[sdiy] TB-303 smps interference

[cheater00 .]

Hi guys,
I'm back on the project of removing hum from my 303's output. (It's
likely present in every TB-303, but I only have this one to test). The
symptom is: while one of the leds blinks, when it's on there's a buzz
coming out of the output. Especially noticeable if you're using
distortion on the output.

I've improvised a small E/H field probe. It's made from one strand of
UTP cable (with stranded core, not wire), and one of those rings that
stay on the neck of a water bottle when you remove the cap. About 40
turns all around (I think the cable was ~50 cm). There's a strong,
very localized source where the smps is located, and nowhere else. You
can see the photos here:

http://imgur.com/a/idtGJ#0

I am thinking that some of this might be coupling back into the audio
output. Perhaps the LEDs turn on, creating more strain for the PSU,
meaning some audio circuit sees a higher impedance voltage rail, and
therefore its rejection ability suffers. To test this hypothesis out
I'll put a load on the 303's voltage rail and see if making the load
higher will create more hum in the output.

Using 3 layers of alu foil, even if not grounded, removes all the
pick-up. One sheet works, too.

The method is good but caution needs to be taken. For one thing I
don't know if the plastic I've used for the core is good. Later I've
noticed that holding the probe would make the display fuzzier, and so
would even holding the probe with the "helping hand" clip. In fact any
metal near the probe is bad. Perhaps taping a pen with the refill
removed could work as a handle. I've had to suspend the probe to get
the last few photos, taken on a different scope, with red graticule.

The signal is some sort of square with huge ringing. Its frequency
modulates randomly, what can be seen in the last few pictures: the
signal becomes fuzzier and fuzzier in the time domain as distance from
the point where the time base triggered increases. The trigger always
happens on a clean edge, so all triggers are synced, but since the
frequency changes the length of each trace getting drawn modulates
which ends up accumulating to a general fuzziness.

I don't know why the alu foil works when it's not grounded. I
understand it's not ferrous (but who knows what they put in it
nowadays). So it should only catch E field, and only when it's
grounded. Does anyone have any insights?

Cheers,
Damian