The fabrication process went smoothly and I was very satisfied with the product. The “snap” provided by the magnets used for preload is particularly satisfying. I often find myself toying with this coupling, and can see myself keeping this around as a conversation piece for a long time to come.
As before, I tested the angular repeatability of this coupling using my trusty laser pointer. One issue I ran into was the difficulty of rigidly fixturing a disc-shaped object in my apartment where I had minimal tools. I ended up attaching the bottom disc (with the grooves) to my desk using three strips of gaffer tape spaced 60° apart, taking advantage of the tape’s flexibility to place the disc under quasi-exact constraint. This is possible because the tape is virtually incapable of applying any lateral or compressive forces due to its flexibility.
Over 8 cycles, I obtained a group of points all lying within 2 mm of each other on a wall 2.9 m away. This suggests an angular error of only 0.04° — a 15-fold improvement over the melon-and-cardboard mock-up! I think at least part of this error can be attributed to the imperfect fixturing — impact loads when the balls engage the grooves may have caused the fixed disc to move slightly. The actual angular error is probably even lower!