We made about 35 significant changes throughout the competition season and off season. One change, just before the last off-season competition, led to one of the largest reductions in our cycle time.
We won Engineering Inspiration and Creativity awards at district events.
Design highlights:
Compact 3D printed swerve drives
FDM 3D printed suspension treads in TPU
All motors on chassis for low CofG.
Passive game piece claw that closes on contact
Gas shock on arm to relieve load on motors
Through robot game piece manipulation
Compliant arm joint to withstand side impact.
The cycloid drive is back drivable - so we could return our arm to vertical by hand. We needed 200:1 overall for the arm when extended with a cone. We achieved this with a 48:1 cycloid drive gearbox and then a chain drive input reduction from the NEO motor and chain drive output reduction from the cycloid to the arm axle.
Nearing the end of off-season competitions, we saw the cycloid drive gears slip (miss-positioning our arm) and so we made a conventional gearbox with VEX gears. This is what the published CAD shows. However, the output stage was at its design limit and we stripped a gear a couple of times when the arm hit a solid surface. So we went back to the cycloid gearbox with a 20% greater tooth depth (Just reprinted the three cycloid gear assemblies). This has continued without issue although it has not been competition tested.
Conventional cycloid gearboxes have many advantages but our 3D printed one is bulkier than an equivalent steel-gear planetary gearbox.
Our design uses three cycloid gears at 120 deg offset. Since about 1/3 of the teeth engage at any one time, three cycloid gears provide nearly 100% tooth engagement. Therefore, a PETG printed cycloid gearbox is capable of unexpectedly high torque output. Also, three gears at 120 deg offset provide almost perfect balance and so there is no vibration issue.
They do make great spools for retracting cord (for our arm extension and claw reset). The two small ones are 24:1.
Thank you for the response! Having 120 degree offset with 3 gears rather than 2 with 180 degree offset is fascinating. Would you guys consider using this in the future seasons? Maybe even in aluminum to make it more compact than a 3D printed one? I have been prototyping a cycloidal gearbox for krakens, but the more I research, the more I feel like custom gearboxes with spur gears would do the same job for frc robots, especially for an arm. I know that there was one team that used a cycloidal gearbox for a climber (I do not remember which year). Other than that, I haven’t seen other teams use cycloidal gearboxes. In many chief delphi threads, it seems like people attempted to 3D print them, but haven’t done anything further. Another question I had was, did it also end up helping your team with backlash when comparing it to when you guys used the VEX gears? Again, thank you so much for the information.
Hi PigTwo, yes 180 does not provide perfect balance where 120 deg does, plus you gain the increased tooth contact. We did use our 24:1 cycloids on our 2021 robot (added at some point after initial design in 2020). We had a pair of these to lift on the hanger. During testing, we found that only 20% power was required to lift but obviously we increased that for competition. They suffered no failure and in fact, we reused that design on our 2024 robot, Synthwave, for arm extension and claw retraction. We have never seen a failure of these 24:1 cycloids (except a cord pulling out because the knot was too small). We like cycloids for spools because the entitle mechanism (sans motor) fits inside the spool volume. We will certainly use them in the future should we need a spool but still 3D printed as it provides quick access to a finished product for us. Machining the gear in Al. would be very tricky!
The problem with a spur gear gearbox is that the last but one gear (typically small) can be subjected to excessive stress like our arm gearbox was in 2023. We did replace this 18T penultimate with a steel gear but it just moved the problem to the final large aluminum driven gear. So, as mentioned, we reinstalled the Cycloid with deeper cycloid gears.
One tip on cycloid design, make the offset as great as possible. We originally had a 1mm offset. The redesign took it to 1.2mm (hence 20% deeper ‘teeth’). I kept adding 0.1 mm to the offset until the formulae produced loops for each tooth, then backed off 0.1mm. I encourage you to try our design as it is now well proven. We use PETG for all printed parts.
Regarding backlash, because we were always raising the arm to a preset angle for game piece collection or deposit, we did not suffer with the increased backlash of the spur gear gearbox. We also had a pneumatic lift for the arm to balance it when horizontal which tended to provide some damping to the movement.
In summary, we use 3D printed parts wherever possible (fully printed swerve dive, wheels and treads, deep groove bearings, roller cages, cycloid gearboxes, etc, etc.). I’d estimate that 50% of the in-house designed parts on our robot are 3D printed. It’s amazing what 3D printing can do when you understand the mechanical properties and design with the layer lines in mind.
David, our 24:1 cycloid gearbox is in the published CAD on this page. If you have any problems downloading and extracting it (I’ve never done it), I can provide a step assembly of just the gearbox. Let me know. Be happy to answer any questions and provide more data if required. Cheers.
I found it, for some reason my work computer was having issues with the file and when I switched to my own I can see it, cant wait to tell y’all how it goes.
David, great!
I recommend that you do a test print of the holes into which the steel pins insert. They need to just press in by hand. We use PRUSA printers, ymmv.
Be very careful on assembly to get the three cycloid gears indexed correctly (use the index markings).
