powering up older machines - is it safe?

Tony Duell ard at p850ug1.demon.co.uk
Tue Jun 3 21:51:13 CDT 2008

> However the  swithers used are less likely to overvolt than fry the
> high voltage side (chopper fails!).   In the case of the Robin that

That very much depends on the switcher!.

In _any_ switching supply there can be a failure in the regulation loop. 
It's not common, and it's not what I'd really worry about, but that can 
make the voltage jump, of course.

And in most swtichers, an open-circuit or high ESR capacitor on the 
output side can put high voltage spicks on the outputs. These spikes 
typically get bigger as you draw more current from the supply. I've had a 
couple of cases where the supply was fine on minimal load, but tripped 
the crowbar when you loarded it, precisely because the output capacitor 
was drying up. And both were in classic computer devices (a PDP11/44 CPU 
and an HP1350 graphics translator). I don't know if these spikes have 
enough energy to damage logic chips, but I wouldn't risk it!

Howeever, there's an even worse case. There is a switching supply design 
that was quite common in classic computer hardware of the 1970's and 
1980s. I've come across it in an HP printer/plotter, the PERQ 1, a 
Sanders printer, and so on. It goes like this : 

Incomiong mains is rectified/voltage doubled to give about 350V DC.

This is fed to a non-isolating step-down switching regulator, giving 
about 150V on the output 

This feeds a free-running oscillator (typically 2 or 4 power transisotrs) 
which drivs the main chopper transformer.

The output of that is rectified and smoothed as you'd expect, There's a 
feedback loop from the output to the _first_ stage (the step-down 
regulator), this controls the voltage supply to the oscillator, 
thuscontrolling the output voltage.

Now for the failure. If the chopper transisotr in the first stage shorts, 
the output of that stage jumps to the 350V input. In most cases the 
oscillatorr keeps on running, and the low voltage outputs of the supply 
jump to over twice what they should be.

Hopefully yhe crowbar fires. If not, then your logic is toast. If the 
crowbar does fire, well, it's not pleasant either, in that the 
overcurrent trip circuit normally operates on the 1st stage. Needless to 
say, if the chopper is shorted, it doesn't do a lot, with the result that 
when the crowbar fires, the oscialltor transistors die too, often 
shorted, which puts a dead short across the 350V supply. This gnereally 
results in current sense resistors,. small signal transistors, and even 
PCB tracks, exploding.

And yes, I've seen it happen and had to sort it out. Fortunately the 
crowbar _did_ fire in my case, I had a nasty PSU rebuild, but the 
1000-or-so chips i nthe PERQ 1A were OK.


More information about the cctalk mailing list