3D printing for a model train layout

I watched the 3D printing technology progress from afar with great anticipation. In August 2020 I decided to take the plunge and see what I could do with the technology.

I decided to record my experiences as well as detail what the process involves in making something that is usable on the layout.

sample items printed with 3D printer

The type of printer that I think is most suited is called Masked Stereolithography (MSLA) 3D Printing. A photosensitive vat (with a transparent base) containing UV sensitive resin, is placed over a monochrome LCD that acts as a mask. This mask lets UV light from below through in some places and not in others, allowing a layer of resin to be exposed with UV light from below, rather like a B/W film negative. This allows an entire (thin) layer to be printed at a time. Once exposed, the machine moves the part being printed up, and then the next layer is printed.

This approach allows for a good resolution needed for fine detail, but it is rather messy as one has to work with liquid resin, and strong alcohol for cleanup.

The machine

(If you are not interested in the gory details in getting a working printer, skip to the TL;DR at the end of this section.)

On 2020-08-23, I ordered a printer called a PHOTON from a company called ANYCUBIC. Their website said that delivery would take a while due to COVID shipping issues. I also ordered their curing and washing machine and some resin. It arrived on 2020-09-30.

The printer is a box-like machine that can be placed on a desk. It has a hinged door that provides access to the vat that is held down onto the LCD mask screen, and a vertical mechanism that can accurately move a build plate upwards as the item being printed is created at the bottom of the vat. There is an on/off switch and a small touch screen control panel on the front. There is a USB port that is used to bring the computer file that defines the 3D object to the machine.

Photon vat seen from above

During October I tried to get things to work, and I followed all the (very poorly translated) instructions carefully. The main problem I was having was getting the first layer to stick to the build plate that lifts it out of the resin vat. I eventually contacted their technical support, asking if perhaps there was a protective layer that should perhaps be removed, or what else I could be doing wrong.

The responses from their technical support was absolutely terrible. Awful awful awful.

All they could talk about was the 'levelling' of the build plate (one has to align it carefully to ensure that it is parallel to the LCD screen), something I had done repeatedly and had also watched YouTube videos on how it is to be done.

Not once did they ask about:

  • Exposure times.
  • The removal of a protective film over the LCD screen that had to be removed (despite me specifically asking about that).
  • Warming of the resin

I eventually decided to not use their default software settings and increase the exposure times. I started heating the resin in hot water, I also warmed the build plate and vat, and I started to get some successful builds.

Then the USB port starting giving me trouble. Part way through a print job, the USB port reading the files off a USB stick would fail and the printer would stop. Back to tech support. I provided details of the symptoms etc., including proof that the USB stick was not faulty (happened with multiple USB sticks), I had reformatted them, etc.

Their response was a set of instructions telling me to dismantle the USB plug connection from the case, and plug the USB stick into the now loose cable. I was also to unscrew the DC power port and wrap insulation tape around the threads before screwing it back into the (metal) housing!

I thought, OK, people who do 3D printing are typically the 'engineer' type of person, so some modifications not typical of an end-user type of product may be justified. I started doing what they said. I found out very fast that it is in fact, impossible to unscrew the female DC power port because the wires inside would get twisted past the point of destruction. Furthermore, if I did somehow manage to do so, insulating the threads and then screwing it back into the metal case would defeat the insulation.

I was able to separate the USB socket from the case and tried printing like that with no improvement.

The correspondence with tech support went from the sublime to the ridiculous with suggestions that made absolutely no sense at all. They decided to send me a new internal USB cable.

I also provided a video of the problem happening and then showing that if I rebooted the printer without touching the USB stick that it could read the USB stick fine, proving that it was not the USB stick, nor the connection, nor the internal cable between the port and the motherboard.

I then looked deeper into the inside of the machine and saw that I could plug a USB stick directly into the motherboard if I disconnected their cable that ran from the motherboard to the housing. I tried that and got the same result, proving without any doubt that the problem was not a result of the internal cable, as it had now been excluded from the system completely.

They then started again with the suggestion to unscrew the DC power port... it was time to give up on them. Via their Twitter account I got some traction and I said I needed to send it back, which I did.

I then ordered a Phrozen Sonic Mini and started again. A very similarly designed machine, this one's lid lifts off, and the vat has vertical walls, which reduces the capacity of resin. The pouring 'spout' in the corner is very small, making it harder to clean. The top of the build plate is also horizontal which means that the resin does not drain off it and into the vat as well as the ANYCUBIC Photon.

What is much better about the Phrozen machine is that they have used a much brighter set of UV LEDs under the LCD mask which means that the exposure times can be much shorter (2 seconds per layer) which makes the overall printing speed much better.

Overall the Phrozen Sonic mini works much better than the ANYCUBIC.


The ANYCUBIC Photon printer failed and their tech support is worse than useless. I then bought a Phrozen Sonic Mini which works.

The process

So now that I have a working printer, what is the overall process of developing an object?


First, one needs to know the exact shape and size of what one wants to print. That entails doing some research and finding the approximate dimensions of the item. One of the items I have produced are replicas of the concrete distance markers that were used in Germany since the 1940s, and can still be found along most lines. I was able to find a 1939 document that defined the exact dimensions as well as exactly where, and how, they were to be positioned along the tracks.

This can actually be an interesting and educational exercise. I was impressed with the level of detail and thought that had gone into the design of these distance markers.


Once one knows what the goal is, one has to use 3D CAD software to produce a three dimensional drawing of the object. I initially used an amazing browser based system called TinkerCAD that allows one to create complex 3D objects. One can either work interactively combining and subtracting various shapes to produce what what is needed, or, use a drag and drop programming system to define the placement of such shapes, in a script like fashion. There is no way to make backups of the code generated though, and my designs were getting a bit too complex to be able to track changes. Drag and drop accidents could also result in fragments of code ending up in the wrong place, etc.

So I looked around for something better and I found an open source program called OpenSCAD which is excellent. It has a functional programming language that one can use to define the shapes needed, and there are also very capable libraries that are available. Best of all, the source code is stored as text files so I can use normal coding practices when editing, and source control software to track my changes, and make backups as needed.

The software draws a quick rendition of the object which can then be viewed from any direction.

Once one is satisfied with the design, one then initiates a true rendering that is more accurate and the design can then be exported as an .stl file. Complex designs can take up to an hour to do the final rendering on my PC.


So now that one has an .stl file, there is still another step before we can start printing. We have to take the 3D model and slice it into layers so that the printer knows what to print at every layer. As one may expect, there is a whole range of slicing software packages out there too. I am using CHITUBOX which supports my printer.

Here is an illustration of a distance marker showing the shape of the layer 512 of 565

Sadly, it is not as simple as giving an .stl file and getting a printer ready file out, because the shape of the item to be printed needs to be carefully considered given that it is printed upside down, suspended from above. This means that one cannot simply orient the model in any position, one has to position it such that every part of it has 'support' from below at every stage of the print process. For cases where it is impossible to find an orientation that satisfies this requirement, the slicing software has the ability to add supports that will be added to the model, and these then have to be removed afterwards. This adds additional work, and also has the obvious disadvantage of leaving nasty marks where supports were removed.


With the sliced file copied to a USB drive, one now approaches the actual printing phase!

In order to maintain a good temperature for the printing, I have placed the printer on a shelf in a closet that can be closed with a cloth curtain. Inside this closet I have also placed a 100W electric foot warmer. I place the build plate and vat on this warmer and I place the bottle of resin in a tub of hot water to get it nice and warm. I shake the resin vigorously periodically too.

I check that there are no dust particles inside the vat and the surface of the LCD mask is spotless. (Any dirt would impede the light from below and not print anything above it.)

I insert the USB stick with the file to be printed and get it selected and ready to print. When the items are warm enough I attach the vat to the printer and pour in some warm resin. I attach the build plate and let the printer start. I replace the cover to prevent any dust from getting in and to prevent external UV light from reaching the resin.

The printer starts and then provides an estimated duration for how long the print will take. I don't know why its estimate is not accurate. It should be able to read the file and calculate the duration to within a second! It takes about an hour to print each cm of height.

I set a timer to come back and see how it went.

After the print

Once the printer is finished, it lifts the build plate up to the top and the item should be stuck on the underside of the build plate, dripping with resin from the vat.

If  not all of it is there, it means that the first layers did not adhere to the build plate, and there are now blobs of resin stuck on the inside of the vat. The resin must now be returned to the bottle via a filter. The stuck bits of resin must be removed and everything warmed up again. Usually, if the item did not adhere to the build plate properly it is because the exposure time was too short, or the shape was too thin, or the temperature was too low. The exposure time of the bottom layers, and the number of layers designated as 'bottom', are parameters that can be set in the slicing software, so if I suspect that is the problem I try different parameters and try again. If the item was designed too thinly, it is back to the design step, or one may be able add supports.

If things went well, the desired item has to be removed from the build plate. Since things now get messy, I put on gloves and a mask. I remove the build plate and drain excess resin into the vat. I use some spatulas to break the item off the aluminum plate. Sometimes this breaks the model. Once off, I wipe off the excess resin and then place the item into 99.9% alcohol to remove the rest of the uncured resin. While that sits in the alcohol, I clean up the build plate and vat, or queue up the next file to be printed and get that going.

Many sheets of paper towel are harmed during this process.

After a few minutes of soaking in alcohol I pour the alcohol out into another bottle so I can retrieve the part. I dry it on some clean paper towel and then place it into the curing machine, close its lid and let it cure the outer layers of the resin using UV LED lights, for about 6 minutes.

When that is done, one can finally inspect the part carefully and see if it matches what one hoped for!


This all means that the steps of designing and printing a part is a long iterative processes that is at times very frustrating as there are so many variables. 

Despite the big learning curve and iterative nature, I am glad to have started it. I have managed to produce some items (and have more in development) that add nice elements of realism to the layout.