Colorado Springs Scrimmage: Three Lessons
October 11, 2026
lessons-learned · mechanical · build-quality
We left Colorado Springs with a robot that mostly worked and three lessons we don’t want to relearn.
1. Test the robot before it goes on the field
Too many problems showed up during matches that a two-minute check on the cart would have caught. Starting now, the pit crew runs the pre-run checklist before every queue, and someone signs off on it. A quick post-match inspection catches damage before the next run.
2. No more 3D-printed pulleys
We had multiple pulley failures, all avoidable with metal parts. The clearest one: a printed 32-tooth pulley on our rack-and-pinion slide.
The setup was a NEO with a 15T pulley, a 9 mm belt up to a 32T pulley on 1/2 in hex, and two 32T pinions on a rack. When the slide jammed, the motor stalled at its 60 A current limit.
- The motor puts out about 1.1 ft·lb at 60 A.
- The 15T→32T belt stage more than doubles that: about 2.3 ft·lb on the 32T pulley. That’s like hanging 4.6 lb on the end of a 6-inch stick.
- A 1/2 in hex pushes on the pulley through six tiny contact patches at its corners. Each corner was pressing into the plastic with about 31 lb.
Plastic can’t take that, especially when every start, stop, and jam hammers the corners a little more. The hex rounded out into a circle. The belt and teeth were fine; the bore failed.
The rule: anything carrying torque — pulleys, gears, hubs — is aluminum or steel, COTS where possible. Printed parts are fine for guards, spacers, and brackets. Not sure? Ask a mentor before it goes on the robot.
Try the numbers yourself in the belt-to-rack calculator.
3. Design for maintainability
Parts of the robot use long hex shafts running all the way through a mechanism. Getting everything lined up is a nightmare, and long shafts twist under load so one side lags the other.
For the next revision:
- Break long shafts into shorter segments joined with shaft couplers.
- Use shaft collars to locate everything consistently.
- Consider driving each side independently, or driving from the center.
- Ask: can we pull this mechanism and swap it in under 10 minutes at an event? If not, redesign.
See the long shaft vs split shafts comparison for the twist math.