Chuck Guzis wrote:
Did her observation reflect reality? Are there
any other
manufacturers that routinely engaged in this practice?
Most guitar amplifiers. If you look at the power amp, you'll see that
most have two big NPN power trannies driven by a cascaded NPN and PNP
smaller power tranny. If one goes, all four go in a bizarre silicon
suicide pact.
I don;t think I've recently described the typical failure mode of the
Boschert 2-satge SMPSU (on-topic, since it's used in classic computers,
including the PERQ 1).
I'll start by describing the basic circuit :
Mains in is rectified and smoothed, giving 350V DC
This feeds a (non-isolating) step-down switching regulator (using one
chopper transsitor and an inductor). The output of this is around 150V
(but see below)
This feed as pair of chopper transistors which are coupled to a small
transformer making a free-running oscillator, the freqeuncy of which is
determined by the transfomer properties
The output of that feeds the main chooper transformer, the secondaries of
which are rectified and smoothed in the usual way.
A voltage sense signal is taken from one of the outputs (almost always
the +5V line), compared to a reference, and fed back (opto-isolator)
across the isolation barrier. This controls the first step-down
regulator, thus controlling the voltage to the second choppers, and thus
the output.
Oh yes, there's a current sense resistor in the first regulator circuit
which will shut that chopper down (and thus remove the outputs) if
there's an overload. Adn a crowbar on at least one of the outputs.
Now for the failure mode :
The chopper transistor in the first regulator circuit goes short-circuit.
The otuput of this stage (should be 150V remember) leaps to 350V. The
second choppers carry on working, so all the outputs try to go up to over
twice their normal value. You'd better hope the crowbar fires...
Let's assume it does. That short-circuits one of the outputs, thus
causing a much higher current in the chopper circuit. This is detected by
the overcurrent sense circuit, which removes the drive from the first
regulator chopper. Alas that's what has shorted, so removing the drive
doesn't do a darn thing!.
The 2 second-stage choppers now go short-circuit. This means that across
the 350V rectiifed mains you have essentially the 3 transistors (all dead
short), some inductor and transformer windings (almost no DC resistance
either) and the current sense resistor. The last burns out. This now
applies 350V to the current sense inputs of the controller IC (a good old
723) which blows apart.
Oh yes, a couple of small signal transistors blow up too, and a PCB track
is blwon off the board.
Finally the fuse fails
And yes, I have seen the results of this happening, and indeed rebuilt
the PSU.
-tony