--------------------------------------------------------------------------

My number one common failure fault finding is the units raw B+ filter cap(s). They inherently have to be mechanically small to fit within the cramped casing restraints. [Seems like someone mentioned one venting on them...hopefully with a....BANG!!!... like mine did.] Therefore, that tends to have them at woefully smaller values than could be optimal. They really take a hammering from the ~40 khz operating frequency repetitions of squarewaves. If you have facilities/capabilities of doing an ESR test on some used units you should find them onsetting to high ESR levels in a hurry. Not having that, just a running of a unit (open cased for access) for a warm up period of 10 minutes and then totally disconnecting from AC power and subjecting to a finger tip temp/warmth test should reveal units that are in ESR excess. (Will be quite warm or hot/versus just cool.) Also one could make connection across a cap at a time with a clip leaded AC voltmeter for a ripple test, bet you will find it quite high with their selected (necessitated?) values of caps.
Some units that I did enact a repair upon , ended up in a location that physical width provided no restraint, so I utilized healthier cap values as well as 105deg temp ratings (for sure). The new caps then extended thru 2 holes of the plastic housing with epoxied plastic insulative "warts" encasing their slight protrusions. 4 years and counting now.

73's de Edd