audio + pcb that work after touching

[coppercone2]

So I had this amplifier that I am gonna remake on a PCB and it had this problem where the sine wave was being distorted and some crossover distortion.

Not really sure what was up, but if I fiddle my fingers around it (I cleaned it good with flux cleaner) and touch a instrumentation amp, it often goes into a low distortion stable output state. If I change frequency or voltage levels by alot it will go back to doing bad things. But I tested it, it ran at ~200mA for at least 40 minutes before the waveform went back to having flat areas and such.

and its not an oscillator, its a power amp

In general has anyone else noticed that some analog circuits can be fixed by touching them? I don't mean holding your finger on it, I just mean gently feeling around and having it work after you remove your hand.

I am pretty sure I am pushing strip board to the limit.. but I only expect this behavior with state devices, not analog circuits.

[floobydust]

If it's not due to it oscillating- I've seen when "something" high impedance is floating and the finger is enough leakage current to bring it back.
Or it can involve an electrolytic capacitor that is unhappy about the working conditions, it might be seeing reverse-voltage that gets fixed for a while with the poke changing DC offset.
Otherwise, it could be just bad connections.

[coppercone2]

hmm its soldered and on a low z load

I was thinking leakage maybe too because it could cause it to clip. on some feedback loop

possibly bad parts too I guess this was salvaged from modified equipment I managed to destroy

[coppercone2]

hmm ok well rigamarol 1052 mislead me, I think what is happening is its clipping weirdly (not like a flat) because I measured quite a high voltage on the coupling capacitor output. I guess I must have been bleeding charge from that cap into a power rail

on dc coupling mode it goes ape shit when you change the amplitude of the wave form.

the dc level is completely different then what I thought it would be. like order of magnitude. i wonder what will happen if I reduce the feedback network impedance by similar. i think maybe the output amp unhappy with drain

[coppercone2]

12 gun salute for the shady ebay apex chip that died due to a errant probe (maybe), but this may be a blessing that bring in new infrastructure like heatsinks, sockets and current parts mounted to a etched board.

maybe I will buy a set of foil or mica compensation caps too so I am not sticking Z5U in there lol