[Keith]

Thanks for all the helpful suggestions.
Hugo: All the components you mention have been changed with the exception of the 470pF sm tuning capacitor - I'll do that next. The fact that the recovery is pretty well instantaneous does point to an intermittent component and of course a dry joint is always a possibility.
Beery and D Cassidy: The 820K line oscillator anode resistor is a new component - yes it had gone high! The ECL80 tests good on the AVO VCM but I ought to try another one in there.

I'm not now convinced the EY51 was a problem - certainly changing it didn't initially make any difference. Can these be realistically tested on a VCM?

To me the associated line linearity shift must be a strong clue. Presumably a problem with the line waveform could also result in poor EHT regulation, either due to lack of line pulse energy and/or low heater volts on the rectifier.

The "problem" now is that the fault seems to still be absent, despite the set having been switched off for 24 hours. Maybe I should be pleased, but we all know that intermittent faults have a habit of returning at the worst possible time!