But surely, phase cancellation isn't what the OP is all about?

It's about different recorded tracks being 2ms apart, which isn't going to cause phase cancellation. Doesn't phase cancellation occur when a copy of the same program material is fed back in with a delay? This is, after all why we do separate mono mixes for AM radio in cases where the stereo mix uses a lot of time-based stereo effects such as chorus and flanging.

Just a thought..

ROG.