tracing out schematics

From: Jules Richardson <julesrichardsonuk_at_yahoo.co.uk>
Date: Sun Nov 21 12:59:17 2004

On Sun, 2004-11-21 at 01:42 +0000, Tony Duell wrote:
> > > > it is? I suspect the SOC chip is an opto-isolator (it only has 6 pins
> > > > and is curiously in a white package). Not sure about the SG chip but I
> > >
> > > Almost certainly an optoisolator. Probably for the voltage feedback loop
> > > (LED on the secondary side of the supply, connected to some kind of
> > > voltage sense circuit, transistor connected to the chopper control side
> > > of things).
> >
> > It appears to be controlling the gate of an SCR, the anode of which goes
> > to the live input via a thermal resistor (which is placed next to that
> > 5W resistor which keeps failing), and the cathode of which goes to the
> > 'hot' ground rail. Some form of overload protection, I assume.
>
> Odd.. I wonder how the voltage feedback works, then. It's _possible_ that
> the chopper control circuit is on the isolated side of the PSU (and
> therefore no opto-isolator on the voltage feedback loop),
> transformer-coupled to the chopper transistors (DEC were fond of doing
> this), but I wouldn't assume that without checking

That appears to be exactly right. The gates for the 8 chopper
transistors link into a small multi-tap transformer. I've not
investigated what's on the primary side of that yet. The 'outputs' from
the chopper transistors driving the switching transformer also hook back
into this small transformer too though via a pair of taps, so I'm not
sure what that's about (yet).

> The circuit you've found sounds like something designed to blow the fuse
> (or at least shut the supply down) if there's a problem. Maybe part of
> the crowbar cirucit (although I wouldn't have thought that shorting the
> input would shut the outputs down fast enough), maybe something to blow
> the fuse if the input votlage selector is set incorrectly (the HP
> Integral PSU contains a triac circuit the sole purpose of which is to
> blow the mains fuse if machine is connected to 230V mains with the
> selector set to 115V).

The latter sounds like a distinct possibility.

The rectifier circuit is actually as follows, running as a rectifier /
doubler depending on the state of J1 (culled from the smpsu repair faq):

                 D1
    AC o-----+----|>|-------+---------+-----o DC (+)
            ~| D2 |+ |
             +----|<|----+ | +_|_
                  D3 | | C1 ---
             +----|>|----|--+ - |
             | D4 | +--o-o--+ +320 VDC to chopper
    AC o---+-+----|<|----+ - | J1 |
           |~ | | +_|_
           +-------------|----+ C2 ---
                         | - |
                         +------------+-----o DC (-)

... except in this case C1 and C2 are actually two capacitors in
parallel rather than individuals, and J1 is under relay control (I've
checked that it isn't accidentally operating and trying to dump 600VDC
into things :-)

> > > I think what I'd do there is connect a light bulb in place of the
> > > resistor (say a normal 100W mains bulb, which should be OK for testing on
> > > light/no load), then pull the chopper transistors and power up. If the
> > > bulb lights brightly you've probably got a short in the
> > > rectifier/smoothing capacitor stage.
> >
> > OK, with the chopper array disconnected (plus everything downstream of
> > it) the bulb lights very brightly when power's initially applied, then
> > gradually extinguishes over a period of about 5 seconds. DC output from
> > the rectifier is 300V. Possible dried-out smoothing cap? The bridge
>
> Maybe OK. It could just be the charging current of the smoothing caps
> (they'll cahrage a lot more slowly with the lamp in series).

I've taken the bridge out of circuit just to be sure and that still
checks out on a meter. I've borrowed a pair of monster 5000mfd caps out
of a known-good PSU so I can rule out probs with the NCR's caps too.

> Now what happens if you add the choppers?

try the whole PSU with the bulb in place, or isolate everything
downstream of the choppers and just try those hooked to the DC output of
the bridge? (i.e. by removing the main switching transformer and the
small transformer that drives the gates of the choppers I should be able
to isolate everything else)

cheers

Jules
Received on Sun Nov 21 2004 - 12:59:17 GMT