"N_Cook" wrote in
:

supervinx wrote in message
...
Hi to everyone, since it's the first time I write in this NG

I've little experience in switching PSUs ...
I've one of them (coming from an Amiga 2000) which shuts off at the
first load.
With no load at all, it delivered 5.9V out of 5 nominal.
I reduced it (with the trimmer) to about 5 with no load.
Inserting a little dummy load (47 Ohm) now it doesn't shut anymore,
but with more serios loads (e.g. an HD), it shuts again.
I checked in-circuit capacitors for shorts and ESR increase, tested
diodes and transistors.
It seems an aging problem (the trimmers were still sealed, so no
tampering), but who's bad ?

is there any OVERvoltage shutdown circuit? some protection crowbar you've
missed? some SMPS have crowbars on the 5v logic rail to protect the digital
logic on the load.I've seen where a zener broke too soon and triggered the
SCR too early and caused shutdown.


As it happens I will try fathoming a dedicated SMPS today that as it
stands is in tick-tick mode, 1 ultrasonic cycle per second.
Putting a dropped LED on each output shows life on all rails.

scope each rail to see how far it climbs.
Maybe one rail is not climbing as much as it should.

Plan of campaign will be
Run up on variac to find lowest point and will still run of sorts,
tick-tick Put a x10 larger timing capacitor over the one on the 555
that feeds the UC3842 PWM controller. Then make up an exerciser to
feed into the ,isolated, LED of the optoisolator, over-riding the
error feed. This will be a monostable of variable period from 1mS to a
second or so with push switch, hopefully to elicit which area is
playing up by gradually ramping up . All active, big Rs seem ok and
caps ESR-check ok. I suspect false error signal.
Anyone else any ideas/advice on a generic plan of action for these
sorts of situations.?




use a scope with a current probe to see if you're hitting the current
limit.
Check the PWM IC's housekeeping cap.Maybe it's not holding the IC supply
enough to keep the PWM IC running.(a common failure)

--
Jim Yanik
jyanik
at
localnet
dot com