The above is an image map, use it to navigate the site. Best viewed at 1280 x 1024 using any browser.
Restoration... what's needed?
First of all, a free confession. I screwed-up. I was under the impression
that this was a 220/240V machine. And I knew it had been running recently.
So I decided to forgo any capacitor-reforming, and performed a quick visual
inspection for obvious faults, shorts, loose parts or cables, and shipping
damage.
And powered it up.
What's supposed to happen: You flip the main breaker. This sends a bunch
of voltages from the 'AC Board' (essentially an auxilliary PSU) to the
Power Sequence Card. This is a jolly clever bit of logic which monitors
things like overtemperature, overvoltage, undervoltage, and overcurrent.
If a PSU fails or a fault condtion causes a trip, the source of the problem
is stored in latches on the board (as long as the main breaker stays on,
that is), and can be read out by the engineer.
It also monitors the operator power on/off switch. When you switch 'on',
it's supposed to pick relays, starting the fans, the main power supplies,
the drives, and the logic, in sequence - hence the name.
Mine did no such thing; throwing the power switch did nothing. Nada. Zip.
Started troubleshooting, Went through the doc. Turns out there are a couple
of manual procedures to bring up power - the simplest being 'operate the
main relay by pushing it in with an insulated screwdriver'! Did that...
lights, fans, disk, CRT comes up to a glowing green dot... woo hoo! Nothing
drastically wrong with this system, but there is clearly a PSU problem.
Check further.
Turns out I was wrong - it *isn't* a 220/240V machine, it's jumpered for
208V according to the serial number plate. Grrrrrrr. Could that be the
trouble? Overvoltage? Hope I haven't cooked anything. OK, restrap the main
tranny (top left in photo below) for 220/240V and retest. Still nada.
Ooooops.
More investigations... see below.
The PSU with covers off. Main transformer top left. AC Board bottom centre.
Power Sequence card is the other side of the bulkhead on the right.
After restrapping for 220/240V, I did more tests. First, the lights didn't
behave quite right - see below. They pointed to a power sequence card problem.
Measured the voltages at the power sequence card test points... all good
except the +5V, which was loitering around +0.7V. Uh oh. +5V is critical,
drives logic on sequence card, operates LEDs on front panel etc.
Damn bugger blast. Perhaps my wrong voltage cooked it.
OK, I have another System/32. Remove the sequence card from it, put it
in this machine for testing. Exact same results! What's going on?
Read the doc again.
Damn bugger and triple blast. The auxilliary transformer on the AC board
- the one that drives the power sequence card. It *also* has to be restrapped
for 220/240V! Use language against myself which is far too immoderate for
a webpage. Swear mighty oaths concering buggering about with mains voltages
& computer restoration when short of sleep due to 16 month old son
with teething problems.
Restrap the AC Board transformer to 220/240V. Retest. Both cards still
behave identically, showing the same fault symptoms.
OK three possibilities:
1. Idiot Computer Restorer has seriously buggered both of them due to wrong
voltage.
2. They both have random faults (perhaps not the same fault) that cause
+5V to go down.
3. The fault lies elsewhere.
Datum: remove the power sequence card and check the +5V supply at the sequence
card backplane (or 'Maple' as IBM charmingly term it). Result? +5.0V, right
on the money. So the problem IS in the sequence card(s). Or is it? Where
does the +5V go, apart from the sequence card? We know the sequence card
takes it out to the operator panel, to run the on/off switch and power
fail & thermal fail LEDs. But figuring it out requires learning IBM
ALDs - Automated Logic Diagrams, IBMs idea of a circuit diagram, which
are, I'm sure, very good once you understand them. Right now I feel they
would be easier to understand if written in Welsh. Or Sanskrit.
Check out a few more things. Do cautious 'finger testing'. Ouch!!!!!! The
+5V regulator on the AC board gets *very* hot, far too hot. Last time I
saw that was on my Fairlight; one of the boards had a short to ground. Pulled the board, it came up
perfectly. Why is the +5V on the S/32 at +0.7V? If it was 0V (short to
ground or dead regulator) it would be easier to understand.So is there
a high-resistance semi-short to ground someplace, enough that regulator
can drag it up to +0.7V, but also enough that the regulator roasts in the
process? Or is the regulator in fact bad?
Well... Datum: even with all output cables disconnected from the AC board
(essentially, the ribbon cable in the photo below), the +5V reg *still*
gets hot. So now I think we're looking at an AC board problem; the regulator
itself, or some other part of the +5V circuit on it. It's *nearly* working,
because remember, with no load from the Power Sequence Board, the +5V at
the backplane is correct, even though the regulator is far too hot.
Stay tuned - or better, if you have a suggestion, get in touch!
Closeup of AC Board - the very hot regulator indicated by the red arrow.
Checking the test points on the sequence card.
Manually pick the main relay... woo hoo! Power! Lights!
Press 'lamp test' - more light...!
...on the operator panel as well. But nothing from 'TH CHK' and 'PWR CHK'
- they are powered direct from the power sequence card +5V, which is where
the problem lies.
Press 'DPLY PWR CHK' and 'LAMP TEST' together; in this mode, the high byte
(leftmost 8 lights) should all light to confirm power sequence card good
(or nature of fault, if faulty). But none of them light - because the fault
is, the +5V needed to drive them is bad on the sequence card (they're driven
from the sequence card in this mode; on a regular 'lamp test' they're driven
from the main logic +5V - totally seperate supply).
Pressing just 'DPLY PWR CHK' - nothing shows. Same reason as above; if
there is a fault, sequence card can't display it due to bad +5V.
The CRT power is fine tho! Tiny green dot, just looks blurred in this photo.
So what does it do at present? I can't boot it when I bring up the power
manually. Not sure why; probably some 'power no good' line is being asserted
and inhibiting everything. One thing I've found does work so far; when
I select 'ALTER MAR IRPT' on the CE panel (essentially the same as setting
an address on a 'traditional' front panel, the LEDs light up correctly
to reflect the address I dial in with the rotary hex switches. Which is
encouraging.
Any other System/32 owners, or former CEs, *please* get in touch!
There's almost nothing about this machine on the web.
http://tinyurl.com/jyul is a link to a usenet post on Google which gives a fair bit of information
about the machine
http://www.fwtunesco.org/musi/wolz/ is an excellent Italian page on the restoration of an IBM 3741 Data Station, a machine with several similarities to the S/32.