tl;dr – I made an impossible gear keychain, which you can check out here: https://www.thingiverse.com/thing:6952946

This took much longer than expected (read: 14 total jobs sent to printers), though the journey increased my capabilities in Fusion360 and tested the limits of my patience.

 

 

 

 

 

I chose to make five impossible objects. This was mainly because I wanted to create a little gizmo, but also because I believed, if needed, modifying the CAD model would be easier.

 

I settled on this gear fidget toy from 8jack on Printables. There were no issues with the initial print, but the gears themselves weren’t that fun to play with – it requires two hands to awkwardly fidget with it. The print was also a little small for a gumball capsule, so I needed to adjust the scaling. I decided this was where a quick modification would come in handy.

 

 

 

 

 

 

 

 

 

Dear reader, it was not a quick modification.

 

 

 

I cut the retaining ring to make the large gear’s teeth stick out, like this Makerbot fidget gear. I also sized everything up a little on the Bambu slicer (increased the length from 1.75” to 1.8”). Here’s a quick rundown of my subsequent print attempts:

 

 

Print 2: First few layers delaminated (despite using glue!). Cancelled the print, tried again.

Print 3: Same issue as #2. I was very heavy handed with the glue stick for print #4.

“Prints” 4 and 5: Scaled the file incorrectly (input 1.8 as the scale factor instead of the x-dim size), so it was too small to actually print. (The printer would lay the calibration lines then report the print as done, since the piece was smaller than the filament).

Print 6: Printed spaghetti.

At this time, the Bambu printer I was using threw errors I didn’t know how to solve and there was no TA, so I moved to a different Bambu printer.

“Print” 7: The wrong bed type auto-selected in the software, so right after I sent the file to print, it threw an error. I didn’t check on the process until about 7 minutes in (halfway through the total print time), so I was shocked to see an empty bed.

Print 8: Came out nicely! But the large gear easily popped out. I decided to modify the file to have more rigidity so the gear wouldn’t accidentally fall out.

 

the various stages of grief troubleshooting this print

 

It was at this point that I had to reckon with my CAD capabilities. I was already extremely familiar with Autodesk Inventor since I used it in high school engineering classes and was design lead for my school’s robotics team. The issue with Inventor is that it sucks my knowledge base does not cover importing and editing STL files, only creating parts from scratch. After grappling with it for maybe 15 minutes, I decided to download the kindergarten version of Inventor Fusion 360, since I was only making simple changes to geometry and could find easy-to-read threads about converting STLs to editable files.

I promptly spent a good few hours fighting with Fusion 360. After adding a bridge to the retaining ring across both gears, the smooth form of the exterior was disrupted. I knew exactly how to remedy this, but the geometry was complicated and Fusion refused to cooperate, meaning I had to find obtuse workarounds. After a bit of agony, I had done it.

 

 

 

 

 

 

I was playing with print #8 throughout this CAD marathon and noticed that the small gear’s teeth kept catching on the large gear’s teeth, so I shaved them down in the model to allow for smoother meshing. I sent this off to print and prayed that the supports between the gears and the new bridge wouldn’t cause an issue.

 

They caused an issue. The gears on this print (#9, for the tally) just did not move. I looked up from my computer and recognized it was rapidly approaching sunrise. I called it a night.

On my dejected walk back home, I kept thinking about how to improve the design. It seemed too complicated to try to resolve the supports issue. I decided that going back to a full perimeter retaining ring would be easiest to prevent the gear from popping out. And to make the fidget more fun to play with, I could just make the large gear more tactile.

 

 

 

 

 

 

This version printed fine. The gears didn’t spin immediately, though after some coaxing they began to mesh smoothly. However, despite my use of rolled over edges for the tactile bits, the raised plastic was pretty uncomfortable to push on. So I completely scrapped this idea. 

 

I went back to the cut retaining ring design, but tightened the angle of the sector cutout from 120 degrees to 90. This, hopefully, would make the gear difficult to remove. I also recessed part of the large gear and made its teeth taller, just to add some kind of grip that wouldn’t be painful like the past iteration.

This printed just fine (#11). I tugged on the large gear and it gave some resistance, then popped out. However, it was much more difficult to remove this time, and I realized it was also fun to pop out. With this quick reframing, a bug converted into a feature: the gear is supposed to be removeable. Now I had my finished design.

I printed 4 more in pink PLA, all with no issues. It was time to try a different printing method. I sliced and set up a SLA resin print, very much ignoring the large “no fun prints” sign. After printing, washing, and curing …

 

 

The gears were stuck!!! The small gear had fused to the retaining ring. I figured this was because the gap between the ring and the gears was just too tight. At this point, I did not want to waste resin while troubleshooting the tolerances, so I decided to do another FDM print with a different printer. 

 

I set up a print on one of the PRUSAs and let it rip. The first layer shifted. I sent the print again and finally finished up my 5th impossible object to round out this project. I quickly removed any stringy filament, sanded down rough surfaces, and placed each keychain in its own gumball capsule.🕺🏽

 

 

 

All I had left was to print slips of paper with the Thingiverse link to the file. I created an account, uploaded my .STL, and hit a wall. Or rather, a banner. At the top of the site in bright red, I was informed that new users must wait 24 hours before publishing. This meant I wouldn’t be able to post anything until 2:26 PM the next day (aka after 12:00 on Monday). Let’s take a step back and record some stats for this weekend’s project:

3D PRINTING STATS

• Hours in the OEDK (Fri-Sun): 22+ 
• # CAD files intended to be the final version: 6
• Meals eaten: 3
• Dr. Peppers consumed: 1

It was time for an executive decision. My impossible objects would have to be slip-less at the deadline. I left a note with my objects that I would make the slips ASAP, cleaned my workspace, and left the OEDK. 

 

 

 

 

 

 

 

 

Here is a link to my impossible object on Thingiverse (thing 6952946): https://www.thingiverse.com/thing:6952946

Below you’ll find a cost breakdown of this project. I split it into cost for the final product (aka only the 5 objects turned in) and the ideation / prototyping cost to get to the final product.

*based on Prusa’s justification to use average wage in the country of operation. Average hourly wage in the US (Jan 2025) is $30/hr. Source: https://tradingeconomics.com/united-states/wages

** assumed similar operating costs for curing + washing machines as SLA printer (couldn’t find reliable quote otherwise)

Link to table: https://docs.google.com/spreadsheets/d/1kIJlYjq_M2OLwzJHWlSc7ev0SPiPXj62SsT9ADeZJuk/edit?usp=sharing

 

All pricing sources

PLA filament / kg: https://us.store.bambulab.com/collections/bambu-lab-3d-printer-filament/products/pla-basic-filament 

Labor, facility cost estimate (FDM): https://blog.prusa3d.com/how-to-calculate-printing-costs_38650/

Clear V4 resin / L: https://dental.formlabs.com/store/materials/clear-resin/

Prototyping engineer wage: https://www.ziprecruiter.com/Salaries/Prototype-Engineer-Salary 

Labor, facility cost estimate (SLA): https://3dprintingindustry.com/news/how-to-accurately-price-for-stereolithography-sla-3d-printing-projects-171977/