That was my original thought but
A) Wood dimensions aren’t perfect, and my 3d printed corner brace. It’s sub millimeter accuracy but what might fight tight on one side won’t fit at all on the other without a hammer. A cut spacer would result in movement on the Z axis because of imperfect 3d printing and wood. The spring I knew would work for any corner that isn’t exactly the same width. They’re all within a millimeter but eh.
B) Only upwards forces are placed by the device. The only downward force is gravity on the belt end when there is no tension on the line. As long as the spring fully resists gravity of the belt + end when it is slack, there is no movement! The spring is pushing up, not down.
C) A spring is universal for anyone else with the same problem
D) Maybe most people are mounting to the top of the frame above the corner bracket. It didn’t dawn on me until I stumbled across a random assembly photo, I didn’t see an explicit callout in the user manual. That’s my misunderstanding. A spring would not be needed there, but then a cotter pin is also better than a bolt.
E) I did manage to get a very good calibration fitness with it, and calibration finished quickly whereas it failed twice without
F) It’s not unheard of, many CNC machines use eccentric nuts or springs or other things for alignment, when the spring or friction will fully resist forces at hand
I’m going to have it draw some shapes with a pencil soon, and if it’s off at all I’ll move to mounting the Z on top of the corner brackets.
There might be an advantage to my misunderstanding - there’s only significant lateral forces on the bolt, so it can actually be not tightened down at all. Any extra upwards force is going right into the frame through the corner bracket instead of pulling on the bolt and on any slop in the cotter pin.