On Sheet – When Things Go Wrong

On Sheet – When Things Go Wrong

any of you might remember that incident we had on Tusk Hill (a division of my Tuscarora layout, where we run an English Midlands session rather than a western Pennsylvania one). Here is the disaster report.

So the designer/builder Steve (a good friend whose help has been critical in the development of my microlayout) came out to the club (where we keep it) and took a look at it. He checked the obvious possibilities (particularly the solder joints) and after two hours gave up. It would take a lot more work. So he loaded it into his car (yay, microlayouts) and took it home.

So what went wrong with my microprocessor-driven interlocking tower?

Tuscarora tower operators lounge away the day, their interlocking down (which is odd, given that most of it should be steampunk pull-rod technology (Photo: Zach B)

The first thing was a blown chip (that handled a bank of levers around 13-16 or so – you’ll remember those signals coming up wrong and the “fault 16” warning). He replaced the chip and it didn’t work. Then he found out the replacement chip was also shot and had to replace it again.

Still not perfect. So yes, there was a solder joint that had once again failed. That was reapplied.

But this next one was the gotcha. He was having all sorts of problems with the downloaded libraries used to compile the microprocessor. His son had done a lot of the primary coding (C++) while Steve and I had done smaller sections (I did the fault coding). So we did the original coding in 2021 or so, and it had worked fine. By early 2023 I had gotten fussy that my facing point locking levers were backwards (from how they should work). This came from additional readings in the wonderful The Classic Railway Signal Tower (by Stephen McEvoy). Steve had a new computer by then, so he had to search for and install many new libraries for the chip compilation.

Even though he did all this (updating them since the blue-lever incident at the start of the year) the signal tower continued to fault. He ended up having his son come out (they live about 30 miles apart) and have a look at everything again. Turns out there was a mistake in the original code – something that would set elements of the microprocessor incorrectly. In 2021, the libraries did not apparently use these values. In 2023, the updated libraries did. And so there it was, the reason that the tower would start up, run for a short while, and then crash.

The compiler.

The solder joint.

The chip.

It reminds me of that old movie The Guns of Navarone where the raiders hide a trigger to blow up the guns, but then they hide another trigger even more cleverly. The Germans find the first, congratulate themselves for spoiling the English plot and go back to shelling the English fleet. And then the guns explode.

Just like my signal tower.

What an adventure that was. I took up smoking just for this, so I could pace around my house and burn through packs of the things, just for the drama as I waited for news.

Anyway, once I’m back from vacation, the boys and I will line up another session. Looking forward to it.


P.S. I’ll relate this – while Steve had the layout and tower on his bench, his four-year old granddaughter came in and looked through the tower windows. “There’s humans in there,” she noted. Steve told me that might have been an issue too.