Stupid simple comparator fuzz

[anotherjim]

Finger out then, and breadboarded.
Result silence. TL072 not happy with output slammed into +ve rail.

Ok - to work with input near -ve supply, how about a CMOS amp? Could it be a TLC272?
I have those and.... bingo! Comparator Fuzz definitely fuzzing.

Mine has 100k to ground on + input and a 10k series resistor from guitar and no clamping diodes/led's -  otherwise same as PBE6 has drawn. Oh, I padded down the output a bit with a 10k before a 5k vol pot.
100k input (actually 110k) because I'm starting to think this is more suitable for this type of circuit than the usual 1M. The signal loss form 10k/100k divider on the input doesn't seem to harm operation with passive guitars -  even my Telecaster is happy.
Unused amp has pins 6&7 tied and 5 to ground.

No, I don't have a stripboard layout 

[PBE6]

Interesting! You'd figure that if one TL072 worked then they all would. I suppose it's possible the function depends on being in a Goldilocks zone for some spec that changes significantly enough from chip to chip to require "getting lucky". If only life were always so much fun

I tried rewiring my prototype to bias it at half supply, sounded mostly the same although there was some intermittent noise present between notes. Going to try adding some hysteresis and see what happens.

[PRR]

> You'd figure that if one TL072 worked then they all would.

They will (usually), IF you follow the rules.

Have a sufficiently-low resistor to a stable bias point. (TL07x can use any resistor you can buy, even 100Meg.)

That bias supply MUST be within the "common-mode input range".

Recommended Operating Conditions
VCM Common-mode voltage VCC– + 4 VCC+ – 4 V

i.e. with 9V supply, you should stay between 4V and 5V.

In fact it "works" (perhaps not to full spec) with a wider range. At +/-15V it will "typically" (weasel word!) function to +15V and -12V. You still have to stay 3V up from your most negative supply rail (zero in typical 9V pedal work).

If you grok the schematic you see why. The inputs two P-JFETs. Their drains must be a few volts negative of the gate (input) pins. Those drains stand on about 1.2V worth of BJT junctions.

So when you set the inputs too close to the negative rails, these JFETs turn off. Well, never perfectly off. Which one of the pair is more-off or less-off? That will determine what the rest of the circuit does in this "illegal" extreme.

It is interesting that it worked once. If my analysis is right (it incomplete!) then it may work on "most" chips if the mask is a little heavy on one FET. OTOH maybe the mask is symmetrical and it is a 50:50 bet. DIYers may pick and sort. A pedal production run that came out half no-good would be a commercial disaster.

[anotherjim]

I was wondering if the working TL072 is an imposter - maybe the marking is smudged. That's why I tried the TLC272 as with a bit of imagination it might be misread. The original scheme with the volume pot lowering supply volts maybe worked better than it should for a TL07x, but the TLC range work down to 3v total supply.