Welcome on Henk's website

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


Producing parts with resin

New techniques are generally advocated with promises about future properties that may turn out to take a lifetime to realise, nuclear fusion is one of those for instance. However in the case of 3D printing the waiting was only 10 years, as with the introduction of the Anycubic Photon, a 3D printer with acceptable quality levels for small scale modelling is now available at consumer price levels.
The big question is however still not answered of course, because even when every household has its own printer than you still need its input!


Dissappointing 3D printed results around 2009/2010 were reason to choose for CNC milling as the best available technique for high quality results. Either I got good visual results but in a plastic too weak for anything or horribly stepped results good for nothing. Of course CNC milling will still hold on as a more versatile technique, because it is material independent, but the current generation home resin printers are now capable of results that can beat it in speed as well as quality for complex and curved parts. There is of course still a limit in dimensions, but for modelling in N it will suffice for most parts.
The Anycubic Photon is a 3D printer using a 2K LCD display to polymerise an UV sensitive resin. The resolution in layer depth is from 0.01 mm upwards, this is fine enough to avoid the stepped surfaces that 3D printed products normally are afflicted with. Of course outsouring using 3D printing services could reduce this stepping effect by use of higher quality and thus more expensive printers, but this is a very expensive way of modelling and fraught with uncertainties. This because you do not control the process itself on how your items are stacked on the printing platform, which is generally an all important factor in the quality of the outcome, but you will be charged full price even if the outcome is good for nothing.
Apart from full control another reason to own such a printer is to enhance the workshop capabilities, for objects with many curved surfaces 3D printing is a faster and sometimes the only production process.
Because Deskproto takes .STL files as input for 3D CNC milling quite some files exist in my collection of drawings that can be directly fed to the printer software to see what comes out.

First results

Here we see the first results of an Y25 bogie for a new wagon series. It is consisting of an inner part with the brakes and the keepers for tophat bearings and cosmetic claddings. Doing it this way certifies that the holes for the tophats can be reamed to size and will fit, this however comes at a cost of overal thickness. Further experiments with digitally optimised versions are therefor likely. The cosmetic parts got a first iteration as the strip in the center fell away being too thin. cosmetic sides
Results of the adapted drawing, this time done in Monocure grey resin which seems a bit softer and less crisp than the green Anycubic resin when it comes out the printer. After UV curing the properties get better but I think I prefer the green one. No idea why the openings are closed, in the STL file they are clearly open.

various parts
Various parts such a NMBS ribbed buffers, a running character and set of side frames for a tender. For the latter one these results show the leaf detail in the springs which is not available in the CNC milled version because being much finer than the cutter. Thus this would allow to model those details that are currently left out such as a series of rivets. characters
Direct comparison of the character with a ModelU driver from the 2mm SA shop. There is plenty of room for optimisation in support structure but the figures compare well. CNC milling such items requires more axles than my machine has, thus here the printer wins hands down.

The boiler of my type 29 which comes out quite well. The chimney has a part missing but this is clearly traceable to a software error in generating the slices. This item was printed with 0.05 mm layers and 11s lighting and took around 3 hours printing time. The boiler is perfect, only on the back of the firebox some layering can be discerned. But that is of course nothing compared to the result from the filament printer. The angle was chosen to see how the support ladder performs, steeper angles will do without, this is a control parameter in the software. firebox

A set of seats for the interior of my 603 autorail complete with passengers. In comparison with that from the filament printer these results speak for themselves .

A set of wheels but here the grey Mococure resin clearly looses out on the green resin of which at least the single printed wheels are far better, further playing with the parameters might improve these results. wheel smokebox door
A single wheel and a smokebox door which is perfect without any stepping. Compared to the CNC version only the hand grip is missing because being too thin. wheels
Direct comparison between a 3D printed wheel and CNC milled wheel in POM (Delrin). These are made with the same .STL file as input. The CNC version clearly has thinner spokes, but these look a bit too skinny but the holes are clearly better defined whereas the 3D one looks a bit fuzzy on the edges but has its spokes more clearly tapered. Both look to be usable as long as they are not mixed. Playing with the parameters such as milling speed, path definition and cutter diameter vs layer thickness and time can certainly produce items that lay closer together. The 3D resin looks to bond better to steel rims than the POM which would be of great advantage. Further the production time for an entire set for a loco will be smaller as the printer can do more than one set in one session where CNC milling needs the same time for each wheel.


The first experiments show that this printer certainly has great potential in fiNe-scale modelling and therefore earns its place in the studio.

copyright: Henk Oversloot
date: 3 Dec 2018