In my long list of projects to tackle some day are the Stockton Station tracks. There are 3 station tracks. We’ve been using the outer-most 3rd track for the Passenger automation track since the beginning. I’ve always wanted to start using the other two tracks at some point. The first pass, completed today, was to clear the tracks, clean them, and vacuum them:

In the picture above, the two tracks at the left are the mainline tracks. The station tracks, from left to right, are tracks 3, 2, and 1 -- track 1 is the one along the station, which makes sense.

Tracks 2 and 1 were occupied by a static display of a Pennsylvania GG-1 and its passenger cars, as well as a NYC Berkshire and its NYC passenger cars. The Pennsylvania GG-1 and the PRR passenger cars have moved to the unused siding at Walong:

With the tracks free, I was able to test them. The turnouts all seem functional. They all have DCC power from the station panel, and I explicitly marked the block boundaries:

A couple weeks ago, I identified an issue with the Zephyrette getting stuck on one of the Brancheline tunnels. We removed the support pillar that the Zephyrette was hitting. Orion repainted it, and I glued it back today:

The engine is still barely closing the tunnel wall, but at least it doesn’t hit the support pillar anymore. That’s how close I’m talking about:

In the picture above, you can clearly see the timber support that looks freshly brown painted on the left of the engine. It was aligned with the tunnel wall, and I reglued it about 5 millimeters inside. There’s probably just enough clearance for a page from le bottin téléphonique, and not much more. Sorry, French joke, and no I won’t explain it.

OK so I thought that was the end of it but no… The Zephyrette was still having trouble and once in a while, it would fail its automation run, get stuck somewhere on the canyon block, and the automation would bring it back home. It happened every 4 or 15 other runs, not every single time. Took me a while to figure where that was since by the time the automation would tell me there was an issue, the engine would be back at its parking spot. I finally managed to reproduce the issue. Turns out the Zephyrette was getting stuck on the other side of that tunnel. Here’s the spot, and how I fixed it:

Today was the very active Winter Holiday Craft Day at the Randall Museum. This is a large event that always has a lot of crowds attending. In the train exhibit, we had a good variety of trains running, and I brought along my latest project, a live camera car:

This is composed of a Raspberry Pi Zero 2W and a Raspberry Pi camera. It broadcasts a video stream over wifi, which I then display on a tablet using my TCM Android software:

Here’s a closer look at the camera car:

More details on the build can be found here on the project page on the electronics blog.

I finally finished the last phase of the Track Cam Monitor project: the 3rd camera is now installed and configured to show all the back-of-the-mountain trains on the tablet display!

Track Cam Monitor (a.k.a. TCM) is a project I started last year: the goal of this project is to monitor a train staging yard with remote tracks out of sight. The "kiosk mode" Android application displays feeds from RTSP cameras.

In the case of Randall, the purpose is to display the tracks which are not visible on the “other side of the mountain”. When the Saturday Operators stand by the main Valley control panel, they can’t see the tracks on the other side of the room. The tablet projects views from 3 cameras in a concise display conveniently located next to the main control panel:

The cameras installed are Tapo C110 cameras. They are excellent for this purpose -- very small footprint, very lightweight to install, and at a very attractive low price. Here’s the latest one that I installed on the top of the window -- it captures the view of the Summit tracks at the top and the Napa tracks going towards the Sultan tunnel at the bottom level:

As is customary at the end of the year, the Polar Express is back on automation during December:

This uses the Père Marquette #1225 class 2-8-4 Berkshire steam engine, matching the “prototype” from the eponymous book and movie.

The RDC SP 10 failed again on the Branchine automation. Courtesy of Orion and the Willcox estate trains, we are now running the Western Pacific Zephyrette on the Branchine automation:

The Zephyrette on the Branchline Canyon Bridge.
Note the tunnel portal on the image above.

This engine is a great Proto 2000 with a LokSound decoder and a great rich sound.

After a few runs, the engine oddly stopped in an unusual place… That’s where we learned that the Branchline at Randall has some “interesting” curve tolerances involving that tunnel portal -- here’s a close up of the tunnel portal seen on the image above:

Interestingly we had been running the Rapido RDC engines (both SP10 and ATSF) there for years, so how come? Well it turns out they are just narrower by half a millimeter. No kidding -- doesn’t take much. They were actually rubbing a very very tiny bit against the tunnel pillar too.

The problem was easily solved by taking the pillar off (it’s just lightly glued in place) and we’re going to reglue it a couple millimeters in the curve.

If you want to know more about the Western Pacific Zephyrette, I suggest this fascinating read, especially all the insane technical details on page 4: https://www.wplives.com/passenger-operations/self-propelled/zephyrette-01.php 

Here’s the latest visible change on the “train exhibit”:

By the entrance sits the “Vision” screen -- it projects videos from a few live cameras around the layout. When there’s no train motion detected on the cameras, it displays pre-recorded videos I made on the layout -- the same ones I have on my YT channel,  I have about 140 video clips now.

Next to that screen is the RTAC tablet display. It shows the state of the automation -- which trains are running, or what will run next, and details like speed and direction of the trains. That little 10-inch tablet is a ChuWi Hi10 that I got back in 2017, running my custom Android RTAC app, and it has lasted a surprisingly long time given that it’s basically on 24/7. A couple times lately the tablet has turned off, but I got it running again after some careful massaging. Still, at some point, I expect it to entirely stop working so I wanted a plan B before that happens.

And the plan B is to merge the display from the tablet onto the Vision screen display. So that’s exactly what I’ve done, the changes are in the Vision project, and we can see the result above. I actually like it and I removed the tablet since it’s now redundant. Amusingly, plan A was to update RTAC to run on a more modern Android 13 tablet -- I have done that and I had it at home ready for deployment. Out of curiosity, I implemented plan B integrating both features on the same display screen, and now I like it better. That’s why I try things. There is no wrong choice here.

Most visitors will only need to see the status line under the videos -- it indicates whether the automation is stopped, or whether a train runs. For my own edification, I have details on the automation shown on the left side:

That display on the side is purposely simple and not too fancy. I thought about doing something like an “old monitor” look or an LCARS interface and opted not to go for the gimmick.

A lot of the work I do on the layout and its automation has to do with maintenance and “invisible” infrastructure -- things people don’t see at all, but keep the entire thing running. In software development, we call these tasks “KTLO” (a.k.a. “Keep the Lights On”). Part of the invisible part of the Randall Model Railroad Automation is all my remote monitoring of the automation. I can keep track of what the trains are doing, if they are doing it properly, with notifications when things don’t run as planned, and I can (in a limited fashion) remotely operate the trains if needed.

As such, I spend about the last month working on an entire new dashboard infrastructure, which results in a web page giving a dashboard like this:

This page gives me an overview of which computers are on, wherever the automated lines are active (the “toggles”), and which train ran last, and whether it completed its run properly.

I recently expanded that dashboard with a “performance” tab that lets me see how the trains behave on their respective route:

Description of Issue

The top of the mountain is an area known as “Summit”. The mainline has two sidings around the mainline. The siding B372 has never been used and both turnouts on either side are non-functional. It now contains a little station which is where the Automation Passenger Train stops and reverses. The siding B371 used to work fine, and although it hasn’t been used in a while, we may restore it at some point.

A view of the mainline track to be replaced -- from before the bridge up to the first turnout before the signal bridge.

Just before turnout T370, the mainline track was of sub-par quality. There was a couple millimeter gap in the track, and it was held in place using hot glue. Once in a while we would get derailments there, and at some point we got a dead spot, and I did some repairs back in 2018.

Turnout T370 had been the cause of derailments in the past and I had spiked it. Anyhow, the Fulgurex motor underneath was disconnected. Here’s how I left the track back in 2018, and you can see the poor track alignment, and that’s when I was amazed to find the track was held using some hot glue:

Back in 2018 when I soldered that bonding wire to fix the track continuity problem, I was surprised when the part below the solder started melting and that’s how I discovered that this was held using hot glue. It must have been done a long time as it had blended with the rest of the scenery. Anyhow, as I wrote back then, that bonding wire was supposed to be a temporary fix. And now we’re finally circling back and taking care of it.

Since only one of the two Santa Fe RDC engines currently runs satisfactorily, I’m currently running a little experiment with matching Santa Fe RDC 191 with the SP 10 RDC:

That combo works well. I checked and both engines are correctly speed matched. It’s certainly a non-prototypical setup though. I had previously run them together as an April’s Fools joke.

On the workbench, I got a new duplicate of the UP GE 9538:

UP 9538 has been diligently working the mainline since May and, although I initially had some reservations on the TCS decoder, I got it working quite satisfactorily. I decided to have an exact replica -- same engine number, same motor -- so that I can just trivially swap them every other month without having to reprogram the automation every time. That should help give the engine a much deserved break.

I also got a new UP SD40-2. This is an older Bachmann design with an Econami decoder. From experience, these work quite well. I’ve tested 3450 on the mainline, and it seems very adequate. I’ll program it in the automation soon, giving me one more engine to swap from time to time.