Sunday, April 1, 2018

3d Printing A Crowd of Passenger Cars: Harriman 60-C-1 to 60-C-4

In any operating model railroad, the engines and the cars are the characters in our play. And just like in a play, crowd scenes need a crowd - only in this case, our crowd are lots of trains and lots of freight and passenger cars. Collecting enough rolling stock for a model railroad can take a while as we build kits, shop, and otherwise put together an appropriate set of rolling stock for our layout.

On the Vasona Branch layout, I’ve accumulated my freight cars over years, slowly getting just the right mix of cars. The layout started out with some freight cars from my teenage years. A few years out in the wilderness of suburban New York City got me a bunch of PFE refrigerator cars, all picked up at Valley Model Trains's great location out at an abandoned mill in Wappingers Falls. I’ve bought and built additional models - some resin, some plastic, but all trending towards the era I chose. I’ve gotten rid of unprototypical cars, bought others that better match my 1920’s California setting, and now my layout has the depression-era look of a sea of brown boxcars. I’ve got my crowd scene.

With the Market Street layout, though, I’m back in the position of not having enough characters for my crowd scene. This time, my problem is a lack of passenger cars rather than freight cars. I haven’t had to model passenger operations significantly in the past, so I don’t own a lot of equipment. And I need a crowd.

My current Vasona Branch layout, after all, only needs two passenger trains - a Los Gatos commute and the Santa Cruz to San Francisco train. The commute is served with some ok 1980’s, Soho brass commute cars. The Santa Cruz trains use a string of 1960’s Ken Kidder cars. Both cars aren’t particularly detailed, but they’re affordable ($100 for the Soho cars, and $40 each for the Ken Kidder cars), take well to modification, and aren’t particularly rare so they’ll turn up on eBay if I’m lucky. I’ve got an assortment of other passenger equipment - baggage cars from the old MDC kits, a Southern Car and Foundry 70 foot baggage car that can’t quite make it around my 24” curves, and one or two odd brass cars - but otherwise that’s it. Great for a branch line out in the country.

The Market Street layout, in contrast, needs a crowd of cars. It needs four or five commute trains, each with at least three cars. It needs several baggage and postal cars that will be added and removed from some of the commute trains. And it needs at least a train or two to represent long-distance trains passing through.

That’s a lot of cars - fifteen or so coaches, four baggage cars, four RPOs, and another train set or two.

Well, I’ve got a 3d printer, and I’ve already made freight cars on it. What could be so hard about making Harriman cars?

These models represent the 60-C-1 to 60-C-4 Harriman passenger cars built starting in 1910. The actual cars were the first steel cars for the Southern Pacific. They ended up in wide use across the SP system. More importantly, they were common cars on both the SP commute and short-distance trains such as the Santa Cruz to San Francisco run. The cars originally had gas lighting, but switched to electrical lights by the late 1920’s - a key detail to add for later cars. The 60-C-1 to 60-C-4 cars were only the first Harrimans; there were other series (both as plain coaches and more comfortable chair cars). But these earlier cars were near-identical and look better to my eye. Later series had different window heights or spacing, differences in doors, and a bunch of other minor differences.

These photos show the first two presentable cars I've 3d printed. There's still some details to get right, and there's still some challenges in assembling them precisely, but these models do show that a crowd of passenger cars can come off of a 3d printer.

The cars have been a bit of a challenge. The cars are too long to fit in the printer, so they had to be made in sections that I'd assemble into a single car body. I ended up making the cars from four pieces: a vestibule and car end, the body in two halves, and another vestibule. I printed the bodies vertically, just like I’d done on the flat cars and Hart gondolas. This meant that only the cross section of the car was supported, minimizing support structure and keeping the support structures off of finished surfaces of the car. The car ends couldn’t be done this way; they print upright, with the support structure attached to the steps and to the coupler pad. The bottom ends up being coarse because the side facing the build platform never quite prints right because of lack of support, but the overall part prints fine.

Like all the 3d models I’ve done, the cars are also challenging because of all the niggling little details I need to understand in order for the car to look right. For example, the roofs on Harriman cars were overlapping steel sheets. The lower sheet always was placed between windows; the top piece always lined up with the windows. That’s a trivial detail, but needed to capture the look. Minor inaccuracies in window spacing is a glaring problem for anyone who's researched the car. Getting the roof curve correct took multiple tries, and I'm still not quite happy. Other little details - like the beading at the intersection of the roof and side - turn out to be more important than I expected. Without the beading, the cars looked wrong, and it was hard to spot the point where the roof curve ended. It also was helpful for brush painting - the brush could be drawn across the bead to paint it neatly, and the bead would stop the brush from going further. I could cheat on any of these details, save myself a bunch of time, and have complete cars sooner... but they wouldn't look right to my eye.

Enjoy these photos of the cars and the pilot models; I’ll talk in-depth about the cars and their construction in upcoming posts.


Thanks to Jason Hill - his experiments using Shapeways to print wall sections for MDC Harriman cars inspired me to try making a whole car.

Monday, March 19, 2018

Comparing 3d Printed Models Against Kits of the Past

When I talk about those Hart gondolas I made a while back, I like to highlight how the models represent freight cars that haven’t been done accurately in the past. There’s no resin kits or brass kits for Hart gondolas, let alone for SP's own version. However, combing through old Walthers catalogs will show two Hart gondola kits: the craftsman kits from Silver Streak in the 1960’s, and Train Miniature’s plastic shake-the-box kit from the 1970’s. Both are good for their time, but we can do a lot better in the 2010’s. How much better? I picked up examples of both the Silver Streak and Train Miniature cars at a train show a couple weeks back. Let's compare them against the 3d printed model and see how forty years of technology has helped model building. (Click on any of the photos for a higher resolution version.)

Let’s first look at the Silver Streak kit. For the time, this is a neat kit. It has the underframe trusses from the real car, brake cylinder mounted on the side of the car, and matches the SP cars with 8 spaces between posts on either side. Just like the modernized SP cars, the Silver Streak Car has grab irons on each end supported by a short vertical piece of wood.

However, the kit lacks the car sill and floor you'd see on the real cars. Instead, it has the car side boards going all the way down to the bottom edge of the car. It’s also missing all the door hardware; the real cars had castings at the bottom of each post, but that’s a pretty tiny detail to include. The modeler who built this kit didn’t quite get the bulkheads at the correct location - they should line up with the outer post.

The model’s a little coarse with 6x6 strip wood serving for the top rail and for the posts, but it’s a fair tradeoff for intermediate modelers building their first car. It’s nice to see the board detail on the inside faces of each side. Overall, the car is a bit oversize, with the sides measuring almost five feet high compared to three on the real car.

Here’s the Train Miniature car. Again, it looks like it got inspired by the Southern Pacific cars that would have been seen in the 1930’s and 1940’s… or they just copied the Silver Streak car. The car has the correct eight spaces between posts. The grab irons arrangement doesn't match any of the real cars, though. More importantly, the trusses are pushed out to be even with the sides - definitely not how the real car was built, and a detail that hides one of the neat details of the Hart design. Like the Silver Streak car, Train-Miniature left out the car floor visible on the sides. Again, the car sides are taller than they should be - partially out of scale, and partially fallout from removing the side sill.

And finally, here’s one of my models. I pulled out one of my “original” cars just to highlight the detail. 3d printing gives us a lot of advantages, including the ability to throw in all that detail for the door hinges , the door latch mechanisms on the posts, and the various bolts all over the model. The truss is lighter than the Silver Streak car, we can see the car side frame and floor sticking out beyond the car sides.

Here’s all three from the top: Silver Streak on the left, the Train Miniature, then the 3d printed model. The Silver Streak car did correctly model the sloping hopper. It’s not perfect; this kit shows the hopper as incorrectly extending up along the bulkheads on each end. But I’m pleased to see they included the supports that ran through the hopper, even if they’re not quite at the right location. There’s no detail on the hopper doors, but then that’s a pretty tiny detail.

The Train Miniature kit’s hopper is hidden by the load, but that brake cylinder and brake gear in each end is completely wrong for the car, and misses the fact that those partitions were meant to be removable so the car could be used as a typical gondola.

And finally, for the 3d printed version. We see the braces running through the hopper (with the notch to hold the 4x10 that supports the doors when closed. We see the end bulkheads definitely look removable. On the far end of the car, you also might see the hinged apron that allowed running a plow through all the cars - a detail that wouldn’t be needed on either of the other modernized cars, but does highlight how 3d printing lets us throw all that sort of detail on the car.

Finally, here’s the underside of each car - Silver Streak on the left, mine in the middle, Train-Miniature on the right. This photo highlights how the other two cars are a bit oversized compared with the actual cars. Both Silver Streak and Train Miniature made some parts oversized (like the trusses) and also placed the trusses differently to make the car easier to manufacture. Both also had to lose some of the interesting detail: braces for the trusses, side sills, etc. in order to make an economical and easy to build kit.

All in all, the Silver Streak and Train Miniature kits are fine for both their time and for what they’re intended for. They had to design parts to be manufacturable. Train-Miniature moved the trusses out to the car edges so they could be injection molded in one piece. Silver Streak made the trusses thicker to survive manufacturing and clumsy fingers, and left off detail on the hopper so the fiddly shape could be made in cardstock. Both kits needed oversized parts for easier assembly, dropped details to keep part counts low or permit injection molding, and did the best they could from the photos and plans they could find.

The 3d printed model gets to benefit from being 30 years in the future. I had access to the SP blueprints which the earlier manufacturers may not have had. 3d printing meant I could make parts closer to scale, and could easily add details and embossing that would have fouled up molds and part ejection. 3d printing also allowed me to refine the models, and quickly make variants: doors up vs. doors down, or the modernized cars without the side dump doors. If I found some railroad back east had a similar car but with a minor tweak, I could make that too. That's a luxury that anyone doing injection molding doesn't have. Cutting a new mold is an expensive, start-from-square-one sort of action. Anyone trying to run a business would want to cut those molds once, and getting a detail wrong isn't enough of a reason. Doing a different model requires a different set of molds; again it isn't worth doing unless you're going to sell a ton of the new design.

All three models also show how model building's changed. Silver Streak's kit dates from the craftsman kit era, and a time when you could run a reasonable model railroad manufacturer out of your garage. As long as you could cast white metal and cut strip wood, you could sell a model railroad kit. Because of the lack of good models, if you had an even partially realistic model, you'd have a hit. Train Miniature dates from the heyday of small-scale plastic kit manufacturers. You need a lot more skill and equipment to make injection molded parts, and even more to print the car sides. Worse, the effort needed to cut those molds meant you had to sell thousands of cars - great for a forty foot boxcar, but not so good for an odd misfit maintenance-of-way car.

3d printing gets us back to those garage days. My Hart gondolas were, after all, made in a garage. (Or at least I wash the extra resin off the printed models in the garage; the printer stays inside to stay away from dust and cold.) And yet the models still have some pretty impressive detail - approaching resin kits, but certainly better than the kits we saw in the 60's and 70's. We'll see more 3d printed models like these in the coming years, and it'll be great to see the prototypes that folks find interesting enough to manufacture in their garages.


In case you'd like a Hart gondola for your Southern Pacific, Union Pacific, or Pacific Electric layout, I've still got 3d printed kits available. Check out photos, prices, and ordering details over at Dry Creek Models site.

Sunday, February 11, 2018

Movie Night XXVIII: Things You Ought to Know About San Jose

History San Jose has apparently been busy. Their History San Jose channel on YouTube has a bunch of new videos and interviews. They also have a pair of promotional videos from the 'teens and twenties, one focused on San Jose and the other on Santa Clara County, both in the same YouTube movie. The first half - the Santa Clara County half - is the more interesting one, showing both the operations at the George Hyde packing house and cannery in Campbell, and an apricot plant pit which I suspect is >Sewall Brown's plant at Vasona. The George Hyde videos start around 11:50, there's a scene showing loading a sulfur house around 17:00. Processing apricot pits starts at 18:00. There's two men shoveling pits from a huge pile, which might explain why the Sewall Brown fire in 1958

At some point, I'm going to have to build that office at the Hyde Cannery; having the video, with the horse-drawn wagons crossing in front of, should be a great starting point.

Sunday, December 24, 2017

Movie Night XXVII: The Track-Pull Tractor Survived

Earlier this year when I reported on the Bean Spray Track-Pull Tractors, I'll admit I didn't think the tractors had changed the world. Even if one of the tractor startup guys had spoken at a Bean Spray employee all-hands about how "Our Tractor" would change the market, there wasn't much evidence that the tractor had much of an impact. Sure, Jim McCollough did another startup with a similar tractor in the 1930's. Sure, Alf Johnson was still testing tractor designs into the 1950's. But it's not like Bean Spray, and later Food Machinery Corporation ever built more tricycle tractors... right?

YouTube proved me wrong. It appears the Track-Pull design influenced Bean Spray products for years afterwards. Food Machinery Corporation, the name for the company after multipler mergers, was selling a similar three wheel design as late as 1945. This 1945 tractor isn't the same design: it uses rubber tires rather than a Caterpillar track, the engine is front and center instead of on one side, and the tractor is surprisingly tiny compared to the tractors that rolled down Julian Street in 1918. But it's still the same, with a similar fork supporting the two wheels, and similar huge wheels to control steering. I suspect it was a great tractor for a small orchard, or for cleaning up a large yard.

1945 Food Machinery Corporation tractor for sale. $6K OBO.

1945 Food Machinery Corporation tractor for sale. Part 2

For a crazy Sunnyvale tractor startup, it's pretty neat to hear their design was still in production 30 years later. A bunch of us out here in Silicon Valley can't claim such a victory.

Thursday, November 2, 2017

Market Street: The Train Departure Board

Short shameful confession time: I built the Market Street layout because it was easier to build than a video game.

A non-trivial bit of my inspiration for the Market Street layout came from the railway switching games made by SIAM Railway Simulations. These mostly-British prototype games provide accurate simulations either of a station or for dispatching a long set of track. The station simulations put you in the role of the tower man and switcher at anything from the end of a country branch to the busy through station at Crewe. The dispatch sections include British settings (Scottish Highland line) as well as realistic versions of Tehachapi Pass, either in the 1920’s or the 1960’s. I’ve played some samples and bought other games.

SIAM's samples page lets you get a taste of the games by downloading demonstration, reduced versions of their simulations. The samples include dispatching the Highland line and Tehachapi Pass, and handling switching in the stations at Crewe and Penzance. The Tehachapi dispatching game gave me a great appreciation for getting trains over Tehachapi Pass in the steam era.

Penzance, 1987 game from SIAM Railway Simulations

Of course, I wanted to try some more American prototypes - handling the traffic coming and going from the Oakland Pier, perhaps. Or why not the San Jose Market Street station? I got a fair way through making an iPad game so I could switch those places, but I found it difficult to control all the switch engines and car movements I wanted to include. After enough tries, I found myself asking "so what would a layout set up like one of these games look like?

So, I’ve built the layout, and I've got an answer to that question.

However, to make the layout operate like the game, I need to know what trains need to come and go. More importantly, if the layout is being exhibited, I want folks to see the names of the trains coming and going so they can understand the sheer number of trains that could be handled in the 1920’s, even if the station was a creaky old Victorian barn.

Suddenly, that way-too-large display on the Raspberry Pi has a use.

Snippet of departure board. Click here to try it out for real.

My inspiration was a European-style flip-card station sign; it’s not quite prototypical for the era, but anyone who’s been in a large train station has seen one of these, so showing a listing of trains should both inform and hint at the sheer number of trains they’re seeing.

And luckily, making such a sign is easy. One of the neat habits of programmers these days is to share programs they’ve written so others can use those programs. A friendly guy named Paul Cuthbertson liked the idea of drawing flip-style departure boards, and based on an inspiring article, went off and wrote a web page to draw one.

Now, I need more than just the departure board graphics for the Market Street layout. I need a sped-up fast clock to show the current simulated time on the railroad, for the compressed track plan means trains won’t take as long to move around as on the real thing. I’ll need a way to control the board when trains arrive and leave. I’ll need a way to manage problems - backing off accidental train departures, or stopping the clock in terms of problems.

I took Paul’s code and created this departure board for San Jose Market Street. (See this Github project for the source code itself.) Pressing a number key (1-9) causes the nth train to move to its next state; a train that’s on its way will go to “arriving” on the first press, then “arrived” on the second, then will be removed from the board on the third.

Paul’s card-flipping web page turns out to require a lot of computing horsepower to flip through all 26 characters for each change. The poor little Raspberry Pi, being the size of a credit card, couldn't keep up. Instead, I changed the program so that the board only performs 6 flips per letter, instead of flipping though the whole alphabet like the real flip board.

So now, when I set up the layout, I just need to set the box containing the DCC electronics somewhere both the audience and operator can see. The stationmaster can use a keyboard to advance the departure board as trains arrive and leave. The operators know what trains they're building; the audience gets an understanding of how easily they could commute from San Jose to San Francisco (or Salinas) in 1928.

And I get a modular layout that's inspired by a video game.


The modern practice of sharing sample code and reusable libraries is a great part of modern programming culture. Great thanks to Paul Cuthbertson for the core of the departure board code!

If the departure board idea might work for your layout, download the sources from GitHub and customize it for your own use. Drop me a note if you find it useful!

Market Street: Electronics and DCC

From the start, I knew I wanted the Market Street layout to have multiple operators working in parallel: switch crews making up and breaking up trains at the station, road engineers coming from the roundhouse, and taking a train out, and freight crews switching the cannery and packing house on the layout. That meant I needed to choose a DCC system for controlling the trains, and I needed to decide what kinds of throttles to use. There were two obvious choices: I could go with Digitrax, the standard for Free-Mo modules, or I could use the same system I had at home - EasyDCC. Neither was attractive. I’m not fond of the Digitrax system; I’ve always had trouble understanding how to use their controls, and have seen too many cases where an errant button press disabled a throttle. EasyDCC would allow me to reuse my existing throttles, but I’d still need to buy a new command station and booster, find an enclosure for both, and then wire all the modules with coax to allow wired throttles to be plugged in around the layout.

Another engineer running a train on the Silicon Valley Freemo-N layout with a phone

Luckily, I was reminded of the local Silicon Valley Freemo-N group. They’d set up their dozen modules (most based on Bay Area scenes) at the 2011 NMRA convention up in Sacramento. When I stopped by to check out the layout, Dave asked if I wanted to run a train. “Sure, but I don’t have a throttle.” “Doesn’t matter,” he replied. “You’ve got a phone - just install WiThrottle.” They were using the iPhone-based WiThrottle, all talking to some random command station connected to a computer. I downloaded the app, and within a few minutes was running a streamliner across their layout.

Fast forward to last year. At the Bakersfield NMRA convention, one of the vendors had the SPROG 3 - a tiny DCC booster and command station that can power smaller layouts. It could be driven by the dirt cheap Raspberry Pi computers and the JMRI software, making for a full DCC system in less space than a cigar box. This seemed like the perfect choice - wireless throttles for the layout meant no wires for a throttle bus. Operators came with their own throttles, so I didn’t need to shell out hundreds for throttles. It could all fit in some tiny box.

Back: Electronics

Front: Monitor

The Sprog / Raspberry Pi plan won. The Raspberry Pi is a full Linux computer; with a keyboard, mouse, and monitor, it could do anything a big computer could (though a bit slower). I wandered over to Fry’s to pick up the Pi ($50). At the same time, I picked up a cheap flat panel monitor with an HDMI plug; the cheapest, at $99, was a 22 inch display - bigger than I expected. It turns out smaller monitors are speciality items now. That monitor set a minimum size for the electronics box.

With the monitor and electronics, I made a plywood box to hold them all in - the monitor set just inside, facing one way, and electronics the other. The box also contains the power strip, extension cables to reach the modules, and power for switch machines - all nice and compact, and easy to transport. Once it’s plugged in, I can use the monitor to start up JMRI and the command station; with some quick clicks on my phone, I’m running a train. All I need to do now is get some additional plywood to make lids for both halves of the box, and I’ll be all set.

Interesting aside: my nephew came over a couple weeks back and wanted to see my trains. The big layout was a mess, so I laid the Market Street modules on the floor, plugged things in, and we ran trains. When we went back inside, he beamed. "Mommy, we ran trains with a phone!"

Now, that 22 inch monitor is too large to use just for starting up JMRI; check out our next episode to hear what else I can do on that screen.