Over the past week, I have been working on the ENGI 210 “3D Printing” homework assignment, which involves selecting an “impossible part” on Thingiverse (a part that could not be practically made with any other fabrication technique), scaling it to fit within a gumball, and 3D printing it using at least two different printing methods. This blog post will cover my process, areas for improvement, and cost analysis for my impossible 3D prints.
Figure 1: My gumballs with their info sheets removed so the prints can be more clearly seen.
My Process
My process for this assignment was somewhat hectic. I ended up going through three different impossible objects before I could get one to print correctly. It is worth noting that my initial prints with each model were done with the Prusa FDM printers on the Maker Bar wall.
Our entire class began this assignment by searching Thingiverse for items that were “impossible.” Initially, I went with an ornament that seemed to qualify as “impossible” due to its geometry. Unfortunately, I could not find a way to slice this object where it was small enough to fit inside a gumball, did not print with poor quality, and did not generate supports that were near impossible to remove without breaking the piece. I gave up on this piece very quickly.
Figure 2: The first impossible object I tried to 3D-print.
Figure 3: My FDM prints of the first object.
Next, I decided to try an impossible object that would be “impossible” due to its ability to move. I began by trying out a cat that could bend throughout its body as well as its tail. I had previously printed this object in its original, scaled-up form which turned out nicely. Unfortunately, scaling this piece down presented lots of issues. When I tried to print this part with a raft, it would work just fine, but the tail pieces were so thin and weak that it was near impossible to remove the raft without breaking the tail altogether. When I tried to print this part with a brim, it would also brim the hollow parts on the bottom edge of the tail, creating internal pieces that prevented the tail from flexing and, similarly, were near impossible to remove without breaking the part. Finally, I tried printing with a skirt, and even after a few attempts, the tail would not stick to the bed properly. The common denominator in all of these issues was the cat’s tail being too thin, and I couldn’t scale it up any further while still having it be able to fit in the gumball, so I thought that perhaps if I could find a similar object that did not have such small features, I might be able to get a successful print with a similar “impossible” nature.
Figure 4: The second impossible object I tried to 3D-print.
Figure 5: An FDM print I made of the second object (notice the raft is not removed).
At this point, I pivoted from the cat to “Flexi Dog,” a model of a dog that bends almost identically to the cat, except that it has fewer degrees of freedom in its tail. Printing the Flexi Dog with a brim or skirt yielded the same issues that were present with the cat, but when I printed it with a raft, with patience, I was able to remove it without tearing the piece apart, yielding my first fully successful print. Excited to finally have a part that worked, I quickly printed two more on the Prusa FDM printers on the Maker Bar wall, and then printed two more with the Prusa SLA printers elsewhere in the Maker Bar.
Figure 6: “Flexi Dog” on Thingiverse.
Figure 7: An FDM print I made of “Flexi Dog.”
Figure 8: The SLA prints I made of “Flexi Dog.”
After carefully removing the rafts from my FDM prints and the vertical supports from my SLA prints, I printed off the paper tags to accompany them inside their gumballs, put the gumballs together, and at this point, was able to call this assignment a success.
Figure 9: Gumballs packed and prepped for the gumball machine.
Areas for Improvement
Looking back, I’m not sure what I could have done differently to improve my final products. Most of the work in this project was done by the 3D printers – I essentially only found a file, scaled it, sliced it, and pressed “go.” My pieces do have a few imperfections associated with the supports generated for each respective 3D printing methods, however.
Like I said, for my FDM prints, I used a raft support, which is essentially a flat surface that is 3D printed under the actual object that helps prevent warping and promote bed adhesion. Rafts were essential for this impossible object – with a skirt, parts of the object would not stick to the bed and would end up attaching to and clogging the printer nozzle. With a brim, flat parts that were difficult to remove would form where the object was supposed to bend, rendering it inflexible. It is clear that the raft is definitely the only support that could produce a successful print of this particular object at this size. However, when removing the rafts, they would leave an ugly finish on the surfaces they had previously been attached to. This is clearly shown on my black FDM parts; the side that had been attached to the raft looks somewhat discolored and spotty.
Figure 10: An example of the discoloration caused by removing the rafts from my FDM prints.
I used a different kind of support on my SLA prints, which also had side effects on the final products. The SLA printer, rather than skirts, beams, and rafts, forms a support that is made up of a solid clump of resin that attaches to the print bed and is connected to the 3D print by means of thin, vertical resin strands. Removing these strands is simple – you just cut them off from the part. However, even after they are cut off, there are still little bumps left from where they had been attached to the part. While not a functional issue, it is an aesthetic imperfection.
Figure 11: An example of the spots created on my SLA prints by the supports.
While these side effects are unfortunate, the supports that caused them were very much necessary for the success of my prints, so I do not think there is any way I could have done this assignment without creating those imperfections. Perhaps there is a method of post-processing that could make them less severe, but I’m not sure what that would be. The best idea I have would be to sand down the bumps on the SLA parts, but I also would not want to do this since it might make the smooth, uniform parts of the surface (a major strength of SLA-printed parts) look scratched and feel more rough.
Cost Analysis
For this project, the elements that could have accrued cost are the FDM filament used to produce the FDM objects, the resin used to produce the SLA objects, the time spent using the 3D printers, and the opportunity cost of my time.
FDM Filament
The cost of the filament used to produce each of my FDM prints (as calculated by 3DPrinterOS) was $0.10. Since I made 3 of these, they would cost a total of $0.10 x 3 = $0.30.
Resin
Since I can’t find enough information online to accurately estimate how much the resin used to produce my SLA prints would cost, I will assume that it is approximately the same as the cost of the filament used to produce my FDM prints, especially since my pieces are small and the actual price likely is very small in comparison to some of the other costs anyway. Therefore, I will estimate that each of my resin prints cost $0.10. Since I made 2 of these, they would cost a total of $0.10 x 2 = $0.20.
Time Spent Using 3D Printers
Although I spent quite a bit of time using the 3D printers due to how many different objects and scales I attempted, I am only going to do this analysis for the final 5 pieces that I am turning in. My FDM prints took about 3 hours of printer time, and my resin prints took about 2 hours of SLA printer time. I can’t find a figure for a standard hourly rate of using a 3D printer, but I have seen a figure a few times online where a $2000 printer would have 2000 hours of expected printing life, which equates to $1/hr. Someone who bought a 3D printer would want to profit, however, so I will assume $5/hr for both types of 3D printers. So, the cost of this element is $5 x 5 hours which is $25.
Opportunity Cost of My Time
Most of the time that this project took was not time that I had to be actively present or working on it – although my process was somewhat hectic and hit a few dead ends, all I did at the end of the day was search for files on Thingiverse, scale them down, slice them for the 3D printers, and press “print,” followed by a bit of post-processing work. I would estimate that I spent about 2 hours of my own time on this project, and my time as a Tech TA is valued at $12/hr. Therefore, the opportunity cost of the entire project is $12 x 2 = $2.
Overall, the total cost of this project was $0.30 + $0.20 + $25 +$24 = $49.50, a low price compared to that of other projects I have done for this class. If I did this project again now, I anticipate that I would be able to dramatically reduce this price. Now that I have found a part that is able to be successfully printed at a small scale, I could dramatically reduce the cost of this project – all I would need now is enough time to find the part on Thingiverse, scale it once, slice it on 3DPrinterOS and Preform, press “print,” and do relevant post-processing for each printing method. This would probably account to be around half an hour of my time at most, reducing my opportunity cost by $18.
