It was interesting to note on stardot that someone mentioned resistor failure
They may have been referring to resistors in the startup chain. Usually, the control section of a running SMPSU actually runs on an auxiliary output of the SMPSU itself, which raises the question: How can the control circuit ever start if it runs on the output from the SMPSU?
The answer is that at switch-on the control circuit is given a brief 'pulse' of power derived from the high voltage DC on the mains side of the SMPSU - this typically consists of a chain of several high value resistors to drop the voltage down from the +350VDC at the top end to the control circuit power feed at the bottom end which is held at a low voltage (6V or 12V, say) by a zener diode. There may also be a high voltage low value electrolytic capacitor which allows this temporary supply current to flow for only as long as the capacitor takes to charge up, during which time the SMPSU should start running and generating the main supply for the control circuit to keep running on.
Things which can go wrong with this arrangement:-
-One of the high value resistors goes open circuit, even though the voltage drop is intentionally spread evenly across two or more similar high value resistors in series so that each resistor bears an equal share of the burden. Even so, they still sometimes fail open circuit.
-The capacitor dries up and goes high impedance or open circuit.
In either of these cases the symptom is the same, the PSU will run for ever for as long as the mains stays on but the next time the power is cycled off and then on again it will fail to start. A typical tell tale for the open-circuit resistor or open circuit capacitor fault is that the main high voltage capacitor remains charged for a very long time after the power is removed, the reason being that the control circuit never starts and therefore never turns on the output switching transistor so no current is drawn from the charged capacitor. The +350V charge just sits there on the capacitors indefinitely.
It is also possible for the zener to fail short-circuit but if it does it will cause immediate failure of a running SMPSU by depriving the control circuit of power.
Note that these observations are not specific to the BBC Master PSU but are broadly applicable to many SMPUS.