We probably should have anticipated that putting a huge knob on the rotary switch would increase the mechanical stresses on the switch. The switch held up to a few hours of vigorous use by the many children at STEAM Maker Festival, and then started showing signs of failure. At one point I caught it acting as an 11-position rotary switch, instead of the 10-position switch it is supposed to be. The detents still lined up with the panel markings, but some (!) of the functions were offset by one position. Then, the end stop stopped working, leaving the knob free to go all the way around in one direction but not the other. Finally, the switch stopped working entirely, making no connection at any position.
We had to alter the software to ignore the rotary switch and stay in one mode (the keyboard visualization with waterfall effect) for the rest of the event.
Back in the lab, I pulled out the rotary switch for failure analysis. I found two problems, explaining the observed symptoms. The end stop relies on a round sheet metal plate with a tab bent down. That plate was bent, leaving the tab at an angle, so it could ride up over the stop in one direction but not the other. It's easy to see how the plate could get bent when the knob is slammed against the stop. The other problem was that the moving contact that wipes over the ten fixed contacts was broken off. I found the contact loose in the housing.
I speculate that the stop plate bent first. The wiper contact then fell off the end of the fixed contacts into the gap between the first and tenth contact. When the knob was returned to a normal position, the wiper contact had to ride up onto the first or tenth fixed contact, which it was not designed to do. It was able to survive a few such transitions before breaking off.
So, we need to find a more robust rotary switch or else go back to a much smaller knob.