BWD Minilab 603B regulator not behaving as expected

[clay1905]

Hello,
This is not a repair how to request, but a request for knowledge, or at least, where I might find it.
Recently I acquired a bench power supply, maybe for just the right price too. But only if I can fathom a few issues.
In the power supply circuit is a LM317T and a LM320T in a mirror image arrangement. They should output +15, & -15V about a common 0V ground.
The LM320 output is -17.97V. This is with the trim potentiometer turned fully one way.
About the only component in the schematic that I think will make much difference is if R320 drifts. However, it measures to within 1%  of 820 Ohms.
The outputs from this supply are connected to the various circuits via wire links on the board. In order to see what influence downstream circuits may be having, I lifted the -15V link. This had the effect of lowering the voltage even further, down to -30VDC.
If the output were weak, I would think that there's a current path that has become less of an impedance, or a short has developed. But no. The voltage is more negative and I don't quite get how that occurs. In my rather limited experience, regulator IC's should be capable of a pretty stable and accurate output if the controlling dividing network is set up correctly.
I would very much appreciate any advice in trying to determine what is happening here.
Attached is the schematic of the circuit in question. Pencilled in are actual voltages found, except the -30V.

[gbaddeley]

Possible bad connection on r320 or r321, or the trimpot is bad. You have a classic piece of 1979’s Aussie test gear. Could be showing it’s age.

[clay1905]

The trimpot looked good a while ago when I lifted it to check. I do know all the resistors in this network test within tolerances tighter than their bands.
It's odd, and I wish I knew where the supply has been. All the tinning on the component leads is quite grey. I wonder if it's spent time somewhere near the seaside. if the tinning is so strongly oxidised, what else? I'm hoping I don't become the proverbial dog chasing its tail.
Now this is significant. After lifting the -15V wire link, I have measured a resistance across the gap. Only 1500 Ohms. There's only a very slim chance, but I want to be sure that this is not due to the 25V connections. Otherwise I've traced all the nearby tracks, and they're all isolated from this supply. Such a low resistance over a circuit board doesn't look good for the future.
I will lift the trimpot again to test, and as I go I'll remake all the solder joints around this supply network. Mostly though, I think I'll need to thoroughly investigate the stray 1500 Ohms.
It would be a bit of a disappointment if I can't save the supply. From what I've read, this was the last of the "Old Tech" model line, with supplies designed around things that glow warmly. Just the ticket for a nostalgia buff such as myself.

[clay1905]

Rats.
This is starting to look less like a technical matter, and more of a physical problem. A seriously leaky circuit board is going to be a tricky thing.
It might work out that making a copy of the circuit board is the best solution. The board isn't very tightly populated, so I think it's worth keeping in mind. If the resistance isn't somehow a spurious measurement, I'll probably need to move this to the appropriate place. I'll be back with more info soon.

[gbaddeley]

What is the output on the LM320 when R321 trimpot is jumpered, or set to 0. It should be close to 12 V.
You could try changing to LM337, and lowering the GND to OUT resistor to 100R.
Oxidisation is not good news, there may be bad joints or failed components.

[intabits]

Surely RV321 has gone open/dodgy/dirty?

With the load removed, it was the only thing tethering the output towards ground.
If it's bad, the whole U302 regulator section has lost its moorings, and drops down to the full -30v input voltage.

A pic of the board showing its condition might be helpful...

[tautech]

BWD 603B manuals are here:
https://www.kevinchant.com/bwd.html

[mikerj]

Note that the LM320T has a 330R resistor across it, presumably to lower dissipation in the regulator.  If the load on the -15v rail is too low this resistor will pull the rail voltage below the regulated voltage (i.e. become more negative).  Lift one end of R323 and check the -15v rail voltage.  If it's ok then find out why the intended amount of current is not being drawn.

[clay1905]

Thanks gentlemen for your thoughts and suggestions. Due to a sudden onset of work, I have been unable to respond until now.
OK.
-17.97V is as low as I can get the output of LM320 by changing the value of RV321.
I have lifted all of the components associated with the negative side of the supply circuit. They all measure well within their tolerance band.
Lifting one end of R323 obtains -26VDC.
One of the things that's been troubling me is the resistance I found between the ends of the -15V link. There is no return path that I can find.
Now, I read in the handbook, right up front, that the unit provides "seven independent instruments"(emphasis mine). Well, they're no longer independent. There are three different current sources, so that A, B and C voltages could be supplied to power a simple radio valve circuit. The heater supply is the only output that remains properly independent. All the others are now interconnected, with no more than about 4.5K Ohms separating them. What I mean by this is every positive output socket has no more than 4.5K Ohms resistance separating it from every other.
Another thing that I'm not at all sure about is just how easily the tracks on the circuit board are damaged and loosened. Any touch with a soldering iron and the tracks are off the board. I'm absolutely loath to lift anything without a very good reason, as a repair will be definitely required.
I have also found that virtually every solder pad is connected to every other pad on the circuit board. Some have many KOhms, others a few hundred.
Between this and the apparent interconnectedness of everything, I have a hunch this is bigger than a local power supply issue. I'm starting to wonder about what can be done to a fibreglass circuit board to make it so conductive, and so easily damaged.
Below are a couple of pictures. One shows where the 1500 Ohms was found. This is the backlit shot, and the total distance between the ends of the wire is about 12mm. The other picture is a general shot of the track side of the circuit board. In the middle region are some ugly blobby repairs. These are mine and temporary if the unit can be repaired fully. They were caused by lifting components for testing. No drawn out dwell times here either.

[clay1905]

Just had another look at the schematics (I was lucky enough to get an original handbook with the unit) and the three power supplies indeed *should* be independent. They all have their own power transformer windings, and no taps. Each is properly independent. Yet they're connected together by the time the circuit reaches the output socket.

I suspect that these boards have been steeping in subtle fumes in somewhere like a school laboratory storeroom. Over a long time of disuse and neglect, these boards have absorbed something that makes the copper detach, and introduces conductivity into the substrate. Whatever the fumes were also made the component leads oxidise quite a bit.

[gbaddeley]

You could try taking out the PCB and give a good wash with distilled water and scrub with a tooth brush. Dry in the sun.

[mikerj]

Lifting one end of R323 obtains -26VDC.

Just to confirm, lifting one end of R323 causes the output terminal of the LM320T regulator to increase to -26v?  That makes no sense at all, are you sure you were measuring at the correct point?

[clay1905]

Thanks, and indeed I will. I'm thinking about going the whole hog, and giving a good scrub with isopropyl alcohol first, followed by a rinse in the purest rainwater, collected by virgins on a full moon in Icelands......
He's at it again.
Thanks again. A good scrub could be just the thing. Just now it's pouring down though. I think it'd be a good thing to wait until some drier, warmer weather.

[trobbins]

If you have removed all external loads from the regulators, then I'd first suggest you add temporary loading and also an output cap as typically suggested by the IC manufacturer.  If you can't achieve local regulation then I'd suspect each regulator and do a temporary swap of cutover to confirm you can get nominal operation.  There are also a maximum input/differential limit of just 25V, so any form of output short, or loss of R320 or R321 connection would exceed ratings with a 30VDC input - which may have happened in the past.