As an aside, it is certainly possible to have a dip on the +5v momentary enough to reset or halt a microprocessor that you would never see on a multimeter. You would need a digital storage oscilloscope with at least 20MHz sampling.
Anyway, hope you've fixed it now.
Thanks. The diagnosis has been crazy on this tbh, took ages to pin down to a specific area but looks like at least two things could have been contributing to it, the CPU socket and diff connectors (than the norm) to the power driver board. Still got fingers crossed here but at least it seems I've cornered it at last.