Those Point Switches
Confession time. The permanent short was the classic error, that I really should have spotted, of there being no isolating gap between back-to-back points. Or, at least there was, but I didn't spot that it had been linked across. Some clever DC power routing I suspect. The short happened when I added extra feeds for DCC.
The point operation is an interesting mix of wire-in-tube and OO rail held in bent nails , sometimes both:
You can just see the OO rail going top to bottom on the underside of the baseboard. It connects to the wire-in-tube a few inches away, which then returns to the switch.
I am now left with four gaping holes in the scenery but at least there's plenty of access to attach servos to the tie bars. All of the switches are make-before-break and cause the command station to shut down. I'm using an EZ-Command for testing. Eventually it will, of course, be a SPROG but I can't be bothered to turn the laptop on every time I want to test something
I don't want to disturb the points too much but two switch rails have come adrift from their tie bars, which explains the large gaps commented on in an earlier entry. It's an object lesson in what can go wrong with soldered joins between switch rail and tiebar. The throw of the existing switches is too great, contributing to the larger than finescale gaps. I suspect the alignment was not quite correct, putting pressure on the soldered join when the switch rail was pushed against the stock rail.
Thanks for all the comments!