PSU = DOH!

As noted in the previous article, it seems there are a few diodes on the back of the relay frame and their orientation doesn't make much sense, given the polarity of the 24V DC power supply.  To this point the polarity of the PSU hasn't had a material impact on the description or operation of the circuit, so I haven't paid much attention to it.  However, with the introduction of diodes, I figure we'd better revisit it.

Original 24VDC power supply schematic

As you've probably guessed, while reverse-engineering the Computox I'm not powering it as I don't want to cause any damage.  This makes resolving things like the orientation of the rectifier tricky.  To this point I based my work on the assumption that the 24VDC relay fuse would be mounted on the positive output of the supply, as shown in the diagram above.

Now I need to state up front that I don't know much about metal plate rectifiers, other than the selenium based ones stink and may present a health risk if they fail.  The rectifier inside the Computox does have a marking on its crusty bracket (perhaps 80/LU 1033B), but I haven't managed to track it down.

Rectifier bracket

I'm also not sure whether the diode test mode on a modern multi-meter should be expected to report anything sensible across such a rectifier, especially when it's in circuit.  Mine appears to measure either 0.12V, 0.18V or 0.29V between the various terminals in both forward and reverse direction, which doesn't seem very sensible.  Using a low current 3V DC input and bulb it is possible show that diode connections exist between the expected terminals, so that's good.  And by using a signal generator to feed a 5V peak-to-peak sinusoidal waveform into the AC input of the rectifier, we can see the rectified output on an oscilloscope.

Rectifier output

The output shows a couple of interesting things.  Firstly, it confirms I've gotten the polarity of the supply wrong, the 24VDC fuse is on the negative output, and I had to reverse the 'scope connections to get this positive rectified output, DOH!  Secondly, its lumpy nature appears to confirm that there aren't any smoothing capacitors on the DC output.  Finally, it seems we only have a half wave rectifier!  This is a bit surprising given the diode measurements, and I think it suggests that at least one of the rectifier diodes has failed open circuit, but I could be wrong on this.  I believe it's difficult to isolate and statically test the individual diodes in circuit because the transformer secondary provides an alternate current path.  However, as the rectified output is in phase with the sinusoidal input, it's possible to deduce which pair of diodes a fault would have to reside with.  I think further debugging would require desoldering the rectifier, which I'm loathe to do.

I also thought it might be interesting to see whether either supply output is actually grounded.  It turns out they aren't.  In fact, the only things that do seem to be grounded are the chassis of the transformers.  Getting the Computox to pass a PAT test might be "tricky".

The corrected 24VDC circuit schematic looks to be the following (where the top-tape and bottom-no-tape names indicate which rectifier terminal the outputs are taken from and whether the wires are marked with electrical tape).

Corrected 24VDC PSU

I probably should have performed these tests earlier, as I now have to correct the schematics that make use of the 24V supply (i.e. most of them).  I'm not going to update the text or diagrams found in previous articles, but once the corrections have been made I'll provide updated schematics for download.  On the plus side, it does mean that the orientation of the EOA and FOA diodes make a bit more sense.  For clarity, from here on I'll use the correct polarity in future articles.