VAX 11/750 power supply fault

[Tony Duell]

>  One of the trickiest things about switchers is that they often
> use the main output to provide the rails for the regulation feedback.
>  This causes a chicken/egg problem. We can see that there is
> no output so no feedback to tell the high side to deliver more

If there is no feedback -- that is the feedback detects no output 
voltage, then the power supply controller should try to increase the 
output voltage. That's why a fault in the feedback circuit -- even 
unconencted sense terminals in some poorly-designed supplies -- will get 
the outputs to go sky-high. Hopefully the  overvoltage protection circuit 
(crowbar or whatever) trips before any real damage is done.

You might be confusing this issue with another one. That is, the power 
supply control circuitry needs to be powered. Once the supply is 
running, it; can be powered from one of the outputs of said supply, but 
at swich-on, the power supply outputs are all 0, to the power control 
circuitry is not powered, so the supply can't do anything, so the outputs 
remain at 0.

Some larger supplies -- even things like the VT100 -- have a little linear 
supply to get things going. Other times there's a dropping resistor 
(startup resistor) to power some bits of the control circuit directly 
from the recifiied mians, just to get it all going. Once it's started up, 
it carries on running from the main supply outputs.

DEC took this to extremes in the 11/44. The main PSU circuits (and there 
are 2 of them) in that machine start from a little SMPSU circuit. That in 
turn has a startup resistor.

-tony