Hmm...my first line of thinking was that the rat's nest was acting like an antenae for the induced voltage from the ballast...now you've got me thinking. So maybe the noise was introduced into the power supply and propagated down the power. Actually maybe both. Too bad noise doesn't give off a color or something so that we can see it. We have to guess and test and snoop rely on the reactions to our shielding...at least in this case.

Someday after getting an LM34 temperature sensor hooked up to an ADC0809 I'll try to shield with a grounded metal chicken wire, replace the 'screaming' bulb and listen. If I 'hear' nothing, I'll assume the mains and throw in the chokes...actually now that I think of it it would probably easier to implement the chokes. I'd have to buy the chicken wire.

Carl