M&K Junction Railroad

M&K Junction Railroad
Another train of eastbound coal crosses the Cheat River

Friday, March 2, 2012

It's Itsy-Bitsy

I've finally finished a project that I've been thinking about since 2007, namely, creating a scale B&O Color Position Light (CPL) dwarf signal. What's the big deal, you ask? It's to exact scale and it's itsy-bitsy.

The story starts at the 2007 O Scale National Convention. I attended a clinic by a gentleman who was using 3D Computer Aided Design (CAD) and a jeweler's 3D wax printer to create O Scale parts. He would design a part in the CAD program, send it to the wax printer which would print a copy of the design in wax. The wax model would then be used to make a rubber mold, the mold would make more wax copies which were then used in a conventional lost-wax process to produce the final parts in brass.

While this sounds complicated it was, for its time, an inexpensive and fairly simple way to get to finished parts if you needed more than one copy.

At the time I had been looking into various 3D printing processes concentrating on stereo lithography (SLA), but the costs of those machines were astronomical and having models produced on them was far too expensive, even for one piece; so the wax printer/lost wax process was promising.

Soon after the convention I settled on the B&O dwarf signal as the model that I wanted to produce in O scale. It had never been done to scale; MTH (and, I believe, Right-o-Way) having produced grossly over sized versions for the 3-rail O gauge market. Also, I had the plans for the General Railway Signal Type 'VA' B&O CPL dwarf signal so that would make it easier.

Here's a picture of the prototype.

This dwarf has two additional markers and a sign on top, but the basic signal is the lower eight lights.

The CPL dwarf is a difficult signal to model. Eight lamps, each with visor, on a small target would be difficult and expensive to produce by conventional machining; even to build a single master for reproduction by casting. Many 3D printing processes, like fused deposition modeling (FDM), are too coarse to produce a small, finely detailed model.

In 2009 I took a job where we use 3D printing (out shopped) to check designs before we commit them to production. I had a chance to sample several different 3D printing processes as well as get some exposure to 3D CAD. Now I had access to the tools and some people who could help me out.

Here's what the 3D model of the basic CPL dwarf signal looks like:

This model is somewhat simplified. The nuts and bolts on the case are not reproduced; nor are the ribs on the backside of the case back; nor are the edges of the case rounded. As it turns out, none of these are visible at normal viewing distances.

I still needed to find a supplier that could produce the models without charging $80-100 for each one. Enter Shapeways. I'm not the first modeler to discover Shapeways 3D printing service. Their service is unique; they offer several different 3D printing processes, they offer worldwide shipping, their prices are reasonable and they are modeler-friendly.

I designed the model to use the full capabilities of Shapeway's "Frosted Ultra Detail" (FUD) material which could hold the smallest features out of all of Shapeway's processes. FUD is a 3D printing process called 'PolyJet', which prints a part using a UV-cured plastic shot from a multi-nozzle print head much like your inkjet printer prints. The signal's shell, which is hollow to allow for illumination, is made to the minimum thickness allowed by the process. With some trepidation as to how they would appear, I made the visors a little thicker than the minimum for strength.

Since the model was done on a heavy-duty 3D CAD program (Solid Works) it passed all of Shapeway's tests for printability. After I uploaded the model to Shapeways, I discovered, to my delight, that the model could be produced for well under $10 (US); so I ordered one.

When it arrived, it struck me just how tiny this thing is. It's so small that I am convinced that it could not have been produced to scale, with scale-sized visors, by most other process, especially casting. Here's what the model looks like.

The model is on the dining room table's tablecloth and that's the weave of the fabric that you can see. This gives you some idea of the size of this dwarf. I'm really happy with the way the visors came out. They are 20 thousandths of an inch (less than one inch in O scale) and although that is thicker than the prototype, you cannot tell from looking at it.

Here's a picture on the layout:

That's an O scale E7 in the background. The O scale figure is 1.43" tall (36.43 mm).

I have made some changes to the back cover of the signal, which is a separate piece, and I have another batch on order. When those arrive I'll experiment with painting (Shapeways could not advise me on compatibility of FUD with paints) and illuminating. I'm sure that this signal can be illuminated with 'chip' LEDs; but that will require me to design and lay out a circuit board, which will take a while.


  1. Did you allow for the shrinkage of the brass castings if you still do lost wax to cast these ?
    Ed K.

  2. The 3-D printed parts came out so well, much finer detail than can be done with casting, and the price is right; so I will not be doing these as castings.


    1. That is sort of short sighted. Investment castings provide a finer finish than any 3D printing is able to do by it's very nature a stepped process. Then again you NEVER have any derailments, right? ANY piece of rolling stop with instantly demolish that resin. Investment cast brass or better still Jeweler's copper will survive a head-on from an 0 Scale steam locomotive.

      So do recalibrate and add that shrink factor.

  3. When painting plastic that is to be illuminated I have found it helps to spray a metallic coat such as silver or aluminum first before another color coat is added. This helps to block the bleed-thru of the LEDs.
    Regards, Ben Brown

  4. Nice! I like it. Have you tried miniatronics for the lighting?