Switching PSU Theory of Operation / Repair Tips & Techniques

From: Douglas Quebbeman <dhquebbeman_at_theestopinalgroup.com>
Date: Wed Jul 26 09:03:22 2000

Good Morning, All!

In his last reply, Tony Duell stated (somewhere in all those quotes):

> > You may not have yet seen the correction I posted, but I was seeing that
> > only when I measured the + terminal of one of the blue electrolytics,
> > the - terminal of the other. Why would I do that? I'd just gotten out of
> Yes, but aren't the 2 capactiors in series, something like :
> ------------+-------------------- +ve
> |
> =====
> -----
> |
> ------------+
> |
> =====
> -----
> |
> ------------+--------------------- -ve

Yes. I've made a schematic of this portion of the circuit.
And while I'm not very good at ASCII art, I've made a stab
at it, as you can see below.
> And you're seeing 600V between the 2 points I've marked '+ve' and '-ve',
> right? That's still twice what I'd have expected -- I'd expect peak mains
> voltage (i.e. 115*sqrt(2)) across each capacitor, or about 320V between
> +ve and -ve in that diagram.
> I am still puzzled as to how it gets 600V without either other large
> capacitors, or a mains-frequency transformer.

If we have a 300vdc source, positive with respect to ground, and we
have a 300vdc source, negative with respect to ground, don't we have
a 600vdc potential between the two sources (as long as they share a
common ground)???
> I will just ask again -- this thing does run off 115V, right?


Now, let's talk failed components. And I suppose I'll take a stab
at ASCII art by adding to what you've already "drawn":

                                                    --------------------- +
to BBU
             +---------{ MCI 307138 }---------------+--------------------
            / \ | + |
           / \ ===== / 12k ohm
          / \ ----- \ 5%
         / \ | - /
    ----* *---------------------------------+----------*
    | \ /| | + \
    | \ / | ===== / 12k ohm
    | \ / | ----- \ 5%
    | \ / | | - |
    | ----------{ MCI 307138 }---------------+---------------------
    | | |
    | | |
    | | ---------------------- -
to BBU
    | |
    | |
    | |
115v Load 115v Neutral

In the diagram above, 'BBU' is the label on a connector to which the
labeled lines run. I think it's for a Battery Backup Unit. Cool, who
needs a UPS when you've got a bank of kick-ass batteries? But I digress...

The components that I've shown as {MCI 307138} are what appear to be
either large disk ceramic caps, or, judging by appearance, more likely
MOVs. The one on the + side is discolored and cracked open along its
circumferential periphery. It was blue, some of it turned green, most
of it turned black. The one on the - side is discolored similarly but
it not cracked open. These components are on a small board which is
separate from the main PCB and which mounts by being screwed onto the
tops of the two electrolytic caps.

The electrolytics are Mallory CGS 2900MFD 250VDC. The one on the - side
was in close proximity to a power resistor on the main PCB. The power
resistor on the main PCB was a Dale 15K ohm 5% 20W unit. One leg of the
power resistor is soldered to a trace which leads to what I think it the
oscillator, which is one of two separate boards that plug into the main
PCB. The power resistor has changed in value from 15k ohms to infinite ohms.

The main PCB was discolored underneath the power resistor. The electrolytic
shown above on the negative side which was in close proximity to this power
resistor may have suffered some damage, as the blue plastic (heat shrink?)
that Mallory uses for the CGS caps is discolored where it was in closest
proximity to the power resistor.

There are some other components, quite far away on the PCB from the area
I've been discussing, that have also gotten hot enough to badly discolor
the PCB. These components include some small 5% power resistors, two other
items that look like resistors but may not be (as the color banding looks
odd), and a couple of polystyrene caps.

I unsoldered one of the little 5% power resistors (four of these, with
little ceramic spacers/insulators on the leads, 51 ohms). It tested OK.
These power resistors are nearby to two transistors (in what I think are
TO-3 cases, the ones that kinda look like hats when viewed on-edge).
Also, in that area of the PCB, the leads to one of the transformers
have heated up and discolored the PCB underneath.

A couple of other notes:

The yellow AC LED appears to be lit from a signal that comes from
a circuit on the main PCB that monitors the line voltage. Unfortunately,
said circuit requires the DC supplies that aren't currently working.

> > > > complex looking boards (although they're only 2-layer) with custom
PALs and
> > > > the like.
> > >
> > > PALs? Are you sure? It's not at all common to have significant amount
> > > digital electronics in an SMPSU. Most of the time it's either simple
> > > analogue circuitry either as discrete components, or one of the
> > > SMPUS control chips.
> >
> > These look like digital boards to me. I'll pull 'em and either scan them
> Well, you've seen it, so I guess you're right. Seems a little unusual,
> > or take a digital photo, as well as record a few part numbers.
> If you could post the numbers on any ICs on the boards, I can see if they
> make any sense. Or have they used house-numbers only ?:-(

The daughter board containing the PAL has a 16L2A, IIRC. A TI part.

I was told that the PSU provides a signal to the Virtual Control Panel board
called "Confidence", which is like the PowerGood signal that PC switchers
out. This daughterboard may monitor the voltages and then provide that

That concludes my investigation to date. I'm starting to feel good
about the possibility of repairing this PSU, even without the test
equipment I should normally have. By replacing the obviously failed
components along with the ones you and others will hopefully suggest
should be replaced due to the failure of the obviously bad ones, I
think this thing may damned well work again.

> > > > If it matters, the room in which I keep the Prime was hotter than
> > > > I usually let it get when I have the Prime running (I don't have
> > > > central A/C, so I have a window unit in an adjacent bedroom and
> > >
> > > In which case look for dried-up (high ESR) electrolytic capacitors.
> > > is a very common heat-related problem.
> >
> > Would I not have to open them to make that determination? I though they
> > contained PCBs or somesuch toxic chemical.
> Firstly, no, don't take the capacitors apart -- you'll ruin them if you
> do.

I assumed that... and I don't want to be out the US$45 a replacement would
cost, but...

> Measure the ESR electically -- either with a special ESR meter (the
> Dick Smith Electronics one is claimed to be very good and not expensive),
> or on an impedance meter. I've never used the Dick Smith meter -- I was
> going to buy one and then I was given a rather nice digital impedance
> meter (it applies a sine wave to the device-under-test, measures the
> inphase and quadrature componentes of the current through it, and
> calculates the resistance and capacitance/inductance). But I've heard
> nothing but good things about said ESR meter.

...I figure the test equipment will cost more than the replacement parts.

Of course, the test equipment has value beyond this one problem. :-)

> Secondly, the electrolyte in these capacitors is not PCB-based (PCBs are
> not electrolytes anyway...). The electrolyte is not particularly harmful.
> Actually, according to somebody who works with _large_ transformers and
> capacitors, yes, PCBs are dangerous _in quantity_, but unless you're
> really silly, the amount in one capacitor, or even all the capacitors in
> an HV PSU, is unlikely to do you any harm. I can't really comment on this
> -- it's not my field.

What, you mean this isn't the Classic Chemistry list? 8D

Hey, I feel like we're making progress. Tony, thanks again, and
to everyone, I kept this on the list rather than take it private
because it just seems like someone else might benefit from it,
and of course, I'm seeking input from all who'd care to contribute.

regards to all,
-doug quebbeman
Received on Wed Jul 26 2000 - 09:03:22 BST

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