Switch-Mode PSU Failure Modes, Repairs, & Parts Substitutions

From: Geoff Roberts <geoffrob_at_stmarks.pp.catholic.edu.au>
Date: Mon Jul 31 20:41:32 2000

----- Original Message -----
From: "Douglas Quebbeman" <dhquebbeman_at_theestopinalgroup.com>
To: <classiccmp_at_classiccmp.org>
Sent: Tuesday, August 01, 2000 12:27 AM
Subject: Switch-Mode PSU Failure Modes, Repairs, & Parts Substitutions

> I hope everyone had a great weekend, I sure did (not
> Classic-Computer related so I won't bore anyone with
> the details)...

Spoilsport. ;^)

> Subsequent to my first post about my Prime PSU, Don
> Maslin sent me a howto on repairing switchers, that
> was written by Keith Lofstrom. The howto is fairly
> comprehensive, but does assume that most repairs
> are taking place due to bad PSU design.

Often the case, more so in PC type stuff though, realistically.

> thus I think the PSU simply overheated
> inside the cage.

Heat and dust are a bad combo in a psu. Thermal stress kills caps and
the like and can cause odd effects in the operation of the system.

> One additional data point that I thought I'd
> mentioned, but may not have, is that: The system
> was up and running fine, no evidence of any problems.
> I shut the system down that night, and then the next
> day when I turned it on, I got nothing. I did not
> see any evidence that the PSU briefly came on and
> then died; it simply never came on.

This sounds like a classic startup resistor failure.
If the supply is more or less conventional (I've followed the thread
about the
mains side - and now that the details are sorted it seems that it is),
there should be
a high value resistor (typically 300-800k) from the positive hv rail to
the switching
circuit electronics. It's probably relatively small, physically, as it
carries very little current.
Under normal circumstances, in most other circuits, such a resistor
would be unlikely to fail,
however open circuit high value resistors are a common failure mode in

> From what I've gathered from the comments posted
> about switcher operation, and from Keith's howto,
> it seems like something failed that is part of
> getting the secondary supplies "jump-started" to
> an up-and-running state. So, as long as it was
> powered up, it would continue to operate. But once
> I turned it off, the portion of the PSU (something
> between the primary and secondary supplies? Something
> like that big power resistor, maybe?) that gets the
> secondaries going, couldn't get them going.

Unlikely to be a big one, that's probably a current limiting resistor of
some kind.
I think you mentioned this resistor was 15k? That would be a little low
for a startup
resistor, unless the design is somewhat different to what I'm expecting.

> So, has anyone else been in a similar position? Do
> I have a fair chance of simply replacing the parts
> I've found that failed, reinstalling it, and being
> successful?

Worst case scenario in a SMPS is when the fuses blow and you find bits
switching transistor spattered about the case, burnt up resistors etc.
That's a major repair job.

>There is a 5v adjustment that needs to
> be made before bringing the system up all the way,
> and while I hate to use the system itself as a dummy
> load, I don't have the Prime bus pinout, so I can't
> easily determine what's what.

Assuming it IS a startup resistor failure, there should be no need to
tweak the supply, unless there are indications of over/undervoltage.

> One of the things Keith discusses in the howto is
> about parts substitutions. Most of his comments are
> w/r/t transistors and the like, and he suggests
> using as replacements devices that have voltage and
> current capacities that are increased over the part
> that failed. He goes on to say you need to test each
> such substition with a curve tracer before finalizing
> each such substition.

This is to do with the switching circuit itself. Components in that
especially semiconductors, have performance characteristics that effect
operation of the circuit, and equivalent may not be close enough in some
Again, if it's just a startup resistor, then there should be no issues
requiring that sort
of in depth investigation.

> However, I don't think any semiconductors failed for
> me. He says that in a good design, there are some
> power resistors protecting the switcher transistors,
> and from my examination to date, it appears that Prime
> did in fact put these in the design. I'm going to remove
> the other 3 of the 4 that are in that area to be sure
> whether any did or did not go open-circuit, as there
> is significant thermal discoloration of the underside
> of the PCB in that area.

This is not abnormal in power supplies. The resistors do get quite hot,
in normal operation, and some discolouration of the board is inevitable.
Assuming the resistors are ok, just check the solder joints for heat
deterioration, in fact, it wouldn't hurt to resolder any point that has
been subjected
to high temp when operating, especially if the solder looks a bit iffy.

> How reasonable a course of action is this? What's likely
> to happen should a problem still be present? I'm thinking
> that worst case, the parts that failed will just quickly
> fail again, putting me back at square one.

Given the evidence so far, I think you are unlikely to be in that


Geoff Roberts
Computer Systems Manager
Saint Mark's College
Port Pirie,
South Australia
ICQ: 1970476
Received on Mon Jul 31 2000 - 20:41:32 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:32:58 BST