Ok, out of warranty, so nothing to lose by having a look. Turns out the SMPS module inside had cooked it's inrush limiting resistor. Having replaced the module with one with appropriate characteristics, it's working fine again now.
What surprises me is the way this feature has been implemented on the original PSU- It appears to have been an afterthought, since one leg of the PCB mounting fuse has been lifted from the board and the resistor placed in series, with the resulting assembly then encased in heat-shrink. For low-volumes/prototypes, I guess this is understandable, but for production volumes of pro-grade equipment? These things aren't exactly cheap, either....
Basically, I'm disappointed in a shoddy and unreliable hack of a feature which ought to have been properly implemented in the first place.
