Will try the RGB @duck_arse points with signal from guitar.
I have a generator and an old 1-ch scope in case there is something useful I should try with them?

[matopotato]

... Finally compare them with your friend's compressor (give him your blueshift, he will agree i am sure). See if anything stands out.

Yes, might come to that. Actually he had a TC E 3rd Dimension (BOSS DC-2 sort of) that is his least liked pedal. So I have it to compare with Blueshift. Sadly Blueshift hasn't sung yet and I owe you the IC swap change. All prepped, but I figured better to concentrate on one brick at a time.
I got as far as noticing the upper deck first IC didn't have proper power so I suspect the power stage, bad solder. I am more hopeful once I can get readings for all ICs that it will actually perform. But that is another thread.

[eh la bas ma]

What did it make you find or discover or adjust?

That the sidechain was working on my build.

I didn't change anything. I tried to modify the input resistor, as suggested by a skilled forumite, from 1k to 10k. It didn't change anything obvious, so i soldered the 1k back on my board. I remembered Merlin said :

The circuit works with the component values in the schematic, changing them isn't going to fix anything, it's more likely to create new problems.

Edit: I notice your last readings are very similar to mine, as you can see in my thread, reply #8. I'd say they're allright, they "make sense" ?

[duck_arse]

neatness - tidy, ordered, sensible, all squared, no wobbly long legged resistors hanging about. and those ground links! short, straight, bare copper wire formed like a staple [stationery] is best, makes sense, non-confusable, not those long looping black wires you show.

and that diode with the red and white paper ----- remove the paper bandoliers, they have and leave problem glue gunk. ANY/ALL parts with that gunk on their leads ends, cut the ends off. never ever shove one into the breadboard, it will only end in straitjacket.

as for your latest voltages, they are looking right, sensible, as EHLB sez. however, ignore what I said about the C5 and C6 points voltages moving with signal - I found they stayed steady, as near as, but the transistor collector voltage did move in sympathy with signal level. you need to be careful measuring this point on your IC, as you can fritz the chip if you slip with a probe.

and as for you mate's box pot positions comparison - if his pots are a different taper to yours, or even a different manufacturor, you are comparing your oranges to his apples, your knob settings won't match.

[matopotato]

neatness - tidy, ordered, sensible, all squared, no wobbly long legged resistors hanging about. and those ground links! short, straight, bare copper wire formed like a staple [stationery] is best, makes sense, non-confusable, not those long looping black wires you show.

and that diode with the red and white paper ----- remove the paper bandoliers, they have and leave problem glue gunk. ANY/ALL parts with that gunk on their leads ends, cut the ends off. never ever shove one into the breadboard, it will only end in straitjacket.

as for your latest voltages, they are looking right, sensible, as EHLB sez. however, ignore what I said about the C5 and C6 points voltages moving with signal - I found they stayed steady, as near as, but the transistor collector voltage did move in sympathy with signal level. you need to be careful measuring this point on your IC, as you can fritz the chip if you slip with a probe.

and as for you mate's box pot positions comparison - if his pots are a different taper to yours, or even a different manufacturor, you are comparing your oranges to his apples, your knob settings won't match.

Thanks a lot.
Still have some learning to digest. I tend to keep legs and paper until they go to be soldered on a pcb or perf/vero/strip board. But point taken.

Will use wire and crocodile clip instead of probe then.

Pot from same place and will check values.
I also hade my Thumb Sucker kit compared with other builders if the same regarding pot settings. And my problem was confirmed. My friends ET is as close to hands on as I can get.

EDIT: Yes, 2 pots differ at least

[matopotato]

Another theoretical thought try from my side, before starting to breadboard this circuit:
what happens to the "fuzz", when the inverting OTA-input (pin4) is not directly connected to Vref (4.5V) as in the original schemo, but through a 220R resistor?
(before trying a 500R trimpot pot across the +/- inputs with wiper to Vref, as another possibility to check whether the whole case is an offset-/symmetry-issue; or even AC-coupling the OTA).

I was not sure how to connect the 500R trimmer. I tried 220R from pin4 before going to Vref, but no change.
I think you mean to put the 500R between pin3 and 4.. As a bridge? and leave all other connections as they are?

[matopotato]

For a moment the breadboard worked the same as the PCB. That is not fuzzy, but with distortion if I take Ratio a bit over noon and Threshold below noon. At my "would like it to sound clean at least here" with R at 3 and T at 9 there is still distortion.
So the Breadboard works just as "well" as the PCB now. ("Ett fall framåt"-A fall forward)
Then I tried some tips to make the OTA offset adjustable, hopefully to some more central point as suggested by @puretube. And I think I did not quite made the connections properly, so will have to re-think/re-ask about more detail. So I put things back, and the distortion had gotten worse again.
So there is something loose and I am touching on the breadboard mocking things up for sure.
Still the breadboard has the capability to act as the PCB does, so I should be able to experiment on it and try things out. If I can get back to the somewhat less distortion mode, then I can evaluate for sure. Else I can probably evaluate maybe some stuff, but it might be a false approach in case some error would hide the solution to another.
Would like to at least try the wiper idea in case the signal is offset. But should try to get back to a more stable circuit first.

[puretube]

Another theoretical thought try from my side, before starting to breadboard this circuit:
what happens to the "fuzz", when the inverting OTA-input (pin4) is not directly connected to Vref (4.5V) as in the original schemo, but through a 220R resistor?
(before trying a 500R trimpot pot across the +/- inputs with wiper to Vref, as another possibility to check whether the whole case is an offset-/symmetry-issue; or even AC-coupling the OTA).

I was not sure how to connect the 500R trimmer. I tried 220R from pin4 before going to Vref, but no change.
I think you mean to put the 500R between pin3 and 4.. As a bridge? and leave all other connections as they are?

One end of the pot to pin 3, the opposite end to pin 4, and the wiper to Vref (4.5V).
Lift the end of the 220ohm resistor that goes from pin 3 to Vref (in the orig. schematic) on the Vref-side and leave it floating, while testing the trimpot-settings.

[matopotato]

Another theoretical thought try from my side, before starting to breadboard this circuit:
what happens to the "fuzz", when the inverting OTA-input (pin4) is not directly connected to Vref (4.5V) as in the original schemo, but through a 220R resistor?
(before trying a 500R trimpot pot across the +/- inputs with wiper to Vref, as another possibility to check whether the whole case is an offset-/symmetry-issue; or even AC-coupling the OTA).

I was not sure how to connect the 500R trimmer. I tried 220R from pin4 before going to Vref, but no change.
I think you mean to put the 500R between pin3 and 4.. As a bridge? and leave all other connections as they are?

One end of the pot to pin 3, the opposite end to pin 4, and the wiper to Vref (4.5V).
Lift the end of the 220ohm resistor that goes from pin 3 to Vref (in the orig. schematic) on the Vref-side and leave it floating, while testing the trimpot-settings.

Thanks. Took a little bit for me to get it right. And it "works" in the sense that it affects the distortion.
CCW less distortion, less volume and mor of the oversaturated?, dying battery fuzz sound, gating?. To CW it gets back to more of its usual self with distortion but more volume and a bit clearer tone. I would like to check with the scope if I can recreate the flattened tops but round troughs.
And then drive the trimmer to see if it does even out the curve. And if so, then try a 500R across pins on my pcb version and some eqviv R from the trim between pin3 and Vref.
So a bit more promising than other tests. Will travel next week so might be a while.

[matopotato]

neatness - tidy, ordered, sensible, all squared, no wobbly long legged resistors hanging about. and those ground links! short, straight, bare copper wire formed like a staple [stationery] is best, makes sense, non-confusable, not those long looping black wires you show.

and that diode with the red and white paper ----- remove the paper bandoliers, they have and leave problem glue gunk. ANY/ALL parts with that gunk on their leads ends, cut the ends off. never ever shove one into the breadboard, it will only end in straitjacket.

as for your latest voltages, they are looking right, sensible, as EHLB sez. however, ignore what I said about the C5 and C6 points voltages moving with signal - I found they stayed steady, as near as, but the transistor collector voltage did move in sympathy with signal level. you need to be careful measuring this point on your IC, as you can fritz the chip if you slip with a probe.

OK

and as for you mate's box pot positions comparison - if his pots are a different taper to yours, or even a different manufacturor, you are comparing your oranges to his apples, your knob settings won't match.

Pot            EngThumb    ThumbSucker
Level         A10k            A10k
Attack       A100k           B100k
Ratio         A1M              A1M
Threshold  B1M              B1M
Release     B500k           B1M

So Release differs, but when testing I keep it quite low. An neither of us have felt much effect from Release nor Attack. In my case likely due to the finer points being overshadowed by the distortion problem. In his case, you can hear something, but neither of us is much of compression knowledgeable so might take some getting used to.
Attack is also set low in the tests so I doubt A vs B has much impact either.
The other three same, so even though I would have understood some differences, what we hear when tweaking knobs still makes my problem (and him not having one) quite obvious.
Might try to film knob tweaking while playing to make it more clear. But that will have to wait a little.

[matopotato]

Lift the end of the 220ohm resistor that goes from pin 3 to Vref (in the orig. schematic) on the Vref-side and leave it floating, while testing the trimpot-settings.

I plan to try this on the pcb version. I think I  can carefully bend pin 3 and 4 out from the socket and try your idea. Only thing is how to lift the 200R between its end and vref. I don't want to cut. Any thoughts? Some parallel R would work in the opposite way...

[puretube]

Lift the end of the 220ohm resistor that goes from pin 3 to Vref (in the orig. schematic) on the Vref-side and leave it floating, while testing the trimpot-settings.

I plan to try this on the pcb version. I think I  can carefully bend pin 3 and 4 out from the socket and try your idea. Only thing is how to lift the 200R between its end and vref. I don't want to cut. Any thoughts? Some parallel R would work in the opposite way...

then take out the 220R completely? (same effect)

[matopotato]

Lift the end of the 220ohm resistor that goes from pin 3 to Vref (in the orig. schematic) on the Vref-side and leave it floating, while testing the trimpot-settings.

I plan to try this on the pcb version. I think I  can carefully bend pin 3 and 4 out from the socket and try your idea. Only thing is how to lift the 200R between its end and vref. I don't want to cut. Any thoughts? Some parallel R would work in the opposite way...

then take out the 220R completely? (same effect)

No, I mean on the soldered pcb.
I have a copy on breadboard which I  tried but the ground problem is worse there. So I want to try your idea on the soldered PCB bynlifting 3 and 4 pinnout fromnits socket. Then crocodiles and loose wire should be possible to join with the trimmer.
Which leaves the 220R. Soldered. I guess I  have to desolder on ts vref side

[puretube]

Lift the end of the 220ohm resistor that goes from pin 3 to Vref (in the orig. schematic) on the Vref-side and leave it floating, while testing the trimpot-settings.

I plan to try this on the pcb version. I think I  can carefully bend pin 3 and 4 out from the socket and try your idea. Only thing is how to lift the 200R between its end and vref. I don't want to cut. Any thoughts? Some parallel R would work in the opposite way...

then take out the 220R completely? (same effect)

No, I mean on the soldered pcb.
I have a copy on breadboard which I  tried but the ground problem is worse there. So I want to try your idea on the soldered PCB bynlifting 3 and 4 pinnout fromnits socket. Then crocodiles and loose wire should be possible to join with the trimmer.
Which leaves the 220R. Soldered. I guess I  have to desolder on ts vref side

If pin3 is out of the socket, the 220R is not connected to it anymore. Okay. But that 10k resistor must connect to pin3!
Bending IC-pins (and later trying to bend them back ...) is a dangerous thing! They love to break off.

[matopotato]

Thanks, point taken. Still, I am reaching the end of ideas. I have one from another thread that will be revealed in July. And yours.
After that, unless fixed, I am considering donating it to science. With just the root cause explained in return.
We'll see.
For the bent legs I'll probably desolder the 220R one leg.

[matopotato]

Today I tried the suggestion from @puretube with the trimmer on pins 3 and 4 of the LM13700 as described in earlier posts.
It looks messy, but as far as I could tell it was connected properly. I lifted the 220R. Bent legs 3 and 4 out and inserted wire as needed.
There was an effect, Low R/CCW made the tone/sound thinker/bassier and slightly louder. And the opposite the other way.
But the distortion is still there and not too subtle either. A bit more difficult to turn the knobs, but I managed to vary within the same intervals as previously and the distortion does not wear off.



If nothing else then at least a smile somewhere perhaps.

What remains is to wait for July and a tip much awaited from another older thread.
I might be getting some replacement transistors as well, but considering previous attempts I doubt they would make any difference.

[puretube]

frustrating ...

[matopotato]

frustrating ...

Thanks, yes quite so. Been  trying to sort it out off and on since February. By no means any record of any sort, but draining none the less

[eh la bas ma]

Did you try what I suggested in reply #78 ?

These tests will tell us if some important parts of the circuit are working correctly.

If they do, as i suspect, there would be a good chance that your circuit is ok. Meaning that you are in fact trying to improve the circuit by removing all distorsion (or even just some of it), which seems much more challenging than troubleshooting, all the more so if you aren't an electronic wizard.

Back when I was wondering about some strange connections while my circuit was powered, Chuck D. Bones, a great member of the PedalPCB community, told me that this circuit is supposed to distort the signal, quite easily. He built the same as yours, I am sure he knows what he's talking about.

https://forum.pedalpcb.com/threads/engineers-thumb-compressor-question.12149/

Your circuit doesn't behave exactly like your friend's, or any other similar circuit, because of potentiometers tolerance and values, if I understand correctly.

Nothing obvious can fix this, or Merlin would have already done it. One simple solution would be replacing pots by resistors, as in the 2-knobs version i guess. This way, every settings would (probably) work without distorting.

Mark said something about using LDRs in reply #6, this could be a start, if you want to redesign the circuit entirely, improving the whole thing with clever ideas, which could take years.

[matopotato]

Thanks!

reflow all suspicious pads (and those that look good), hunt for shorts, perform a ritual sacrifice to the diy goddess (while listening to The Bloods),

No, you're right. I still have the ritual sacrifice left. And find out about the Bloods.
I did measure as @duck_arse recommended you. And posted above and my values seemed to check out. Also a full-ish measure of all interesting junctions.

Yes, Chucked helped me a lot on my thread at PedalPCB. Linked at the beginning of this thread. I tried gis suggestions. Then there might have been some slightly more long-shot ones left, but he did not come back with details about those.
I think he breaded in parallel and there is distortion in the more extreme settings for most builders. And as I hope I wrote earlier, this is no issue. But the tyoe of distortion I get at 3&9 (Rat/Thrs) others are not having.
I think me and Wook22 have it, not 100% if you do too, but felt it was less disturbing and opted for a tighter interval setting anyway. But by now, mine is pretty bad in the main part of any useful interval.
I am hoping for sone last ideas in July in another revived thread on my subscription list.
If that does not help I might be willing to "donate it to science"