Well, after a LOT of checking, re-checking and checking again, one tiny bit of feedback started me thinking and led me to the solution. Or rather, showed me what an idiot I'd been. I'm so used to DC voltages that I never thought to ensure the -ve lead was on after the bridge recifier... So bring on the jibes, jokes, slagging and insults
So I used a 10 ohm, 10W resistor pushed through either the 5a or 5b header and fixed the other end to ground (yeah, DC ground before anyone says ). This provided enough 'load' to test the 5v rails and mine show rock solid at 4.98v now. -5v and 12v show correctly without any load resistor.
Thanks to all who helped out and look at it this way, with the diodes, caps and power regulators all done, this should be great for decades to come!