For this ENGI 210 assignment, we were tasked with the goal of printing a handful of impossible objects small enough to fit in a gumball machine container. While this project would be somewhat straightforward on its own, the fact that I had no access to the OEDK and am currently living just over 1,000 miles away made it near impossible. Below, I walk through the entire process of printing these impossible objects.
To begin the process, I started with the selection of an “impossible object”. To me, the objects that seem to be most out of reach for traditional subtractive machining are components that have mechanisms “printed in place”, such as gear systems printed continuously, or loose objects in closed-off internal cavities. As mentioned in a previous blog post, I have a lot of experience with machining aluminum and steel, so seeing objects that show off these newer manufacturing techniques is exciting. The object I chose is very simple — a sphere enclosed in a small cubic frame. I got the object from Thingiverse and didn’t have to make any major modifications. I was considering pursuing a “print in place” piston, though for reasons I go into more below, I felt like I needed to have a simpler part for this particular project.
I would normally consider myself generally proficient with CAD; I have taken multiple classes for the MECH major, and have used CAD for a wide variety of projects. However, the only way I can access the software that I am most familiar with, SolidWorks, is via remote desktop, as my MacBook laptop cannot support the software. Unfortunately, my laptop also doesn’t do well running the remote desktop software and SolidWorks simultaneously — most of my efforts result in long loading times, crashes, and frozen screens. Given the circumstances, I chose to not modify the part found on Thingiverse, and print it “as is”.
Next came the actual printing process. Again, this has been a normally straightforward process for my time here at Rice, but this particular semester I’m having to adapt. I would normally go over to the makerbar and print these objects myself on the computer. I’d be able to see all the printers, which ones are currently working, and ask for lab assistant help if need be. After I start the print, I’d also be able to check for first layer adhesion, make sure the filament is coming out cleanly, and check to make sure the nozzle extrusion looks good. Because I’m in the OEDK so often, I’d also be able to check my prints periodically and cancel them if something goes awry. Finally, I’d be able to take the prints off the bed myself, remove supports, and post-process accordingly. Alas, none of that was possible given the timing of this assignment.
I was able to print well remotely with 3D printer OS, a functionality I’ve actually never taken advantage of before, with relative ease. The layout and slicing of the part were identical to the way it’s done on the makerbar computer at the OEDK, which was very helpful. However, some aspects were just not the same. For example, the built-in camera that was supposed to be facing the printer bed was off, meaning I just had to hope the bed was clear when I sent my prints.
To complete the assignment, I had to rely on lab assistants to check on the prints, post-process the prints, and pretty much every step described above after simply printing the part. On my end, I just set up the prints online then just crossed my fingers that things would turn out all right on the other side. I found that the easiest way to organize this was to communicate with my friends who are lab assistants and are currently on-campus. I timed my prints so that the majority of the printing process happened while they were on shift, and I would ask them to send me updates on how the part turned out after they got it off. I would even ask them to send me pictures of the parts (which is how I got pictures of any physical printed parts). This was especially difficult given the iterative nature of the 3D printed part design. It was difficult to build off my mistakes when I couldn’t even see the mistakes I was making. Their help was especially appreciated when using the Form (SLA) printers, which are now in the wetlab and not connected to 3D Printer OS. Being remote, I was completely reliant on them to print using those printers. The entire process was so incredibly inconvenient, and I am so grateful I had their help. Their labor cost is totaled into the part cost below, even though their services are free for OEDK users.
For the manufacturing itself, I learned some key lessons about 3D part layup. First of all, a raft was absolutely necessary to create an initial adhesion of both the cube and sphere to the bed. Without it, the print would quickly fail. The sphere especially had difficulty sticking. Next, the recommended layer height of 0.15mm for smaller parts seems to be best — I tried to increase this height to speed the process up, but the layers were quite noticeable in such a small part. Finally, the supports around the overhanging cube portions were critical. Many of my prints would fail towards the end, and the printer would appear to be printing onto the air after portions of the cube would collapse. I feel like I never truly fixed the issue (the software auto-generated supports were never quite the best) but adding supports helped tremendously. All of the above lessons apply only to the FDM printers — given the circumstances, I was not able to learn any lessons from SLA printers, as I didn’t have access.
As for the cost of the project, here’s my breakdown:
PLA material cost (FDM): $0.03 x 7 = $0.21
Resin material cost (SLA): $2.00 x 1 = $2.00
Lab assistant labor: $15/hr x 4 hr = $60.00
Gumball Containers: $0.10 x 5= $0.50
Total: $62.71
Note that this cost estimate doesn’t take into account upfront machine costs (such as buying a 3D printer), or associated hourly machine usage fees (if rented). Also, note that none of these prices are charged to Rice students; the lab assistant labor, especially, is for demonstration purposes only.
Overall, this project was helpful and interesting, though I can only imagine that it would be much more so if I were in-person. The 3D printer OS software enables students to print remotely on OEDK machines, but that doesn’t mean it along with some youtube videos can effectively teach students how to print. The brilliance of 3D printing is the fast-paced, iterative process. It’s fun and helpful to quickly prototype and design using the technology. However, I believe much of this is lost when the results of the print have to be communicated via email from a friend or professor. In order to experience the manufacturing process, and really learn what works and doesn’t, you need to be there. Maybe once I get back to Houston, I’ll print some more impossible objects for fun. In the meantime, I’ll do my best to be flexible in this less than ideal situation, and hope other remote students didn’t have as rough of an experience as I did.
– John Perez
