It's my understanding there is no max effective separation. the further apart the two collectors are, the higher the resolution, at any wavelength. Of course, there are practical limits, but that is engineering, not physics. In a radio interferometer, the time lag of the wave front arrival at the two receivers is what is actually measured, so when the wave front hits the first receiver, a signal is dispatched to the second receiver, so that it "knows" when the first receiver got the signal and can subtract the two arrival times.

If the two receivers are too far apart for a signal to be passed between the two, then they use synchronized atomic clocks to measure the time lags. This calculation was once done with analog oscillators and circuits, I imagine now its done in software.

There is probably a minimum effective separation, though. When the distance between the two receivers starts approaching the wavelength,
the effect becomes less useful.

What is amazing about VLBI in astronomy is that they are getting arc-second resolution with microwaves, which is about the limit you can get with optical telescopes because of the distortions in the atmosphere. So a radio "image" can show you as much detail as a photograph, and you can combine false color microwave images with UV, IR or visual imagery. This can really give you an insight into physical processes in distant objects.