With the easily accessible Prusa i3 machines and 3DPrinterOS, I don’t think I’ve ever had an easier time manufacturing anything. Admittedly, I still iterated through a few designs that were a bit too finely detailed for FDM, but I finally settled on one that worked, albeit with an artifact or two. Here’s how:
Level 1: Thingiverse!
As stated above, I tried a few designs before settling on one that worked. I recently got into Magic: the Gathering, so I wanted to make a D20, which some of the cards call for. I found some awesome icosahedrons on Thingiverse, but they ended up being too detailed to make on the Prusa:
While cool, there was no way I was going to be able to make any of the impossible D20s I found online without getting a glorified blob of PLA, so I ultimately had to scale back and go for a D12 (dodecahedron) instead. This is the design I settled on, made by Brendon O’Keeffe:
The impossibility of this design comes from the “hyper” geometry in the interior parts of the object. The holey parts of the interior could not be manufactured using any method other than 3D printing.
Level 2: Printing
The issue with holey interior geometries like this one is that the supports generated on 3DPrinterOS are a nightmare to remove completely. I thus decided to lay the die down flat on one of its faces and print using no supports, opting for a 75% infill instead to provide extra sturdiness. This ended up working, generating dice like this:
As you can see, the geometry printed well and the numbers are legible, which I am quite happy with. However, printing without supports as I did does have its drawbacks. The core of the die, where the numbers are debossed, overhangs the face on the bed, and cannot support itself well when printing at first. This leaves a significant artifact on this face of the die, which was present on all of my prints:
The number on this face is thus illegible and the strings of PLA deposited by the extruder are highly visible. While I am slightly disappointed in this, I believe it to be preferable to having support artifacts in the holey parts of the hyper-geometry. After all, it’s only one number, so you can always decipher it by process of elimination! If I were to change anything about this process, I would add my own single support column on this face by extruding to surface in a CAD software. I am unsure if this support would be easily removable, as the printer interfaces parts of the file and auto-generated supports differently, but I believe this option is worth exploring in future projects.
Another regret I have is not being able to print via SLA on the FormLabs printer, as I believe SLA would serve well in producing geometry like this hyper-dodecahedron. Instead, I experimented with using another FDM printer, the Ultimaker 3. This print was slightly slower than the Prusa, but printed nearly identically to the others, even including the bottom-face artifact:
The Ultimaker also records a great timelapse, which can be seen here:
With this print, I had my 5 identical D12s complete, all impossible to make using a process other than 3D printing!
Cost Estimate
~59.85 grams PLA at $0.02/g: $1.20
~2 hours labor at $15/hr: $30
Total Cost: $31.20
