Nothing like eating your words... one day later.

This weekend, the local model railroad design crowd hosted the "Layout Design and Operations Meet" for something like the tenth year running. They even managed to sell out the usual venue at the Santa Clara Depot; the 130 attendees forced them to move it to a local hotel to have enough room. Like other years, Saturday was filled with talks and presentations, and on Sunday, attendees were invited to participate in operations on local model railroads.

I got my chance to visit Ed Merrin's Northwestern Pacific layout. This was a big deal for me; his last layout really inspired me when I saw it in 2000. I loved its small size (a bit less than a two car garage), great implementation of a double-deck layout, and lots of places for switching. Ed's current layout has a similar theme and shape; it's a bit larger than my layout, but it's again got a nice mix of heavy switching and nice rural scenes.

When I list the things I did well on my model railroad, I always point out that I chose not to have an explicit yard on the layout, but instead filled the space with industries that would make life fun for the switching crews. Yards are fun because they provide a full operating session's work for the yardmaster switching boxcars, because they're focal points as crews start and stop trains in the yard, and because they're good for showing off engines and freight cars.

On the bad side, yards take at least the space of one or two towns. In my specific case, no model yard would ever look like the huge mainline yards of San Jose and Santa Clara, so I didn't think they were worth building. Instead, the trains start in staging, come "on-stage" to do their switching, take their bows, then go back off to staging. It's seemed like a great solution for small layouts.

Ed's layout is the opposite. His three major towns - Petaluma, Santa Rosa, and Willits - were reproduced from the original track plans, and each one supports a small but decent yard. Each has a yardmaster who's responsible for sorting cars, keeping the town under control, and working with the trains coming through. Scattered through each town are several smaller industries for switching. In some ways, his layout is the opposite of mine; rather than build large industries, the yards really are the large industries, handling lots of freight cars as they get changed between trains or sent off to local industries or nearby towns, with the nearby industries helping give the yard a reason to live. Sometimes, those industries are off-layout as cars get sent "off-stage" towards their ultimate destinations - Los Angeles, Oakland, New York. Because each car gets handled by one or more yards, and because each hand-off requires the yardmaster and engineer working together, the process requires the crews to work together, and keeps everyone quite busy and having fun with the operations game. [Photo: John Sing and Rick Lopez manage the yards at Willits and Petaluma.]

So, yards aren't as bad as I thought. Yards can be small enough to be realistic, and can generate enough traffic to be an interesting part of a layout. I'm not sure I'd change my current layout to add a yard, but I'll be more cautious about declaring "no yards" on my next layout!

Could you live without a yard on your model railroad?

What Went Right... and Not So Right?

Post-mortems are absolutely great. (Unless you're the one getting post-morted, of course.)

We're human - we learn by making mistakes, and sometimes we learn by hearing other people make mistakes. We'll share stories about the nightmare contractor we shouldn't have hired, or the great way to spend $400 with a $7 quick fix for our car. Hearing straightforward discussions of what well and went badly in other people's work helps us avoid those mistakes.

I used to really love getting Game Developer magazine because every issue would have a post-mortem from a video game design team. (See here for all the post-mortems they've published.)They'd talk about the parts of creating the game that went well, and the parts that went badly. The general topics didn't change very often. This team didn't do enough testing before launch. That team had misunderstandings with the publisher explode into contract disputes at the wrong time. Another team underestimated the time for a critical piece, or didn't put enough effort into AI algorithms to make the game's behavior interesting. The articles always told a lot about what the real software development process was like, and realistically explained what could go wrong, and how the team recovered (because most of the projects were, after all, the successful ones.)

I've gone through the same sort of reflecting on my model railroad, and I used the Game Developer "What Went Right / What Went Wrong" structure to put together a presentation for the annual Santa Clara layout designs and operations meet this weekend. (Slides are here if you're curious what was on my list.) I was a bit surprised at the positive response; most of my stories concerned problems that we know can be issues for model railroad - deck height for multiple levels, reliability, space, and staging - but folks still enjoyed the talk. The talk felt best when I was telling my stories of when things had gone horribly wrong for me, or the switch work that I ended up redoing several times, or the fun of watching people spend an enjoyable 45 minutes switching one industry.

Two particular points seemed to resonate with folks. I'd been really unsure of how to start the model railroad five years ago, but at some point, I realized that this wasn't going to be the last layout I ever built, and it didn't have to be perfect. I promised myself this would be a "five year" layout - I'd assume it would have a lifetime of five years, and I might tear it down after that. (I've since heard the term "chainsaw layout" to refer to a model railroad that's being built with the knowledge that it'll have problems, and it'll get torn apart really quickly.) The "five year" declaration got me confident enough to start this layout, and I'm happy I did so five years ago. Multiple folks remarked that the single line: "This is not my last layout." was the message they most remembered.

The other big issue came from a friend after the presentation. I'd wondered whether I'd actually said anything that memorable, and he remarked, "Come on - people don't really want to hear a presentation where you just say you're a genius and everything worked perfectly. They want to hear that you built something, that you made mistakes, and that you either fixed the mistakes or that they weren't a big deal in the end. They want to know that they might succeed if they start."

So what's on your list of the things that went well, and the things that went... less well? Or what's keeping you from building that next great layout?

What Makes a Town "Not Work"?

It's short, shameful confession time again. I play favorites on my layout; there's specific towns I like that get more attention, and there's other towns that... don't get that attention. I just don't like all my towns.

Does that make me a bad layout owner?

Campbell's always been one of my favorite towns. It's small - maybe 8 feet long from end to end. It's kind of dark because the upper deck is too close, but it's spread out, with space behind the tracks for scenery, and a large shelf in front of town where the Campbell station will go. It's also got a fair amount of switching--two canneries on the far side of the track that require cars to be spotted at specific doors, and a couple of smaller sidings on the near side of the tracks to keep trains out of the way. The switching at the canneries isn't particularly complex, but there's always some work to be done here. There's a big passing siding, so trains usually keep it in mind as a place to get out of the way of other trains. It's also got a main street that's three storefronts long, and I've been having fun building each storefront on each side of the street: the real estate office, the Spanish Revival bakery made from a milled Plexiglas face, the weird "Cricut scrapbook cutter" building, a grocery built from a cheap plastic kit, and the eventual home of the Campbell movie theater in the old bank.

Not the most exciting of towns, but it's an interesting setting. More importantly, it's felt like a success story. I've been able to build interesting buildings, reproduce prototype tracks, build a realistic scene, and do all that while making it an active and important place on the layout. That's what makes Campbell work.

On the other side of the aisle is beautiful Los Gatos. It's again got a passing siding, it's got a cannery, it has space for some interesting buildings such as the side view of the Bank of America building. It's a much more constrained town though, with open staging in front of it, and a backdrop right behind it hiding the helix and the ramp up to Alma. On the positive side, it's a destination - trains from San Jose turn around here, flipping the engine around on the nearby wye at Vasona Junction. It's also a necessary passing siding so trains can wait for others coming down the hill.

Los Gatos gets none of the love, though. It's unsceniced, on a narrow shelf with few options for scenery. And it's been that way ever since I first thought of the layout. Los Gatos was always sort of a compromise, a town crammed into too tight a space, added because I needed a destination but didn't have room to do it justice.

I suppose the idea of "liking" and "not liking" different projects in a hobby is universal. A painter might like how this painting turned out but not that one. A gardener may prefer some area of the garden. We're humans, we have opinions, we'll have better or worse experiences.

But is there any way to change that? Was Los Gatos an unsatisfying town from the first time I drew it on a plan, or did it get that way because I wasn't sure what to do with it? Is it always going to be unsatisfying, or is there something I can do to make it less so?

If it's really an unsuitable location, then maybe the answer is there's no h0pe, and I should put my effort elsewhere. I can leave it as an afterthought, and half-heartedly finish the details I planned. I could also decide that there's something wrong with Los Gatos as planned, and try some changes to the design that'll make it more appealing.

If there's just something wrong with my the design, maybe I should I get rid of the town completely, and just add a country scene extending from Vasona Junction? Or are there industrial areas in Los Gatos - the oil distributor at Farley Ave. or businesses along University Ave. - that would be more interesting? Do I keep the passing siding, or sacrifice it in order to make a more attractive scene?

Maybe I just need to rethink it and get positive - try building some great building flats that will get me inspired, or try some alternatives for scenery?

So do you have any towns on your model railroad layout that are the ugly ducklings? How have you gotten a town out of its unloved state?

Ah, so that's how the pros build architectural models!

Here's a nice video of how Sweet Onion Creations uses a 3d printer for the interior of a house model. They still fell back to basswood, wood, and metal for the final structure. They were making a physical model from the winner of a SketchUp house design contest.

http://sketchupdate.blogspot.com/2009/12/making-of-sweet-physical-model.html

Lots of the model making looked awfully similar to building structures for my model railroad, except they've got much nicer tools. Printing the entire interior in one fell swoop was mighty impressive!

Progress on the Drive-In

Here's some more shots of the 1920's Drive-In Market (or what we'd call a strip mall) in progress. The walls have all been glued together, but I still need to glue on the main roof, and paint it to simulate a tar and gravel roof.

Overall, I'm more impressed than I expected. The model isn't the best model I've ever done; it's pretty easy to see the places where the diagonal extruding pattern is visible, and some of the detail's a bit rough. However, I'm really pleased how easy it was to make identical parts, and the roughest piece - the storefronts that I left pretty much as they appeared from the printer - look pretty decent from about three feet away. (See my Makerbot blog for a detailed view of the storefront and some of the tricks I used to get them more acceptable.)

Even more importantly, it was easy to steal the design of the roofline from the arched walls, and reuse it in the tower's roofline. It would be pretty easy to follow the Spanish Revival theme with other wall patterns, just like Design Preservation Miniatures wall sections let you do for brick buildings. I was even thinking of replacing the end walls with a solid wall section matching the arched wall. Luckily, I decided (1) I wanted to finish this model, and (2) the Drive-In Markets were probably inexpensive enough that the real builder would have left the walls plain, too.

3D Printer - Better for the Inside of Models, not the Outside?

I'd mentioned last time that the Makerbot did an admirable job of making the pyramidal roof for the Drive-In tower trivial. Rather than figuring out angles, cutting forms, then carefully piecing on the outer skin, I just drew the shape I wanted, printed it, and was done. A bit of spackle to smooth things out and a coat of tar-like paint made it look just fine.

But a simple pyramid's easy. What about something more complicated - like a pyramidal roof with an octagon in the middle supporting a tower? After all, I'd had to do that to build my model of the Rio Grande gas station near Santa Barbara.

So I dashed over to SketchUp, drew a version of the tower in about an hour (mostly trying to figure out how to build the dome), printed it, and got the result seen to the right. The dome got a bit squished because the plastic couldn't cool before the next layer was applied, but it's not too bad. The tower's still a bit rough and not-to-scale, but I could scratchbuild that all out of bits of plastic. However, getting the roof lines and the octagon base is much more trouble. For the model on the left, I think I used a hunk of balsa wood for the octagon, and fit the roof pieces and underlying forms during a few hours of further work. I'd much have preferred to print out the rough roof shape and octagon, finish both with a file, then apply the roof tiles on top of that base.

I've heard about people building structures this way in the past. They would make a substructure from a chunk of wood using a table saw, then glue the walls and roof onto that chunk. Here, I'm just making that process easier, and I don't have to try to keep my fingers away from the table saw's blade as I set up awkward cuts on a small piece of wood.

Printing a 1920's Drive-In Market

I've been doing less work on the layout for a bunch of reasons, but one of the larger offenders is a new tool, a Makerbot 3d printer. This squirts out threads of styrene plastic to build up three-dimensional objects.

I've been bitching lots about the lack of Spanish Revival-style buildings in HO, and fabricating the curves of these models is a pain to do by hand. I've also been eager to build a 1920's drive-in market, and Spanish style was insanely popular during the '20's and '30's, so this seemed like a great project to try. My best luck in the past with such buildings was the Campbell storefront. In that case, I used a metal milling machine to cut a Plexiglas sheet for the facade - not the easiest of jobs, and a lot of trouble to remake.

This model (still, obviously, in the process of being built) is a combination of sheet styrene and printed parts. The main facade is made of four identical arches, spliced together. The storefront doors and windows are separate printed pieces. The back walls were made with sheet styrene, and I'm using door castings from my scrapbox for the back doors for the stores. The central tower is an interesting mix; I knew printing such a large object would be slow, so I made the body of the tower out of 1/16" sheet styrene, but printed the top decorative piece. Note how I had to make a tab that glues to the back of the styrene to get a good joint; edge-gluing printed pieces doesn't work well, probably because the pieces are hollow and somewhat rough.

What I've learned about 3d printing from this project:

* The Makerbot's great for making duplicate parts. It's perfect for the main building arches, and it was really easy to borrow the middle section of each facade for the tower's false front. I mentioned on my printing blog about the trouble I had keeping the arches from warping; the smaller pieces for the tower were much faster to print and less likely to warp.

Like the Cricut cutter, the 3d printer is a bit less interesting if you're only making one object, for it might be easier just to make the piece by hand. But see my note about making roofs before assuming a 3d printer only helps when making lots of identical pieces.

* The Makerbot's reasonably good at doing curved edges, but it's not great at doing finished surfaces. The extruded plastic is 0.13mm diameter (or about 1" HO), and the gaps between the extruded noodles makes the surface stripey. With the building facades, I covered the face with spackle, gesso, or acrylic medium to try to smooth out the edges.

The storefront door and windows shows what things look like without filling. These are printed pieces straight from the printer, and were only spray-painted. I wasn't sure if these would be acceptable, so I printed them out quickly to see how they'd look. They turned out better than I hope, but they're still pretty coarse, and not quite good enough for a model. These were all printed with a 45 degree angle fill; I'm hoping to print some storefronts with a vertical or horizontal fill pattern in hopes that'll look more like clapboard or another typical material.

* My big discovery today is how great the 3d printer can be for parts that are difficult to fabricate. For the tower, I wanted to have a low, tar-and-gravel roof behind the false fronts. I knew that doing a pyramid hipped roof (where all four sections of the roof meet at a single point) can be a pain to build. I've done them by building up forms for the roof line and carefully cutting and fitting each roof piece, but it's tedious work.

Instead, I just went to Sketchup, drew out a 14 foot square base, found the center of the base, drew a vertical line 3 feet high, then drew lines out to each corner. I then printed the object, and got my perfectly-sized pyramid. I'll glue some simulation of tarpaper on top, and no one will ever know how easy it was to build the roof.

More photos as I finish assembling and painting the model...