Welcome on Henk's website


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

 

An exercise with the modelling tools

Visualising a model of the class 85 diesel shunter and printing it

This article is about building up experience in 3D printing with the Anycubic printer and using a computer on saving time in producing a new model.

Method: Design and 3D print

trial body print
The first trial print of a class 85 Belgian diesel shunter.
trial body print
The first trial print of a class 85 Belgian diesel shunter.

The choice for this model was just because of the dimensions were already available, it dates from 1956 and these machines saw mainly use in Antwerp harbour. Both its overhanging roof on the cab and the struts on its front are details that make it a useful starter for building up experience with the printing process. Actually I have more use for a class 84 shunter which saw wider use over Belgium. The class 84 has a slightly higher motorcompartment with both sides having identical louvre doors and the cab has rounded roof edges. The body is designed in C4D and printed on the Anycubic Photon with Monocure grey resin.

Clearly there are some errors in the first print attempt, but on the whole this looks pretty useful. The outcome is robust and it also shows the fine detail in the louvre doors in right hand side! Where the layering errors precisely are generated is not entirely clear at the moment, it can be either the STL interpretation or the 3D sliding software. The Anycubic 3D slider output shows some spurious layers on the inside of the cab, precisely at the heigth of this deformation visible on the outside. Thus some cleaning up of the STL file is necessary, but even such as it is now, this result could be used with a little bit of hand work with cutting, filling and sanding.

C4D doesn't come up with meshing errors in this particular area, although it showed some minor manifold errors elsewhere. I used to have something called Nettfab, but this doesn't run on my current OSX and seems no longer free open source. Thus I now tried checking the STL file with a repair program called Meshmaker, which does run on OSX, but it literally pointed to 100's of problems over the entire surface, however none in the area where the layering error seems to originate. Thus all-in-all this is a rubbish program as the printer obviously doesn't have any problem with all these supposed errors. My conclusion was that we have a specific Anycubic slicing error here, and it can be easiest traced by inspecting the slider output beforehand. The image below illustrates the forming of it.

trial body print
The error can be seen developing in the LCD viewer layer 1087 is OK but layer 1091 shows the inside cab filled but it should show only a circumferencing wall.

Eventually it was repaired by retrieving incremental save #2 and redraw the inside of the cab before its boolean subtraction from the outside. This solved the slicing problems and the resulting cab part was copied into the latest version #19 without any problem. Simply another feature of the Anycubic slider to account with and careful inspection of the slider output before sending it to the printer can trace such errors.

The latest drawing iteration #19 with everything in place, details are accentuated by increased depths around the doors to enhance the visibility in the print. You can control the slider with your mouse to any in between position for inspection of the model.

The trial print serves the purpose of seeing how the body comes out and where changes are necessary. A first conclusion is that the model comes out pretty well and the chosen wall thickness of 0.8 mm seems OK. Particular the corner stand between windows of the cab look to be robust enough to survive handling. The second conclusion is that surface detail such as the doors need extra accentuation for better visual impression, otherwise these get lost in the noise of the 3D print, which is softening the sharp edges in the print. Amazingly even the tiny steps in the louvre doors on the left hand side are visible in the print but this only under the right oblique lighting conditions.
The next question to answer is whether to include the footplate and buffer beams on the main body or keep these as seperate item.

body print
front of Belgian class 85 diesel shunter.
body print
righthand side of class 85 Belgian diesel shunter. body print
left side of class 85 Belgian diesel shunter.

This new outcome looks a quite useable result, although there are some items to note. First of all the placement of the support structure is far from optimal, this is the best I could generate. The support placement is a sort of random process with lots of misses and few hits. It helps trying to rotate the model over a few degrees to influence the positioning. Here they are all on the inside, but they overlap the line of the footplate (its widest point but with only 0.05 mm). I hope this can be repaired by sanding these struts away. The best way to avoid this is, is to also draw up the support structure in CAD for keeping full control. A certain minimum vertical angle is necessary to prevent needing struts on the roof overhang and the 2 beams on the front of the cab. Strut density here is 35%, higher gives a crisper result on the lower edges, but also leads to having more cleaning up to do. About 1 strut per 3 mm perimeter as rule seems a good approach, actually most of the struts on the inside can probably be dispensed with here. Another Anycubic slider feature seems to be the underestimation of the total printing time. The slicer software predicts around 6 hours, but the printer clearly shows that it will take a good 9 hours and of course it really didn't finish faster. The printer does work unattended thus its not a problem and it doesn't make any noise.
The positive side is that even the very fine door grills on the righthand side come out crisp and complete. Don't ask me why we see the horizontal thickness reflected on top of the tank, but this is very faint and I expect you won't notice it after painting. There is a bit of layering visible on the left hand side, but if the primer doesn't hide this, than a treat with fine emery paper can smooth that. Building this model from scratch by cutting metal and scribing/milling or alternatively etching and soldering would take weeks longer than producing this 3D drawing. Also that way would probably not lead to such fine louvre doors on the right hand side, those on the left however are quite feasible to obtain by milling.
Of course if you want to produce several identical models then this is a good route as another run only takes some cleaning time for the printer and there probably will fit 3 on the platform. All in all an useful exercise.

mechanism
cross section of frame to be built

For finishing, the frame parts can be a straight copy of the class 80 diesel shunter, well predating the digital age thus there is no documented article available of that one by lack of photographic evidence, but it is of a similar construction, a massive frame with split axles, such as that described here: V36/type 230 diesel shunter
This one needs 8 mm wheels and is several mm longer, it will get an 8 mm Tramfabriek motor. It is probably large enough for fitting a sound decoder as option.


copyright: Henk Oversloot
date: 29 Sep 2019