Dead PSU

From: jpero_at_pop.cgocable.net <(jpero_at_pop.cgocable.net)>
Date: Mon Jul 6 13:13:19 1998

> Date: Mon, 6 Jul 1998 18:48:28 +0100 (BST)
> Reply-to: classiccmp_at_u.washington.edu
> From: ard_at_p850ug1.demon.co.uk (Tony Duell)
> To: "Discussion re-collecting of classic computers" <classiccmp_at_u.washington.edu>
> Subject: Re: Dead PSU

> > Oh, bit about the fan jump a bit with power removed. I'm 100%
> > certain something is shorted in the secondary outputs and by the way
>
> The symptoms as I understand them are that the fun runs if the PSU has no
> (or very light load), but the PSU trips if the load is increased.

But Max later said when he pulled the power he noticed about a second
later the fan jumped. I have seen that before too.

> > most of time usually has none of scr or clamp downs to block
> > overvoltage, the PWM ic usually '494 and LM339 monitors the output
> > voltages.
>
> For normal operation, there's a feedback loop applied to the PWM chopper
> control IC, normally via an optoisolator.
>
> I've just looked at a dozen or so SMPSU schematics. All of them have some
> independant shutdown other than the main feedback loop. PC power supplies
> tend to have a 339 applied to the shutdown pin of the 494. Just about all
> others, including Apple ones, have a crowbar thyristor. One obscure one
> has a thyristor which operates a separate optoisolator to apply a
> shutdown signal to the primary side.
>
> Whatever method is used, a PSU that trips is detecting some error and
> shutting down. Either overvoltage or overcurrent. A defect in the current
> sense resistor is unlikely, so it's likely to be an overvoltage error.

That also would not work when either secondary caps and diodes
components is shorted. That is what it's like when PSU tries to spin
into a low resistance load but PWM noticed that and kept off but when
power is removed the PWM disconnects from safety shutdown mode as
input power by that mains filter decays causes PWM to run at lower
power but enough to make fan jump.

>
> >
> > Remove the main transformer and check all diodes for dead shorts and
> > leaky capacitors ONLY in the output side, about 6 or 8 diodes and few
>
> If it'll run at all, then a shorted diode is not likely.

That is not what I meant, PSU are supposed to be loaded always for
proper operation and shorted components adds up to lower resistance
that trips the PWM controller safties. But with all the loads
removed, that PSU is running on that shorted diode as a resistance
load at just right resistance.

>
> >
> > Don't be fooled, at least 2 diode pairs are in two "transistors"
> > casings, but there's markings that looks like two diodes back to
>
> Often, but not always. Also watch for diodes in 2-lead TO220 cans (look
> like a power transistor with 2 leads).

That would be very unusual to me in PSU but someday I would run into
these but I only dealt with common stuff, I only saw one or two
machines that used this 2 lead in TO-220 case.

>
> In genral, transistors are designated by Q1, Q2, etc and diodes by D1,
> D2, etc. That's one way to spot them.

Not USUALLY. :) Cheapo and in good stuff usually does not mark very
well. Use your eyes on circuit layouts and components # used to
pull down docs off the net if needed.

> > But this does not beat the info if you really trace out where all the
> > outputs out on that PSU and pick out most likely components to check
> > with your meter and catch the baddie red-handed. Pure joy when that
> > happens! :)
>
> One way to find high ESR output capacitors is to put the supply on a
> load where it works, 'scope the outputs, and increase the load on each
> output in turn, gradually. You'll see spikes appear on the output with
> high ESR capacitors.

Remember, most guys have no OSCILLSCOPE. :-( Mayor $$$ and very
slow to aquire one.

>
> I hesitate to mention the next trick, because it can damage some
> capacitors, but here goes. It's handy for field repairs if you've not got
> a 'scope with you.
>
> Remove the capacitor from the circuit. Charge the capacitor - must be
> more than 1000uF - from a 9V radio battery, observing polarity. Disconnect
> the battery, and wait a few seconds. Now short out the capacitor with a
> screwdriver. If you get an arc (audible click, spark), then the capacitor
> is probably OK.

Or you're done "zapping" one and putting back that "good" but now
blown caps, adding to the confusion. Also I don't relish doing that
and can't hear the tiny pop or zapping sound.

> > the cost is so cheap. Get one from "Bob Parker's" DSE's ESR meter
> > kit from under and to the left without a second thought.
>
> Yep, an ESR meter is on my list of things to get... I'll probably build
> the one in 'Television' - it's simple (a few op-amps), and cheap to
> build. The problem with the Dick Smith one is that I believe it uses some
> programmed (and copy-protected) device, which means I'll not be able to
> keep it running for ever.

Oh, I tried to obtain the "Television" from a special inporter
mag reseller but they can't find or get them. Well, develop one and
make code freely available with circuit layout then otherwise could
anyone obtain that cicuit layout and send some to interested party
like me.

> > Boo hiss. :)
>
> Yep, nothing annoys me more than companies that won't sell spares. Sharp
> are top of the list at the moment - I'll not buy _anything_ from them
> again.

That one thing but I meant usual parts like caps, specfic semi
components and such that is easily available in US and some places in
canada but none in my hometown, has to rely on substitution heavily.

> of a problem than custom ASICs and mechanical bits.
True...byond that most semis not those jellybean ones are what I need
and most places have those jellybeans but not those I need.

For ASIC's, I resorted to collecting large # number of similar boards
that uses same chipsets and take good ones to repair high quality
boards.

>
> -tony

Jason D.
Jason D.
jpero_at_pop.cgocable.net
Received on Mon Jul 06 1998 - 13:13:19 BST