This is very cool.
No wonder the capacitive switches and buckling springs are so popular They are both ends of the spectrum for tactile keyboards. Capacitive switches are very smooth with a long "post-activation" curve, and buckling springs obviously have that SNAP in the middle of the stroke, while being incredibly linear on both sides of the buckling.
Everything else is an approximation of one of those two. I wonder why the other non-linear graphs are so choppy when the capacitive ones are so smooth.
His
test rig is not very accurate. If I understand correctly, he has a scale with supposed accuracy of +/- 2g, and he moves the probe down some arbitrary distance, reads the total weight (of the keyboard plus the weight pressing the key down), moves the probe some more, etc. I'm kind of surprised that he got anything that even looks like a force graph. A force gauge would really be needed to get any data I'd rely on.
A cheaper, and possibly precise enough method, would be to get a high accuracy scale, like
this (+/- 0.01g, it claims), set it on a stand that can be moved
upwards very precisely, with a rod on the pan connected to a cam that pushes another rod
down onto the keycap. The force required to move the rods and cam could be determined by running it with no key under the final rod, and compensated for. (If it's properly engineered, this force should be pretty minimal)
Something like this:
(Yeah, I know, very crude drawing.) Scale on the left, key on the right. The cam can freely turn, and the rods are hinged where they connect to the cam.