Strange things afoot with a gain control

[edvard]

I have a breadboard that I use to experiment with dirt circuits.  The default front end is an op-amp clean buffer/booster (non-inverting, set for a gain of 5 so it doesn't clip itself), to a volume/gain control (100k audio taper pot wired as a voltage divider), then to a distortion generator (gain, clipping, tone shaping, etc.).

Here's the problem:  With the knob set at 10, I get a lot of noise (big surprise...  ), but when I back it down to 9.5-9, the noise level goes down substantially, but the output barely changes at all.  In fact, with the knob set just behind that "knee", I can get a nice Metal chug with much of the fizz rolled off (depending on the dirt circuit I'm messing with).  I can take the pot and rotate it back and forth over that spot and it's a substantial change within like 3mm of rotation.  I tried a different 100k audio potentiometer thinking that this one was defective, but the new one exhibited the same behavior.  I tried different values, like 50k and 10k, but they all did pretty much the same thing. 

"But Edvard," you may say, "why are you doing it this way?  Can't you just wire the pot as an adjustable feedback resistor?"  Yes, I tried that, but for some reason, it doesn't work.  I set up a second inverting op-amp (so I could turn it down, too) with a 100k pot between output and inverting input, but it went all fizzy and sputtery on me.  Attaching a plain resistor in place of the potentiometer across the same spot sounded fine, and all tests of the pot with my multi-meter said it was good. 

So, what is happening here?  Is this sort of thing common when using a pot like this?
In the grand scheme of things, this is really a minor issue, but darn if it isn't the raspberry seed in my dentures right now... 

[PRR]

> substantial change within like 3mm of rotation

Your *circuit* oscillates at "10" but not at "9.5" (or whatever your numbers are).

Better layout.

[edvard]

Well, whaddya know... I know that breadboards are famous for not allowing optimum lead dress, but I took a second look at where signals were coming and going and found an input cap on a high-gain input parallel with a resistor from a downstream output.  I fixed that, and no more 'sweet spot'.  Now the noise comes up smoothly with the gain all along the full sweep.  I gotta admit, I kinda miss it now... 

[iainpunk]

you can still put a small value capacitor from one side of the resistor to that input cap, to make it hiss again.

[edvard]

you can still put a small value capacitor from one side of the resistor to that input cap, to make it hiss again.

LOL, I was talking about the "sweet spot" on the pot, where the hiss went away and the gain backed down just the right amount to let some clarity through the chaos.  I can still do it, but it's less obvious and a few more mm's down the sweep.