Welcome on Henk's website

find all about fiNe-scale modelling
in a scale of 1 : 160


Using the 3D printer to produce coaches

The consumer range of 3D printers is continuously getting better and delivering more accurate results, thus we can use these for our more serious projects.

On the wish list there are some coaches that form a perfect object for testing the latest generation 3D printer.

It turns out not to be a case of press the button, wait a few hours and there is your coach. There is a lot to learn before leaving the spoiled print domain behind you.

Drawing up the coach in 3D

Generally a coach can be regarded as a box with a curved roof. The coach side normally has windows and doors and it may have riveted strips or panels on its side. The coaches of my interest here are the M1's of the NMBS dating from mid thirties. These were the second range of full metal coach and a bit longer than the earlier type L coach. There are drawings available from H. de Bleser, these drawings are fine for the general layout, but you need more detail when you want to model one. Thus visit your preservation railway with ruler and pencil and find the particular coach and check the dimensions against the real thing. Thus measure width and height of windows and distance between, also doors and the various strips and distance between strips and rivets. These details are necessary to produce your own detailed drawing. A set of photo's is of course helpful for the general layout and details on things like bogies, battery boxes etc.

Original M1 bagage compartment
An example of an M1 with bagage compartment at the far end, CF3V in 2006, Marienbourg

class 2/3
A better preserved example in two tone livery of class 2/3 showing its rivets, TSP in 2009, St. Ghislain

Personally I find it easiest to start with a 2D drawing, a simple line drawing which gives a side view with windows and doors. This can be done in most CAD packages by switching to 2D mode. In my case the drawings were done in 2006, long ago, in Corel Draw, after checking these again to my measurements they were exported as .ai (Adobe illustrator). My experience is that this format is much better than .dxf or .drw. This 2D drawing is imported into my 3D package, Cinema 4D. Being a long user of Mac's the use of this package dates from the time that there was no other choice on the Mac in error free 3D packages. For this subject I used C4Dv21. There used to be a cheap actual version limited in options, but alas this is a thing of the past, every decent package nowadays seems only available by subscription. Viacad as oneoff option is considerably cheaper, but V12 is not error free and there seem no updates available. For hobby use you may use Fusion 360 'free' and there is Blender as open software both are of course only available in exchange for your data, thus actually not free at all.

Learning a new package generally takes a lot of time, count on at least half a year (this of course depends on how much time you spend on your hobby). Thus I tend to stick to what I know, even if it is not the best choice anno 2022. Thus if you start out new, then first do your homework on what to learn. As incentive, having practical knowledge in 3D designing may help you further in your job career, it is never a waste of time to learn new things. I think today, I would probably try Fusion 360 as first option.

In general there are 2 approaches for drawing up a coach. For a flat wall a sideview with openings can be extruded to wall thickness. For a coach with curved sides like the type L the approach is the same as for a roof. For curved sides and roof you export half a cross section of the wall, make a split at the roof if you want a seperate roof. Extrude this cross section lengthwise along the length required and than cut out windows and doors with a sideview of the openings. In both approaches you have to decide early on the wall thickness to use.

explaining extrusion types
An attempt in 2D explaining the extrusion function using a 2D profile to get a 3D object such as a wall or roof.

Thus there we are having an imported 2D side view in our 3D package. The first thing to do is to reset the coordinates of this to some comfortable point in space. Thus nullify the existing object coordinates and set its zero position to say left hand center line of the wagon floor. After this all coordinate values for items will be positive and symmetric along the center line. This helps as you only have to toggle +/- into -/+ when copying parts from one side to the other.

Sometimes coaches are symmetrical and they even may be symmetrical in 2 planes. In the latter case you only have to draw up a quarter of the coach and only clad these with non-symmetrical details, in this case it makes sense to set zero in the center of the coach.
In my case the first chosen M1 coach was a bagage version and not 100% symmetric due to a smaller toilet window at one side. In such case use a drawing of each side. The most upper photo shows a bagage version having 2 toilets and a second less high window at that compartment, thus apparently there are more versions then I was aware of when I started this project.

Wall thickness and window cut-outs

The wall thickness to use is mostly dependent on the behaviour of plastic/resin over time, not on what your printer can print as thinnest wall unless thicker of course. My experience with wagons and coaches milled from polystyrene show that 0.5 mm or thinner will seriously deform if left unattended over time. With 3D resins having high shrinkage and higher regasification of volatile compenents this can only be worse. On the other hand, certainly for flush windows, a thin wall thickness helps to produce more realistic windows as the window thickness wiil be observed on close inspection. In my case I choose 0.75 mm for wall thickness, the windows itselves lie deep at least 0.2 mm from surface. This gives a 3D component with holes for window and doors that can be positioned at the correct half coach width. You can then do same for the door sections and the halves of front and back wall of the coach and position these on their specific coordinates. The generated half coach side looks like this:

A first view in developing a coach side
A view during drawing up a coach side with all the main parts in position

In the case for the M1 the ends of the coach have inward angled sections with a step in height for the buffer beams. The angle is 5 degrees. I exported these 2 parts as seperate elements, which after extrusion to 0.75 mm can be set to the correct angle. The next step was producing the center part of the roof, thus that part between the angled sections. This part was based of a cross section of the coach roof with the same wall thickness and extruded to required length.

This leaves generating the double curved roof ends. This part is based on the application of a loft function. Taking the outer curve of the roof as inward line and the curve of the top of the end wall as outer line. In order to get a smooth transition I drew up 3 additional Bezier curves in between. The width of these is set to the angled ends. The actual curve is copied from that of the roof and re-positioned further outward and edited for lower height and width. This generated loft surface then gets copied and one of them is extruded by -0.75 mm as wall thickness. On this copy all the normals are inverted, thus you need to invert these such that they will all point outward and then you can connect these 2 lofts into single item. The circumference is now closed with the close gaps function. The individual points can be edited to required position in horizontal plane. This leads to a smooth curved roof end, that is placed into a symmetry function. The other end is of course also a symmetrical item. In practice drawing these ends was a feedback procedure as in the first print the roof ends turned out to be too thin for printing, leaving gaps at the corners due to lifting.

When you have drawn up half a coach as a collection of single objects, then you best group these objects into functional groups with recognisable names (roof, sidewall, front etc.) to keep track in your drawing and group these into a half coach and put this into a symmmetry function generating your first impression of an entire coach. After reaching this stage you can start with detailing side and roof with various strips with rivets, all spaced appropiately to respective measurements. Rivets are added by introducing a hemisphere of the right dimensions, I choose 0.05 mm radius, toggle the hemisphere option and duplicate to required numbers and length. In larger scales you may flatten the hemispere somewhat as rivets are not half spheres but a bit lower. Actually I didn't expect these rivets to be visible at all, but that turned out to be an underestimation.

detailed bagage coach
view of coach in 3D

Bogies and frame

The pilot coaches got a 3D printed bogie. Actually it is the same design, the common Pennsylvania (Swann neck) type, as that used for the type L coaches, but instead of milling these were now printed, which allows the brakes etc. to be integrated. The design is for 14mm axle length, this is long enough to prevent wheels dropping out and allows fiNe-scaled N wheels to be fitted.

view of the bogie in 3D

Studying photos is necessary, because contrary to the Y25 Bogies for the Fal wagons, the coil springs turn out to be symmetrical. Thus left and right handed wound coil springs, you can only detect that on photos of the real thing! But also here there are in reality many versions, the Bleser states at least 14 versions. My photos show several options for the axle boxes and more or less rounded swann necks and various numbers of oblong holes in lower carrier to make them lighter. I have also a photo where the 4 cross springs are replaced by a set of 2 deeper lying coil springs
The simplest form of coach frame is a piece of flat metal, I used 1mm brass to give some weight for the pilot coaches. This is just a plain strip with some cut-outs for the door openings and 2 knobs as rotation point for the bogie. But there is very little clearance for the wheel flanges, these will need a cut-out. They may be replaced by a 3D printed version which allows to integrate the battery boxes, the rest of the steps and to add some interior with passengers.

frame in 3D
A design for a frame in 3D with the exterior detail such as the battery boxes and interior with passengers for the mixed class 2/3.

This frame was printed sideways with lot of supports for the tops of the seats and passengers. A print run of sitting passengers seperately to be installed during construction will probably give more freedom.
The lower footboards were drawn, but excluded from printing, there exists a serious interference between the end of the lower steps and the bogie due to non-scale track radius and wheel dimensions for the 3rd class. Thus I will first print one and then try to find a practical solution for these footboards, also the required minimal weight of the coach is of concern with a printed frame. The time gained to reach to the end result faster is not negligiable.
There exist no easy solution for the footboards for the 3rd class outlay, I actually ended up installing only half a footboard (CNC milled 0.3 mm brass), the other half can only be integrated as part on the 3D print on the inner end of the bogie.

view through coach with passengers
Combination of coach and printed frame.

A view through the coach with passengers with the test print of the frame. This combination looks OK for a glass case model, but the weight of the coach is far below that for serious Fremo use. Thus it is back to the metal frame with seperately printed inserts with seats and maybe the battery boxes


Above you see a quickly hand made full metal frame from 1 mm brass metal, milled cutouts for wheel flanges, soldered on rotation knobs for bogies, brass battery boxes (weight optimised = too deep) and the boards for coach numbering, all flame soldered. Between these 2 number boards there exists space for some additional weight (lead), but I don't expect that will be necessary. Actually the length of these boards turns out to be an error, I just forgot to divide by a factor 2 to correct for the 1:80 drawing I worked from. This may get corrected at later stadium.

Preparing for printing and battling printing errors

The first thing you have to do is find out the printer settings for the particular resin you want to use. That is a seperate process, refer to your printer supplier. It comes as a to do with some test prints using a reference object that shows you how to tune the print settings for optimal performance on basis of the amount of detail on the object depending on the outcome. Actually the only thing I did was reducing the time for lighting the base layers as it turned out to be impossible to clear the platform. The resin used here was Anycubic transparent green with 20 s for the base and 2.5 s for layers with a rest period time of 5 s, layer thickness 0.02 mm on a Phrozen Mini 8K printer.

Of course you will be eager to print your design, but before you try this you have to take a lot of measures to ensure a good outcome. For instance printing objects that don't have a solid base require a lot of support posts. My rule of thumb is that one post per 3 mm is minimal required to get a flat surface and even then if you loose a post for some reason then you have a spoilt object. Thus for objects like wagons and coaches that need a clean lower edge I do design a support sledge in 3D. This needs to be cut off when freeing the object from the rest of post base. This sledge consist of some heavier balks under the sides of the coach, say 3 mm width and 1.5 mm high, these will help in keeping the object straigth and flat when it comes out of the resin bath and the model doesn't have much strength yet. Because you still need to support every 3 mm these get the form of a continuous row of 0.5 mm thick rectangles with a square hole in them. The square hole assists in cutting off the balks and the part above it can be carefully sanded away by using a sanding strip with 180 emery paper.

But there are more measures to take. If you try to print the roof horizontally you will end up with a stepped curved roof due to the stacking of layers. Toggling the option of anti-aliasing may combat this, but that is generally leading to loss of detail in other places. I find the best way for printing a roof correct is by tilting the object some 5-10 degrees, lifting it from the horizonal plane. This makes an entirely different cross section of the roof and with much less surface thus giving also less suction and it will print correctly. Having done that you may think of a first print and probably will find some errors in your drawing, things you forgot or items not positioned correctly etc. There are numerous ways in which Murphy attacks.

Keeping the tops of the windows and door openings flat and straight requires to set posts. The alternative is to fill them out with break-out sections, these are better in keeping the intended curves. These break-out sections have a ridge of only 0.15 mm at the circumference and a mid-section of 0.4 mm. This way they are easily removed after having done their work. A photo of a complete coach ready for export as .stl file looks like the photo below

coach on sledge with break-out sections in windows
coach on sledge with break-out sections in windows

The first printed coach came out quite well, superficially. It showed the rivets to my surprise, although with a bit of help of a filtscriber for more contrast. But after curing it did look entirely different.

deformed roof on curing
Deformation of the roof on curing

The mid section of the roof was entirely deformed as the sides curved inward, maybe due to too long times for UV'ing and probably getting too warm. But this necessitated incorporating some internal struts to prevent bending of the roof. These turned out to lead to new errors that were not there before. Streaking on the outside due to internal structures! I don't know the origin of this error, we may have to do with suction or alternative a firmware error. After this experience UV curing was simply entirely skipped, after cleaning leave the model standing for a day or two in an area with clear daylight, certainly do not put it in direct sunlight. Time helps evaporation and daylight does the curing.

streaking along the roof
Streaking just below the roof edge originating at the cross supports of the roof

Although the model has paint on it, it was scrapped for reasons as the top of the windows and streaking above it, but it can perfectly function as a practice object for painting exercises, help in mixing colour, testing viscosity and cover by layer thickness etc.

lifting due to too small posts
Another error that may turn up is lifting of the ends due either too small or not enough posts.

This was a simple case of not enough attention on setting the posts. But here you can see the posts at the door openings and those at the window tops with break-out sections.

I didn't found a way around the streaking other than seperating between roof and coach. The coach can be printed flat, the roof with a tilt. It is a bit more work to combine the two again but they are a good fit. First thing to do is relieve the roof of its sledge and fit onto the coach and then relieve the coach from its sledge and fix the roof with superglue. After that the coach is rigid again.

Painting and glazing

I did put on primer using a Tamiya spray can, but this was not a real succes. It is actually too coarse and doesn't form a nice and even film. Thus the next batch will certainly be treated different. Experience gained. In hindview there also seems a 'fine' primer, which may be better, you are warned now.

current state with paint and glazing
Current state with paint and glazing

The glazing was done by using the milling machine based on the same drawing of a sidewall with windows. The contour was made 0.05 mm smaller to assist easy fitting. The material is 1 mm clear polystyrene from an old CD box. These were glued with micro crystal clear, a sort transparent drying wood glue, this doesn't fog up the PS glass and it has enough grip for normal handling of the coach.

simmple cutting pattern for the glazing
Pattern used for the glazing of one coach.

current state with paint and glazing
Current state with handrails and buffers, right-click and open image in new tab for inspection in high resolution

The to-do list is now very short with only the gangways still to be printed and a coupling to be attached. I didn't paint the handrails in steel colour as these are hand folded wires and this would draw too much attention showing that they are not all identical, etched ones in NS is may be a way forward here.

I think this result proofs the qualitiy of current day 3D printers. Now we only have to wait and see how these resins keep over time, that is the easy part.

copyright: Henk Oversloot
date: 14 July 2022