still having capacitor problems in the IRIS PSU

I'm starting to get a bit peeved - I've tried randomly throwing capacitors across the output, scientifically choosing a capacitor to have a reactance of 1 ohm at the trouble frequency, and taking that cap and putting another .022?f unit in parallel - and the noise doesn't get attenuated at all, it's still at around 500mV, in high audio frequencies. If it was going down (to ~300 mV) I'd just build a LC choke filter and be done with it, but its stubborn spurning of something that should work is bothering me. Is the ripple carrying that much current that the resistance of the capacitor is an issue? I don't think so- this is a secondary supply (-5V!). I also don't get why there should be a relatively equal AC component in 3 secondary supplies that do share much after the transformer primary, and yet a low level in the main +5V. Could this be something bizarre in the connection of the scope? I'm doing a very bodged scope -> BNC cable (one of the bundle that usually runs to the monitor) -> 2 alligator clip leads stuck on to the relevant bits of the other-end BNC connector. Been trying to work out this board. It's one daughtercard in the PSU, the chopper transisors are on another. Yet, there are 3 TO-3 cased bipolar (2Nxxx parts) scattered amongst the rectifier diodes. Similar units are not present in the +5V channel. Would this have linear regulators on the non-main channels?