That’s a major milestone for my automation project. I spent the last two years on-and-off rewriting my Conductor automation software based on the experience of the first version.

The first version of the Conductor automation software has worked really well, but I have reached a limit in the 2000-lines automation script, which state machine was getting increasingly complex and tricker to modify. Thus a huge goal of Conductor 2 was to have an easier to understand automation script. In these two years, I wrote and discarded two early attempt prototypes which were not satisfactory, and the current version matches my expectations.

Ironically, from an external perspective, nothing has changed. The automation works exactly like it did before -- which was one of the key goals to achieve. Under the hood, there’s a whole new scripting engine based around the Kotlin scripting language and a whole new route/block based way to describe the automation which makes the automation script easier to write and understand. It gives me more flexibility to add new features.

The new UI with a live block map, sensors’ state, trains’ status, and a new Kiosk Mode feature.

One of the goals of the new scripting architecture is that I can now have some kind of automated recovery. In certain conditions, even if the trains are not at the right place when the automation starts, the script can take an educated guess and bring them back where they should be. The old behavior of Conductor v1 was simply to panic and stop everything, and let human intervention take care of it.

The new recovery mechanism only handles simple cases for now. For example, it purposely does not handle the case of a train stopped on two blocks -- simply because I needed to prioritize features so I had to start somewhere. I will add that support next.

The BD20 sensors on the mountain are being problematic for blocks B321, B340, and B370.

They sometimes “flicker” when the trains run, and do not detect the trains once they are stopped.

The issue has been going on for a while, and depends a lot on the type of automation train I’ve been running. When we started the automation, the passenger train had two Rapido engines in push-pull configuration, and I initially calibrated the sensors for that train. Then we had a set of “older” engines such as Athearn, etc, which triggered the sensors just fine. A few years ago I changed this for modern Walthers or Bachmann SD70s, and the sensor sensitivity was borderline at detecting them -- I’m guessing these engines just use less current than the older kind.

With Orion and Allen, we fixed the lighting on top of the Richmond yard. This area had been fairly dark for a while since one of the fixtures did not work, and changing the other one required some acrobatics. Orion took care of changing the dead fluorescent lights by crawling on the yard to change one of the tubes!

The middle fixture had not been working for a while. It’s one of the 60s-era ballasts that is both antiquated and dead.

It took quite time and effort to remove the obsolete fixture:

We replaced it with a Lithonia CMNS L48 2LL (80 CRI, 4500 lumens, 4000 K).

The area with the new lights in place:

The little saga with the new Walthers Mainline SD70ACe continues as UP 8330 today totally failed to stop while running under automation and kept circling around the layout, unsupervised.

That was the initial issue with these decoders out of the box and it was supposedly fixed by the first firmware update. Apparently not quite. That’s not promising.

Like ScaleTrains, these Walthers use the new ESU "Essential Sound Unit" decoder and so far I have been less than impressed by this decoder. I don’t care that it has less features -- I don’t need any of the features that are missing. What I care about is solid reliability, and I’m afraid this decoder seems to have skipped the QA line for now.

I shall note that these decoders do not even have a product support page on the ESU website. It feels like a half-baked product. Apparently each OEM has their own unimpressive OEM-specific page to list CVs instead. In the Walthers case, it’s a random footnote in the engine’s listing. That is not a practice I like, and I will make sure to not select any more engines with this type of decoder in the near future (or at least till they iron out all the bugs).

The current Branchline engine is one of the two Rapido Amtrak F40PH we have. Here it is in action:

One of the most popular events at Randall Museum is Bug Day.

For the occasion, Jim and Orion ran a dedicated “bug train”:

We’re happy to announce that today, and today only, we’ll be running a fully staffed Junior Engineer Day at the Randall Museum Model Railroad.

To do so, we have acquired the latest most realistic locomotive controller from Walthers, the RailDriver Desktop Cab Controller.

However this controller requires a Windows XP computer. Since we don’t have that (our automation is all Linux-based for 24/7 maximum reliability), we need patrons to provide their own genuine CD-ROM of Windows XP with their own activation code for us to load it on our computers before they can run.

Orion is leading a “spring cleaning” project. While I vacuumed the banisters behind the glass doors, Orion took on dusting off the water stream next to the Sultan canyon:

The water stream, before vs after.

We’re proceeding cautiously. There are some little details (fishermen, boats, etc) that we don’t want to damage.

I also dusted the roof of the buildings on the Branchline (YouBet and Bear Creek) while he was dusting off a bit of the Fairfield area to get a sense of the task.

We decided our next move is to take care of the entire Fairfield Industrial town. Once we’re more confident with our techniques (as in “do not destroy any of the delicate structures”), we’ll deal with the larger Stockton town another day when Allen is also available.

Recently I’ve mostly spent time at home continuing my software development for Conductor 2 and SDB as well as working on electronics projects that may serve for the Randall layout later.

Yesterday I swapped #1072 for #1067 on the automated freight train:

I should try to rotate the engines more often. #1072 was getting a bit sluggish. I need to open it and lubricate all gears and axles.

I’ve also added these two little trolleys here in Fairfield:

The image above clearly shows how much we need to dust the layout. Orion is going to drive that effort next week-end. Once our dusting effort is done, I want to spend a bit more time decorating the unused track.

These two trolleys in the image above are part of a donation we recently got. There are a few interesting buildings and rolling stock I want to display. I still need to go through the boxes, which is going to take some time. The rolling stock is all DC and I can’t run it on our DCC track, but it still makes for some decent static display, especially if we can keep it on a rotation.

Most of the donated rolling stock is Tycho trains, which to be honest are not exactly the most detailed around. So the purist railroaders may just snobbily pout when they see that; however I think our little young visitors and their parents will appreciate the material just quite right -- I, for one, enjoy these old-fashion no-thrills HO rolling stocks. They are colorful and offer quite some variety from the modern too-much-life-like equipment.

My goal is to have enough variety so that we can rotate the displayed items once in a while, as there are a lot of young visitors who just come back regularly and it would be nice if they get to see different things once in a while.

A few updates here. The Walthers UP 8749 was struggling a bit recently, and I swapped it with the new Walthers UP 8330 engine:

I got two of these engines a few months ago to place them on the mainline automation, and it had been a little bit of a saga, so it’s nice to have finally completed that task. Integrating UP 8330 in the automation script was fairly easy, that only took me an hour or two.

I’ve also tried to apply a bit of ATF to clean the track used by the UP engine, and now 1072 is struggling in the large curve after T311. I don’t think it’s a coincidence. Normally I’d go in there and scrub the track with the cleaning wand but this area is out of reach. I’ll have to create a makeshift extension for the cleaning wand some day. In between, I’ve added one of the cleaning pad cars to the freight train, hoping that will slowly clean the track:

I’ve added this currently static display in front of the Stockton Station, with the engine being a temporary DC one courtesy of Orion:

We’ll see who notices :-) More on that one later.