I watched the starting of operational RNLAF F-104s from close-by on two occasions, and the noise of the little jet starter cart was very impressive. I think the type was only used by the RNLAF. As far as I know, it was built by Fabrique Nationale (FN) and used a Boeing 502 jet engine. I always wanted to scratch-build one, but I never found enough information, and it seems not a single one was preserved.
|Two old issues of the IPMS-NL magazine MIP from 1978 contained a simple drawing with a few dimensions, but it was not sufficient for a model. Recently I discovered the NIMH Beeldbank, and found 700+ RNLAF F-104 photos. A small subset of these photos showed the jet starter, enough to make a 2D drawing. It is shown here in unfinished form, with a rough coloring.|
|Next I converted the 2D drawing to a 3D model, so it could be 3D printed. Fortunately I still remembered how my old 3D CAD software worked. In one evening I did the top half, the next evening the bottom half. Both are preliminary designs, with many details missing. Shown here is an STL export, using this online STL viewer.|
|The third evening was spent detailing the model, with some twenty changes and improvements, including an instrument panel, piano hinges, a cooling air regulator, and beams for forklifting.|
|Work on the fourth evening was mostly spent on the rear side (towing hook, rear light), right side (fire extuingisher, parking brake, battery box, corner reflectors) and top side (exhaust with lid). On the front side I added a fuel gauge and improved the instrument panel. Left to do are the tow bar, and the wheels and tires.|
|The wheels and tires were by far the biggest challenge for me. I did not have all measurements of the wheels (like the depth of the dish), so I had to design something rough and then tweek it visually. Interestingly, the wheels had two bolt patterns, the outer one serving to assemble the multi-part wheel itself. Over to the tires, where the thread was the difficult bit. It became a compromise after crashing the software too many times. Still, I think it will do for a small model like this.
For this and subsequent picture, I used another online STL viewer that did not have the extreme perspective of the previously used viewer.
|In the last round of work I added the wire basket and the towing bar. The wire basket is better made from metal wire than 3D printed, but I wanted to include it for a good evaluation of the 3D CAD model. Having it in 3D will also allow the development of a jig or a mold to help make the wire basket.|
|The finished model seen from the rear. It lacks all the handgrips and lifting hook by the way, that wil be made from wire.|
|Here's how I broke up the starter cart in four parts / groups of parts for 3D printing. It's roughly based on how I would make a silicone rubber mould for resin casting. Let's see whether this works for 3D printing too. Still to be designed is a mould / jig to build the wire basket.
The large-diameter exhaust will be made of K&S Precision Metals 3/16" tubing with .014 wall thickness (4.8 mm x 0.4 mm). I can choose between #8104 aluminum, #8114 copper and #8129 brass. Maybe aluminum is the easiest to paint like the original, using transparent paints.
One detail that I forgot is to make the air hose couplings. My reference photos barely show them, so it will be difficult to recreate them.
Regarding the STL export, I still have to learn what accuracy is required. Of the four paramaters that the CAD software offers, only the 'Surface deviation' and 'Normal deviation' parameters are important (in my understanding). Here are some results with different parameter values, of the top part of the start cart.
| Surface deviation |
| Normal deviation |
| Binary STL file size |
| Facets |
In my experience, an ASCII STL file is a factor 5.62 larger than a binary STL file. Regarding the number of facets, for an ASCII STL file 1 MB equals ~3,600 facets. For a binary STL file, 1 MB equals ~20,500 facets.