Some power supplies will self destruct with too low an input voltage.
The
duty cycle pegs and the transformer saturates.

First part I agree. Second part here is the issue:

It doesn't make the transformer saturate, to do so will present a too
high output. The amount of FIELD in it determines the output and that
does not increase, except when a bad fault exists in the regulation
loop or in some cases tuning. If the snubber network opens up, or part
of the Xformer core pulverizes or cracks it might well cause this
unless there is a working shutdown circuit. If destruction is not
immediate, it is because the chopper is running at too high a frequency
to bring it into regulation, at least with most SMPSes.

The problem with low voltage is twofold. One it usually causes the
chopper to run at a lower frequency. This may cause the collector or
drain current to go beyond it's peak rating. (Ic peak). It is still
charging the magnetic field to the same level if it is in regulation.

Two, the drive optimization circuit is outside of design limits, which
means in the case of a bipolar chopper it may run out of base current
and overheat. The same is true of drive voltage for a FET chopper. Even
though the input impedance of the FET is high, the driver usually slams
against some diodes and this can cause exactly the same thing, although
this particular failure mode would be more apt to happen with a bipolar
chopper.

Actually I'm surprised that the OC protection didn't keep it from
booting, it must be very lightly loaded. When the OC protect is on the
hot side it is usually a resistor on the source or emitter of the
chopper, and even at the same output THAT current would increase,
actually double, almost exactly double. Of course then there are only
about =BD the number of spikes, so some OC protection circuits might not
catch it.

I do agree with you, but I don't think that such a power supply can
actually get past a certain duty cycle.

[Hot damn, I just thought of something. Since we're talking theory a
bit here. I think I might know a good reason that Sony doesn't have a
DC fuse after their SMPS.

To take Devil's advocate, if they put a fuse on it, when the HOT shorts
the time waiting for the fuse to blow would result in a tremendous
spike, possibly destroying all the other electronics in the set. They
could filter it after the fuse but if that filter went bad it would be
an insidious fault. Remember what happens when the 503Khz crytal stops
while the countdown has the HOT turned on. Those are fun to run down
because you get no indication of what happened.

This of course offers them no reprieve from the fact that they didn't
put any current limiting in it. This is technology they had in the
early 80's. These things are higher tech ? That's like touting a
supercharger on a car and putting crappier brakes on it.

I already own all the Sony product I ever intend to, unless I get
gifts.]

Anyway, in a nutshell, the computer PS runs out of drive current or
voltage for the chopper I think, while I know that alot of the driver
chips can do a 100% duty cycle, I think you'll find the feedback amps
designed to prevent this, but they probably let the duty cycle go
higher than it really should. This is to prevent a brownout from
causing a restart, but that is supposed to be temporary. The OP had a
permanent brownout situation. IMO, even with good fans it probably
overheated due to Dv/Dt. Just like an HOT with insufficient drive.

JURB