Overdrive produces a whislting sound when nothing is played

[Echan42]

Redording: https://vocaroo.com/181Vb4Fi1kqT

I built this TS808 based pedal, mostly the same as the one on this veroboard.


Only when both the drive and level knobs are over 90% I get the following issue.
When the effect is disabled there's a very VERY faint whistling sound that is amplified to horrendous levels when the effect is active.
This happens using battery or AC adapters.
The sound is only produced when no sound is being played and happens even when the guitar is unplugged, as soon as a single note is played on the guitar the effect works as intended.
The values for those knobs are as suggested by the original build, I only changed the treble knob to a 10k.

I've taken upon myself to read the liked Debugging Page by R.G. Keen's GEO and I feel most suggestions don't apply to the kind of problem I'm having.

[antonis]

Place a 1 - 10 nF cap between INPUT tip & sleeve..

[Echan42]

Place a 1 - 10 nF cap between INPUT tip & sleeve..

I've just figured out I had the cap on the output buffer section on the circuit wrongly biased, I think it's safe to say that's the culprit.

[Echan42]

I thought the issue was with my electrolytics but I was mistaken, I've stared at this board for so long I was seeing things.

The sound really reminds me of resonance in active filters.
Mounted a 1nf polyester between the tip and the sleeve, as per antonis' recommendation, both with the instrument connected and disconnected.
In both cases the tone of the whistling was changed, and volume either lowered or raised, but never dissapeared.

I've revised the positions of all my electrolytics and examined the circuit up and down to no avail.
Here's the schema I drew https://king.seedhost.eu/echan42/Schematic.PNG

And here's a video I recorded that illustrates the issue way better than I've described it.
https://www.youtube.com/watch?v=7CN41ppXcAg&feature=youtu.be

[aron]

I feel like this is a classic case of parasitic oscillation, usually fixed by increasing distance between input and output and shielding and by routing wires so that the input and output are distanced. I don't recall the TS-808 having that much gain. I wonder if somehow the op amp or some other stage has incorrect values and therefore creating much more amplification than it's supposed to have?

[Echan42]

I feel like this is a classic case of parasitic oscillation, usually fixed by increasing distance between input and output and shielding and by routing wires so that the input and output are distanced. I don't recall the TS-808 having that much gain. I wonder if somehow the op amp or some other stage has incorrect values and therefore creating much more amplification than it's supposed to have?

Parasitic oscilation? Got some bedtime reading to do!
Both your statements sound reasonable, the wires do run paralel a short stretch and they do enter the PCB close, I designed the PCB with a generous ground plane in hopes to avoid anything along these lines.

I do find this TS to have more gain that I remembered but the gain was high in all stages of development, from protoboard, to veroboard to final PCB, so I considered it a quirk of this schematic, so I guess I'll probe up and down for anything suspicious, when you mention incorrect values what do I have to keep an eye out for?

Edit: I mean yeah, I can assume the way to go is probe components and jot down voltages but is there any way I can cross reference these values? Maybe replicating the schematic in LTspice and simulating the circuit?

[niektb]

Start by double-checking your resistor values, It's a bit hard for me personally to read Tagboard Layouts (without schematic) but look at the feedback resistors in the op-amp loop, those primarily define gain

[Echan42]

Start by double-checking your resistor values, It's a bit hard for me personally to read Tagboard Layouts (without schematic) but look at the feedback resistors in the op-amp loop, those primarily define gain

haha I'm with you on that one! tagboard are so complicated to follow, it's like snakes and ladders. I uploaded my own schematic based on the veroboard build, but it's a little spaghettish. I've double checked all resistors and checked them while assembly just to make sure none where bad or grossly close to tolerance.

Place a 1 - 10 nF cap between INPUT tip & sleeve..

On second examination I've done this test again and it does stop all the whistling, I'm sure this wasn't happening before but I've disassembled this pedal twice now so I can't vouch for consistency between tests.
It also disappears when I plug in my guitar with the volume knob at 100% any lower and it comes back, tone knob makes it come and go.
Does this indicate a problem in the input stage?

edit: I've re-routed the PCB becuase under close inspection there where places where input and output ran pretty close, I also re-wired the enclosure to keep the wires as far away as poissble and wound them with ground to no avail, the interference is identical to before I made these adjustmends.
This post has provided me with invaluable info https://www.diystompboxes.com/smfforum/index.php?topic=101538.0

[Echan42]

Ok! It's been a couple of days and I've leaned a lot as I go, here are my findings.

The oscilations are between 20kHz and 8kHz depending of what the knobs are set at.
20kHz oscillation means it's definitely not oscillating by slew-rate alone, so changing from the 4558 (4V/μs) to a tl072 (20V/μs) made no significant difference.
I've read up on noise supressing and unintended oscillation and I've taken the following measures that were unfruitful.
Separating input and output cables and winding these with ground made absolutely no difference.
Examining the grounding to the chasis I can confirm the jacks, chasis, power source and PCB share a common ground, I've read that multiple groundings produce unintended oscillation, does this occur if multiple leads are grounded to the chasis or if the PCB has multiple ground planes?

The only conclusion I can draw is the original Veroboard guitar FX schematic has too much gain due to incorrect resistor values.

Solutions:

1- I can substitute C2 (originally 56pf) which acts as a LFP in paralel to the diodes to cut the high frequencies out and cut-out the high freq oscilations, but this largely affects sound.

2- Modifying the original schematic and raising the value of specific resistors. Specifically raising R2's original value of 1k to 300k (big jump I know) affected the range of my level knob, so instead of going from 0 to 10 (0 being silent and 10 being loudest) it now goes from 2 to 8, which is a very acceptable mount of boost anyway.

Both these solutions solve the problem outright, and in addition a 1nf cap across the input and GND do make the sound slightly cleaner, (is there any reason I shouldn't remove it?).

All that's left to do is compare the veroboard schematic to original TS808s and see if I can tame this beast without sacrificing that much gain or limiting my level range like with the second solution.

You guys are being very helpful and insightful I'm grateful of this comunity, if I've made any mistakes in my assumptions or taken a wrong turn troubleshooting this any correction is encouraged.

edit:
I've also realized the veroboard layout I followed doesn't bias the negative input to ground at the diodes after passing thru a CR wich is standard operation in a tubescreamer's clipping stage, biasing to ground makes no difference as far as I can tell

[Echan42]

Here goes a third solution.

Putting a very low impedance between the second op-amp's inputs makes the whistling completely dissapear. Impedence any higher than 6ohm gave crackles and pops with high output pickups at the highest volume.
This is a very good solution and determines that the oscilation is ocurring in the tone shaping stage.
The severe and pretty bovious downside to this is that treble (that feeds right into the op-amps positive input) now acts as a volume control +  treble.

[antonis]

The severe and pretty bovious downside to this is that treble (that feeds right into the op-amps positive input) now acts as a volume control +  treble.

Pretty anticipated due to variable voltage dividing effect..

[Echan42]

The severe and pretty bovious downside to this is that treble (that feeds right into the op-amps positive input) now acts as a volume control +  treble.

Pretty anticipated due to variable voltage dividing effect..

Indeed! I misspelled obvious 
I feel like I'm slowly closing the circle and I have high hopes of hitting the nail on the head sooner or later and sharing the solution here!

[antonis]

I feel like I'm slowly closing the circle and I have high hopes of hitting the nail on the head sooner or later and sharing the solution here!

There is absolutely nothing bad into this..