Another old postcard recently added to my collection shows the Copper River Railroad (incorrectly labeled Railway on the card) grade on the south side of the Odiak Slough. The Copper River Railroad was the original line built out of Cordova in 1906 by Michael Heney (of WP&YR fame), and sold to the Alaska Syndicate and the Copper River & Northwestern that fall. You have to love those horse-drawn narrow gauge construction cars.
Progress this year has been slow. Between my real job, Iowa Scaled Engineering and projects around the house, I haven’t spent a lot of time in the layout room since March. The biggest thing I wanted to get done before getting back to the layout was to upgrade the furnace and finally, after 20 long years, add air conditioning to the house. Believe it or not, it wasn’t terribly common out here until about maybe the last two decades. There would always be a couple weeks in the summer that were terrible, but otherwise opening the windows and having a few fans was adequate. However, with working at home every day now, the higher heat load from all the computers and such, and the pervasive smoke from forest fires in the West in recent years, I decided the time was at hand.
I’d been procrastinating because adding air conditioning to this house (built in 1977) was also a non-trivial exercise. The house didn’t have a large enough electrical feed to handle it, the breaker panel was full, and my old furnace couldn’t handle it anyway. So it was an exercise is what programmers refer to as “yak shaving” – the seemingly endless series of small tasks that have to be completed before a project can move forward. “But to do X, I need to also upgrade Y, and to upgrade Y I need to also fix Z…”
I finally got fed up with it and pulled the trigger in July. Since the furnace is in the layout room and any AC lines would need to work around the layout, I wanted to get it done before any scenery started this fall. And it’s truly amazing what you can accomplish when you just open the checkbook and get some true professionals in to look at the job. In two days, I had a new heat pump, furnace, and the house was updated to 200A power. And now the whole house – including the layout room – stays nice and cool all day long.
So what I have I done on the layout lately? Not much. About the only thing I really have to report is identifying a source for new signals. While I love the detail of the Century Foundry searchlight kits, getting enough light through those tiny fibers and just the time it takes to build them isn’t the best use of my time. So I’ve been looking around at other options.
At Spring Creek’s Deshler train show back at the end of July, I found some signals from Custom Signal Systems at Azatrax‘s table. They’re quite nice, though not quite as detailed as the CF kits, and relatively affordable. They’re also nice and bright, which will be important for an operations-focused N scale layout.
Today’s old photo is of the snowsheds over the line, as shot from what appears to be the fireman’s cab window. The lower part of the line receives unbelievable amounts of snow, and where it ran along the steep banks of the Copper River, avalanches were a frequent occurrence. This is the only photo I’ve ever seen of what I believe to be four snowsheds along the Abercrombie Rapids section of the Copper River, just to the north of Miles Glacier bridge.
But there’s the nagging question of which locomotive is it?
That box with the tube sticking up visible at the bottom part of the frame is unmistakably a slide valve on top of a piston, which means this is an early locomotive. Of the CRNW fleet, slide valves limits it to the 0-4-0Ts, 20-22 (23 appears to have piston valves, or perhaps acquired them later in life), 50 (and likely 51 if it existed), and 100-102.
The bevel below the slide valve box into the main cylinder removes the 20s, 50, and the 0-4-0Ts from consideration. So it’s probably one of the 100s, but then there’s the two foot ledges you can see in the picture. That doesn’t fit any of them that I’ve seen pictures of, so perhaps we can’t really tell which engine this is. Thoughts?
A few recently acquired photographs landed in my mailbox today. I’ll post them over the coming weeks. The first one is one of the CR&NW’s steam shovels loading rock into side dump car #13, with car #85 right behind it. The location and date are unknown, but judging by the trees it looks like a forest fire has been through the area recently.
Records of forest fires in Alaska from these early years are pretty sparse, as there was little way to monitor and track them systemically. However, we know the Gilahina Trestle burned in 1915 from the Sourdough Hill fire, which was attributed to the railroad and burned some 384,000 acres from Chitina to Kennecott. There was also the 1915 Kennecott Fire, which burned 64,000 acres between the Kennicott and Nizina rivers. Still that’s just a guess, and even if correct doesn’t really narrow it down.
While I haven’t posted much lately, work has been continuing slowly in and around work and personal commitments. I’m pleased to say that I finished configuration and checkout of the final control point tonight at South Chitina. CTC now works over the entire length of the mainline. (The actual trackside signals will be installed once base scenery is done – for now it’s just the fascia signal repeaters.) Most of the rest of the electrical is also complete. The only thing really remaining is to create and wire control panels for the various yards. Then it’s on to starting scenery.
I was cleaning part of my workbench today and ran across the camera car I hacked together back in 2016 to do my original “tour around the layout” video. I wouldn’t call it a great camera, but it’s relatively small and it produces okay video for its size. My other option was keep installing switch machines and track wiring in the Eyak smelter yard, so putting together an updated cab ride sounded like more fun.
Unlike the last one, which started at Kennecott, this one will start out on the (fictional) Nizina Branch and enter the main line at Nicolai Junction (again, fictional, and between McCarthy and Kennecott). The Nizina Branch is located above my workbench, and provides some staging room and a little more operational variety for ore trains from “other mines” up around Dan Creek.
From there, it’s an engineer’s view all the way to Cordova. Stay tuned at the very end for a couple bloopers. Turns out my camera car is far wider than the loading gauge for a typical N scale car and a bit top-heavy, so it doesn’t always behave itself now that some of the layout clearances have gotten a bit tighter.
Some 11+ years ago, I started messing with servos and PIC microcontrollers as switch machines. Originally it was just for Ron Renner’s Wind River in Denver, where we didn’t want to spend the money to upgrade all the twin coil machines to Circuitron Tortoises. In an afternoon, I had a primitive workable design, and Ron’s layout eventually got dozens of them installed. Each was hand-built and installed in place using double-sided foam tape for both the servo and the PCB.
Those early prototypes then went on to become the basis for Iowa Scaled Engineering’s MRServo line a couple years later. The MRServo matured over the years, moving to surface mount components and then proper 3D-printed mounting brackets to make installation more reliable. But fundamentally, the design remained little changed from the earliest units.
We (Iowa Scaled Engineering) discontinued MRServo about three years ago because sales had tapered off, more large players had entered the market (Walthers) and we never got any sizable market share away from the dominant vendor, Tam Valley. There was nothing about our offering that was fundamentally compelling and differentiating from the others. Plus they were a pain to build and support for relatively little profit margin.
The MRServo Version 4
That’s not to say MRServo died, though. It just went from being a product back to a project. My layout still uses them exclusively as switch machines, with probably close to 100 installed. They’re great for me because they’re relatively inexpensive (especially if I’m building them myself), they’re low profile (so they fit up inside the benchwork without issue), and they take a logic level input to control them (so they’re easily run by the signal system). A few months back I got the itching to again improve the design. I wanted to improve the input protection structure, which often got static-zapped, and also move to an AVR so that I could make changes in C rather than PIC assembly. When I designed the original servo controllers, I was still using PICs regularly and programming in their assembly variant. However, having gone to AVRs and GCC around the same time, I haven’t used a PIC in a new design since.
One other minor change is that I’ve started putting 2-piece terminal blocks on the boards. That makes it significantly easier to connect wires under the layout, as I can pull out the connector, screw in all the wires, and then just plug it back in.
Like all of its predecessors, the MRServo v4 is open hardware. The source code and design files are all available on Github.
Since I haven’t posted pictures of the whole layout in quite some time and it was all nicely cleaned up for the CR&NW’s first work night last week, I thought I’d show you the state it’s currently in. The track is complete, except for a few spurs and industry tracks that have yet to be positively defined. The backdrop and fascia is 99% complete. The lighting is done, the signaling is 95% done, and track power wiring is 95% done. I’m still installing switch machines at Chitina and Cordova, but that should be done in a week or two. I’ve also posted pictures on the backdrop to inform others and remind myself what some of the scenes are supposed to look like. It’s a combination of my photos from visits over the last decade combined with historic imagery.
Once I get those last few benchwork and electrical to-dos resolved, it’s on to scenery and painting!
The prototype CR&NW, running only a through train or two every day even at the peak of operations, never really had a use for signals. Operations were handed with train orders telegraphed to stations, and honestly from all evidence that survives, it worked just fine. There’s no record of any major cornfield meets anywhere on the CR&NW during its ~30 years of operation. (An astute observer may try to point out that there are no cornfields on the CR&NW, but my point holds.) The only signal we know of on the original line was a wigwag crossing signal in Cordova.
The N scale CR&NW, however, was always intended to be signaled. I love signals. I’ve been fascinated with them since I was a kid. I have two full size signals in my back yard and part of one in my kitchen. Since my Copper River is set nearly 80 years after the original went out of business, I can take a few liberties with how the line evolved. Plus, with keeping 3-4 operators busy during operating sessions, that’s going to be a fair amount of traffic in only ~10 scale miles of track.
I’m going to write a series of articles on the operation of and the technology behind the system over the coming weeks, but in the meantime I’m going to put up a screenshot of the Computer-Aided Dispatch (CAD) system that will be used by the dispatcher to route traffic. Visually it’s very nearly complete, and about 75% of the functionality is there under the covers. The display is a little boring since I’m developing it from Iowa and while it’s communicating with the layout back in Colorado, the layout is mostly not on and thus not sending any data back. The fast clock is the only thing up and running.
I’ve always liked the idea of fascia signal repeaters, just so operators on through trains don’t need to push to push their way into every corner just to see those inevitable awkwardly-placed signals. Is it prototypical? Probably not, although in either cab signal territory or now as we’re pushing into the new world of PTC, it’s not that far off. It just makes for a smoother operating session, in my opinion.
One option is 3 LEDs per signal head – one for each major color. That’s a lot of LEDs to make room for on the panels. Another option would be red/green bi-color LEDs, energizing both to get a mixed fake yellow color. That has the advantage of one hole per signal head, but the problem is there’s often significant variation from one LED to the next (requiring tuning the color of each individually), and they have a strong color shift based on viewing angle. What I really wanted was a true three-element LED, with true red, amber, and green emitters. The problem is that most of the very few true red-yellow-green LEDs are surface mount, which doesn’t lead itself to being installed in fascia panels.
There is one option, though – the Lumex SSL-LX5097SISGSYC. It’s a standard 5mm LED, albeit with four leads, and available off the shelf from Newark. The red and yellow are reasonably bright, but the green element is a bit weak. Still, some experimentation on the bench showed that they were more than bright enough for panel indicators, and with the proper resistors could be nicely evened out to the eye. At 5V, putting 1k on the red lead, 560 ohms on yellow, and 220 ohms on green provided subjectively even indication brightness.
Since I didn’t want to solder resistors and leads on to something like 60 LEDs, I built a small PCB that integrated a place to mount the LED, the three dropping resistors, and a small JST SH-type 1.0mm 4 position connector. Panelized into groups of 7, it cost me $30 to have 210 individual boards made at PCBWay. After that, it was a matter of getting a solder stencil, slathering on paste, placing components, and then reflowing them in my toaster oven that’s dedicated to PCB manufacture. For those interested in making their own, the schematic, PCB, and Gerber files are all under the “ckt-sigcon” files in the ISE CKT-SIGNAL project on Github.
Once assembled, it was a simple matter of soldering in the LEDs and installing them in some of the new control panels that we’ve been 3D printing. Connections to the little LED boards can then be made with 4-wire cables available from a number of sources. I have a large supply, since Iowa Scaled Engineering uses them for connecting to our newest versions of the TrainSpotter detectors. However, Sparkfun also sells them for their Qwiic I2C network system, and I’m sure you can find more sources. Ours are custom made by DirtyPCBs.
The end result is a very nice looking signal repeater panel with very clear indications in both normal room lighting and in “night” mode.