Hi mod_critical,
I've been having a bit of a look at some of this stuff, and
that page you posted is well worth a good look.
From:
http://www.btinternet.com/~netsurf/SudburyRC/coxamp.html
"
A critical part of the amplifier design is to ensure the 100nF
/ 1ohm combination is as close as possible to the output of
the amplifier. These two components effectively surpress any
ambitions the LM383 has of becoming a high frequency
oscillator - which this amplifier chip has a nasty tendency of
doing. If it does oscillate, you will find that the amplifier
will start drawing 500mA from the batteries, and the heatsink
tab will get hot very quickly - confusingly, it may also be
working well as an audio amplifier, so making troubleshooting
difficult. If you have problems with this, try increasing the
value of the decoupling capacitors on the +ve supply to the
amplifier, and make the 100nF capacitor bigger on the output -
ensure you use a low inductance capacitor for this, such as a
Polystyrene one."
This is from that post earlier, and it seems to confirm that
the problem is some sort of feedback causing oscillation,
the site this came from explains pretty well what can happen,
and how to deal with it.
The reason that two capacitors in parallel are used on the
output is because the way the layers are wound on can give
a small inductive effect, so another tiny cap is often put
across it, to make sure the higher frequencies are effectively
shorted to ground. These frequencies are way above the audio
range, and should not have any affect on the audio output.
The 'de-coupling' of the earlier stages will almost certainly
clear your problem, if its working properly. What this does
is to take a feed from the supply thats feeding the output
amp, and feed it to the pre-amp through a resistor, and put
a smoothing capacitor so that ripples and changes on the
supply to the pre-amp are smoothed, and hopefully dont find
their way into the signal path.
If you dont have any de-coupling between your pre-amp and
your output amp, that would explain this problem.
If you do have, then maybe it needs looking at.
Also, the 1 ohm and 0.2 mfd at the output should be very
close to the output terminals on the chip.
That site also shows that he has included small capacitors
across the smoothing caps in the power supply, and at the
de-coupling, a worthwile precaution i think.
I doubt if i would start making the PC board till i was
quite happy with the loose-wired prototype, and i would
most likely stick to the same components when making the
final item, i have had things go wrong just cos parts werent
quite the same, even though they were marked the same.
Please come back and let us know how you're getting on.
Best of luck with it,
John