pot scratching with high gain amp

[sarv]

I'm testing out a simple clean boost circuit that's just an op-amp in a non-inverting amplifier configuration (see schematic). I'm getting an issue with pot scratching at the highest gain portion of the knob, and I can't figure out how to get rid of it.



From other forum posts (i.e. https://music-electronics-forum.com/...hlight=scratch), it looks like DC across the pot is a typical source of this kind of issue.

When I check the DC level on each end of the pot, I get 4.01V on the C4 side and 4.61V on the other, when the pot is at its min position, and 4.58V and 4.61V when it's at its max position. So there definitely is some DC current flowing through that pot. Interestingly enough, the DC differential actually gets smaller in the region of the pot where I'm getting the scratching (towards it's max position), but I guess with the gain being so high there it probably takes even less DC to pose a problem (assuming that is the source of this issue).

I'd be interested in your opinions on whether you think the DC on the pot is indeed the issue, and if so, how to fix it?

I've tried subbing in a 1u film for C4, thinking it was related to the ESR, but it didn't seem to make any difference.

I have a PCB made up for this circuit and one thing I've noticed is that the scratching is less severe when everything's soldered to the PCB. Especially I noticed that when I took C4 off and put it on a breadboard via a couple test leads (to experiment with changing the component), the scratching was much worse (almost unbearable). So maybe something could be improved with the layout? But not sure what – it's a simple circuit so everything fits pretty nicely. I've got 24mil signal traces and a bottom layer GND plane that's relatively uninterrupted (can post screenshots if people want).

As a final sidenote - I don't think it's related to the problem but I'd just really like to understand it anyway – I realized that the DC levels at the op-amp inputs aren't equal... at the non-inverting input I'm getting about 2.35V, and at the inverting I'm getting 4.61V. I'm wondering if that means I'm operating the op amp in some kind of region that it doesn't like very much?

Any help would be much appreciated!

[Helmholtz]

Two short remarks:

1)With an intact foil cap or a low leakage E-cap in place of C4 there cannot be be DC across the tone pot. Did you measure any DC directly between the pot terminals? Do you get pot scratching even if you short the input to ground?

2)With a differential input voltage of more than 2V any opamp will be in total DC saturation and would pass no signal.


So either your meter is defective or the circuit is oscillating at a high frequency which messes up the meter readings.

[sarv]

Did you measure any DC directly between the pot terminals? Do you get pot scratching even if you short the input to ground?

I'm using an oscilloscope to measure. I measured both ends of the pot using GND as my reference (using the same GND point for both scope probes). So it would seem that there is indeed DC across the pot. I understand that it doesn't make sense theoretically.

With a differential input voltage of more than 2V any opamp will be in total DC saturation and would pass no signal.


So either your meter is defective or the circuit is oscillating at a high frequency which messes up the meter readings.

Again, I understand that this is strange behavior but I'm using an oscilloscope and would see any high frequency oscillation. Also, the expected level is 4.61V (that's the VGND level), so I would think oscillation would raise the level, not lower it (to 2.35V), right?

[sarv]

I realized I forgot to mention another potentially important/useful bit of info: the lower the pot value, the less severe the problem. I originally wanted to use a bigger pot to be able to get to nearly unity gain, but the scratching was fairly significantly worse with a 500k pot, for example.

[Helmholtz]

I'm using an oscilloscope to measure. I measured both ends of the pot using GND as my reference (using the same GND point for both scope probes). So it would seem that there is indeed DC across the pot. I understand that it doesn't make sense theoretically.

Do you get the same DC readings on both scope channels if you connect both probes to the same circuit point?

The opamp amplifies any differential input voltage (AC or DC) by its open loop gain of maybe 100,000. It simply can't work if there is more than say 0.1 mV between its inputs.

Do you get pot scratching even if you short the input to ground?

[Dave H]

When I check the DC level on each end of the pot, I get 4.01V on the C4 side and 4.61V on the other, when the pot is at its min position, and 4.58V and 4.61V when it's at its max position. So there definitely is some DC current flowing through that pot.

If you are measuring the voltage wrt ground using a 1M input impedance 'scope it will be loading down the voltage at the C4 side through the 150k pot. With the pot set to 150k 4.61V at the R5 end gives 4.61*1000/(1000+150) = 4.01V at the C4 end.

As a final sidenote - I don't think it's related to the problem but I'd just really like to understand it anyway – I realized that the DC levels at the op-amp inputs aren't equal... at the non-inverting input I'm getting about 2.35V, and at the inverting I'm getting 4.61V.

This looks like it could also be the result of 'scope loading. The 1M scope input is forming a potential divider with R1 dividing the voltage at the non-inverting input by two.

[sarv]

Yes I'm using a scope with 1M inputs... the thought had never occurred to me! That totally makes sense. Thanks!

Well in that case, I guess the DC across the pot must be zero or at least very minimal. I'm not sure how I can measure it then with my scope. In any case, still open to suggestions for where else to investigate to get rid of this scratching..

[sarv]

Do you get pot scratching even if you short the input to ground?

Yes, I do. I will say that there appears to be some variability between builds. I've actually built about 10 of these, and they all have some scratching, but I'd say half of them are acceptable and the other half are pretty annoying. It seems like there might be some correlation where the ones that aren't as bad don't have any scratching with the input grounded, or with a non-mircophonic input for that matter (I tested with an iPod and a Teenage Engineering pocket operator), but with the guitar plugged in (and non-zero volume), they scratch.

[Helmholtz]

I'm not sure how I can measure it then with my scope.

10:1 scope probes and DMMs typically have 10M input impedance.

[Helmholtz]

Yes, I do. I will say that there appears to be some variability between builds. I've actually built about 10 of these, and they all have some scratching, but I'd say half of them are acceptable and the other half are pretty annoying. It seems like there might be some correlation where the ones that aren't as bad don't have any scratching with the input grounded, or with a non-mircophonic input for that matter (I tested with an iPod and a Teenage Engineering pocket operator), but with the guitar plugged in (and non-zero volume), they scratch.

Any DC across the pot must be produced by the (varying) leakage currents of the E-caps. There can't be DC with foil caps.

[sarv]

10:1 scope probes and DMMs typically have 10M input impedance

Good call - my probes have a 10x setting so I was able to test this. I'm getting 4.53V when the pot is at its min, and 4.61V at its max, which still matches the math of a voltage divider with 10M impedance (4.54V = 4.61V x 10000/10150). So it would seem there's no DC across the pot.

There can't be DC with foil caps.

I also tested with a 1u film cap (soldered on the board), and there's little to no difference in the scratching. I took a couple recordings and I wanna say the one with the film cap is ever so slightly lower volume, but really it's not definitive in any way.

[Helmholtz]

I also tested with a 1u film cap (soldered on the board), and there's little to no difference in the scratching.

Means that the reason for the scratching is not DCV across the pot but probably caused by a rough carbon trace in combination with the high gain. Maybe try a better quality pot, e.g. Bourns.

[Helmholtz]

Good call - my probes have a 10x setting so I was able to test this. I'm getting 4.53V when the pot is at its min, and 4.61V at its max, which still matches the math of a voltage divider with 10M impedance (4.54V = 4.61V x 10000/10150). So it would seem there's no DC across the pot.

Don't you have a decent DMM? That would allow floating measurements without a ground reference.

What about the DCV difference between the opamp inputs?

[g1]

Means that the reason for the scratching is not DCV across the pot but probably caused by a rough carbon trace in combination with the high gain. Maybe try a better quality pot, e.g. Bourns.

Since he mentioned it is only right at the end of the travel, I was wondering if it might be the 'bump' where the wiper contact hits the lug at the end of the carbon track.

[Dave H]

What about the DCV difference between the opamp inputs?

I don't think it's real. Connect a 1M scope to the inverting input and it will show half the supply voltage (4.5V) because the opamp will drive its output to ensure there is no voltage between its inputs and the non inverting input is at half supply voltage. Connect the scope to the non inverting input and it loads it down to 2.25V (because of R1) which will cause the inverting input and output to be at 2.25V also.