PDP 11 gear finally moved

tony duell ard at p850ug1.demon.co.uk
Thu Jul 23 13:14:49 CDT 2015

> * I once talked to a power supply guy about over-voltage sensors and
> the like, and he said that over-voltage conditions (like a power
> supply freaking out and giving TTL 8 Volts or something) due to
> component failures are exceedingly rare. When it does happen, it is
> almost always due to human error - mis-installing sense lines,
> cranking a trimmer too far, setting line voltage improperly, and so
> forth.

This is presumably why all decent PSUs, even those without remote sensing,
without twiddlepots to set the output voltage, etc, have crowbars. Manufacturers
do not fit components for no good reason.

I've had several instances where a PSU has 'gone crazy' in use. One time it was
a dry joint on the sense resistor. Another was in a simple 3-terminal
regualtor circuit where the common wire fell off the regulator [1]. The last
which has happened several times, applies to common switch-mode designs,
if the output smoothing capacitor goes high ESR (or in once case there was a
bad VIA connecting it to the rest of the PSU -- and on an HP board at that) you
get spike on the output of several times the desired output voltage.

[1] 5A regulators are often in TO3 metal cans and the +ve one has the common connection
to the case. If this is made by a solder tag under a fixing screw and said screw works
loose then the output voltage can go sky-high. A neat trick if possible (i.e. when there
is a separate input transsformer secondary/rectifier/smoothing cap for each regulator) is
to put a tag under each fixing screw, input -ve to one, output -ve (system ground) to
the other. Then if either screw comes loose you get no output as the circuit is opened.



More information about the cctalk mailing list