The unit is a Seventeam ST-250BLV, which is also curious in that it's the only model I've seen with two ferrite coils on +12V (despite its age). It uses a TOP222Y (which runs at 100kHz and is allegedly capable of up to 25W out with 200VAC, or 100VAC doubled) for the standby supply, while the main supply is a then bog-standard half-bridge. The primary capacitors are 470µF Panasonic UP, the secondary ones are CapXon KF but seemingly intact. Maybe, as PCBONEZ suggested, they weren't as bad back then but have become worse and worse in recent times...
I shorted the +5VSB to check its response...which causes the main supply to turn on (and regulate normally - with no load attached, anyway)!!! Remove the short and operation returns to normal. (Never short a two-transistor oscillator, though, or you'll blow it up.) I can't say how any other unit (with an IC-based standby supply) will react - the boring response would be for the unit to just "wait" for the short to be cleared. But seemingly, the restart attempts alone transfer enough power to run the main PWM controller in this one (though driving the switchers with a higher duty cycle would probably drain that small amount of power).
(I'm not sure why DM311 runs at only 67kHz and VIPer22A at 60kHz. The smaller transformer on the standby supply ought to be usable at a higher frequency than the large main supply transformer, given similar materials (which is probably the case for half-bridge and forward PSUs). And even general-purpose electrolytics are rated to handle more ripple current at 100kHz than they can at lower frequencies. All TinySwitch-II/III/4 models run at 132kHz and later TOPSwitch models are switchable (see the documentation) between 132kHz (for smallest transformer at lower output) and 66kHz (for reduced skin and proximity effects and EMI at higher output). HiperTFS runs the standby supply at 132kHz and the main supply at 66kHz.
The entire choice of operating frequency is really just a balancing act between the size of reactive components vs. skin and proximity effects and EMI. I suppose most household appliances would work better if designed for and run at 1kHz, but at the expense of the massive power distribution wiring (this is the entire limitation that switching supplies are meant to sidestep in the first place). There is "litz" wire that has multiple thin strands, insulated from each other and twisted/braided in a defined pattern, to minimize the losses - but power companies wouldn't be happy with having to use it throughout the grid…)
