i saw this vid the other day,
this one has much thicker spring mechanism than yours, and it has permanent warping pretty easily, so please check to see how your system handles it
I thought this was obvious just by looking, but you're making me feel like it's not?
Show Image
Here a thin unsupported part is holding the entire weight of the plate/pcb combination against the force of the typing, this is exacerbated by the long travel distance that the force has to travel laterally
Looking at this, it's easy to see that it's liable to bend. The whole 'flex' that this 'leafspring' generated relies on it bending or deforming. The design is banking on the mounting point being close in proximity in the up/down [edit: for clarity] north/south dimension to compensate for a lack of support provided by the mechanism.
My implementation is the complete opposite approach, where the 'leafspring coils' themselves provide support. It's obvious to me that a more tightly bound coil, as opposed to no coil at all, will provide more support. It's probably something to do with when the force travels around the corner of the bend, and would deform as in the example of the board you have provided, that deformation and tearing force is resisted by the bend in front and behind. That simultaneous transfer and resisting of force is very different from an implementation which has no support and intends to create a typing feeling through only deformation of the plate. I'm just someone who designs intuitively, so I have difficulty trying to explain what I innately understand.
Show Image
I don't think you can expect any plastic deformation from normal use, although it would be quite easy to permanently deform from improper handling, so no holding it by one mounting point and then waiving it around to see what happens.
To comment on the 40% video. A single flat bar supported only at both ends will bend in the middle. If the cross section stays the same, the longer the distance between the supported ends the more it will bend with the same applied force. With sufficient flex the metal will plastically deform and the bend will be permanent. To avoid this your options are: Change the shape to make it stiffer (why I beams are a thing), change the material, make the unsupported length smaller or redistribute the load. That said it looks like the observed bend is due to improper handling as in use the load should be sufficiently distributed in normal use.
Your 'coil' design has short straight lengths between the bends so it is unlikely to deform in the same way. The force has to go into the metal somewhere, however. In the case of your coil, I would expect the deformation to be mostly torsional at the bends. With the first bend twisting most and the last bend twisting least.
But for people worried, just look at the video in the first post, empirical is always better than theoretical. In the video the 'coil' handles a large displacement elastically without any problem
. Also, since only the bottom Bumpon is supported in the demo, it shows a much larger flex than you can expect when both top and bottom Bumpon are supported.
My solid mechanics are a bit rusty these days, so don't ask me to do any calculations to show my working
.