tracing out schematics

From: Jules Richardson <>
Date: Wed Nov 24 17:18:41 2004

On Wed, 2004-11-24 at 01:02 +0000, Tony Duell wrote:
> >
> > Interestingly, there's no feedback between the low-voltage side of the
> > switching transformer and the circuitry surrounding that SOC603B 6-pin
> > IC, so maybe it isn't an opto-isolator at all. All the circuitry
> That does not suprise me. Since there's a pulse transformer in the
> chopper base circuit, the controller IC is on the isolated side of the
> PSU (the isolation is performed by the pulse transformer), so there's no
> need for any optoisolator in the voltage feedback loop.

Uh huh, that makes sense.

> One question. Have you found how the controller IC gets its power? Is
> there a separate (maybe linear) PSU for this?

Hmm, no linear PSU. I do hope it doesn't rely on the battery backup for
the memory in order to start! How do typical SMPSU's start up given that
there's no initial drive to the choppers to kick things off?

> > surrounding it is on the 'hot' side of things, and its sole function
> > seems to be to control the SCR which sits between the live input to the
> > PSU and ground on the bridge output.
> >
> > The main PSU board seems to be responsible for generating +5V only; the
> > second board (which I haven't even looked at much yet) handles +12V and
> > the battery backup control.
> It may also pwrovide +5V. It sounds like the second board handles the
> battery-backed lines, presumably for memory.

Quite possible it does do +5V too.

> > I think whoever made this thing was involved in some "design the most
> > complicated PSU possible" competition :)
> No, that honour goes to the supply in the PDP11/44. The official
> scheamtic is over a dozen A3 pages (!). It's 3 choppers running off the
> same (coke-can size) capacitors

Funnily enough, guess what capacitors I used to rule out that I didn't
have a cap fault with this NCR PSU? :-) They're screw terminals so easy
to borrow for testing, plus I knew they worked.

> , one to provide the rest of the PSU
> control circuitry with power, one for the +5V/+15V/-15V logic supplies
> (the +5V line is rated at 125A IITC), one to provide a 36V line that can
> be battery backed. The last is then regulated down (more switching
> regulators...) to provide +5V/+12V/-12V for the memory. Oh, and there's a
> full-H driver for the fans that runs off the +36V line.... As I said,
> complicated...

Ahhh... I didn't realise there was an H-bridge for the fans; I thought
they were AC direct from ~ 30VAC on the transformer. The fans decided to
keel over in one of our 11/44's after one of the feed wires got chewed
and shorted against the case - everything else is still operational.
Probably toasted one of the driver transistors then...

> > I need to check for shorts on that second board, though. It gets fed DC
> > output from the bridge, but when I tried the lightbulb trick the other
> I will refrain from flaming you just yet :-). Seriously, it would be
> useful to know which, if either, board has a dead short!

OK, now it gets interesting. The second board has unfused DC fed to it
from the bridge output on the first board, plus it also gets DC from the
bridge via an 8A fuse (said fuse is 'upstream' of the choppers etc. on
the first board too)

So, better safe than sorry, I left the 8A fuse out of circuit on the
main board so that it wasn't trying to drive the choppers etc. on the
main board, then I ran the normally-fused DC supply to the second board
via a 2A fuse, just so everything didn't go sky-high.

I left the unfused DC supply to the second board unfused as normal.

Light bulb still connected in place of that 10 ohm resistor which kept

Turned on with the 2A fuse out of circuit first, so the second board was
just plugged in via it's unfused rails. No problem there; light bulb lit
brightly then gradually went out as the caps charged.

Tried it with the 2A fuse in place next. Light bulb lit brightly,
started to dim, then there was a faint pop which sounded like it came
from the big silver filter brick right on the mains input. Light bulb
went out.

Seems like that filter brick has some sort of thermal shut-off in it,
and *that's* tripping now. If I wait a bit I can re-apply power and get
the same thing; i.e. it's not damaged at all. 2A fuse stayed intact, as
did the light bulb. Putting a meter across the filter caps, it seems it
always gets to about 200VDC and then trips.

I assume that the fused supply to the second board runs something which
takes a while to start up (or needs at least 200V to start working), at
which point it tries to do something that relies on the unfused supply,
and this is causing a problem...

Will do some more digging tomorrow sometime... I'll get there
eventually :)


Received on Wed Nov 24 2004 - 17:18:41 GMT

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:37:18 BST