This weekend at Week Zero, Team 3467 experienced back-to-back failures in separate matches on our new MK4i modules. These showed initially as diminishing drivebase control in playoff matches which led to lengthy repairs between matches upon further inspection in the pit. These failures eventually led to our robot barely being able to drive in our final playoff match.
We believe there is an incorrectly spec’d screw in the kits teams received that contributed to this failure. Our intent is simply to raise awareness for all teams using these modules to prevent similar failures that are unrepairable without a collection of spare parts or extra modules on-hand.
The source of this failure was identified as the 12-24 countersunk screw that secures the offset gear to the intermediate drive shaft. This screw tore out of the shaft, taking interior aluminum threads from the shaft with it. The torn out threads are the little circle on the pulley flange pictured below. This then renders the shaft useless without a way to retain this gear.
This ultimately caused the steering of the module to bind when the screw jammed against the gear, locking the rotation of the module.
It’s important to note that these bolts were not over- or under-tightened and that Loctite 243 had been applied correctly to each one.
The bolt did not back out or shear - rather, the interior threads ripped out of the shaft due to inadequate thread engagement. The actual full thread engagement is ~1/4" since the bolt has a taper at the end and doesn’t engage near the top due to the countersink. With the intermediate shaft being machined 6061-T6, we’d expect at least 2x the major thread diameter of engagement, which is not achieved in this case. The use of a countersunk screw was necessitated by the clearance to the bottom of the idler gear (0.06375" of clearance as currently designed when measured in CAD). However, it is odd that this hole was drilled and tapped as deep as it was, but the hardware used in the CAD model made use of ~half of these threads and the bolt actually shipped out with the modules makes use of even less.
(As a general note - countersunk bolts are measured from the top of the head when the length spec is listed so a 1/2" countersunk screw actually only has 0.365" thread length to begin with.)
This image is of the MK4i CAD of the assembly with a 1/2” long screw that has been sent in the kits as specified on the hardware page and what was received with the modules. We suspect this wasn’t a one-off screw issue and more teams have these 1/2” bolts.
This image is the 5/8" long bolt that is in the CAD model provided by SDS which while this thread engagement still seems low to me would probably have been adequate.
This image is of the 3/4" long bolt that we have ordered to replace all of these bolts in every module we own to hopefully avoid this in the future. Because of the odd hardware size of 12-24, we highly recommend other teams with these modules order them as well.
Based on analysis of the CAD model and some crude caliper measurements before leaving the competition, there should be more than enough thread depth in this shaft to accommodate the entire bolt as the threads in the hole did seem to get very close to the bottom.
We are also very curious if any other teams have seen this issue either of the threads ripping out of the shaft or constant tightening of these screws. Talking with other teams, we have heard reports of frequent maintenance on this bolt in addition to Loctite being applied.
Also curious if @PatrickW could provide any insight to the inconsistencies between the CAD model and the actual hardware shipped with the modules or has any solutions to offer to aid in this very critical failure.