That's a fuckton of information in a single picture, so we shall take the switch appart at first. The switch has a total of 10 parts consisting of the spring (I could model it but I hate doing them), slider, spring guide, 4 2x2x1mm magnets, housing, clicker and capacitive element. Anyways, the parts can be arranged in three groups according to their common purpose: housing (boring), slider and clicker. Let's take a look at the slider at first:
This thing is quite simple, you stick a normal MX keycap on it and it goes up and down, but with two added oddities that make this more interesting. The spring is the same as in Cherry MX switches, but because the design lacks a bottom housing to hold it in place, I had to design a spring guide, which is basically a moving part that prevents the spring from scrap the PCB or bend. The other two pieces are fixed magnets, their only purpose is interact with the other two magnets in the clicker.
Clicker:
This part is the trickiest, it took me a while to think the best possible way to make it work. Talking about the operation principle, we press the slider over it, the slider's magnets bypass the clicker magnets and because of the antiparallel configuration, it shoots the clicker up, producing bot a tactility and sound feedback and moving away the capacitive element under the plastic part (yes, this switch is normally closed like beamsprings). The thing is, yes, so easy, right? Right !? NO!!!! HOW DARE YOU!
As you might see the shape is weird. I wanted to make a very stable, maybe stupidly stable switch so I needed the moving parts to be as wide as a 1U key can fit (you can always make the switch taller, but this will make the entire keyboard taller). Also the force applied on the clicker must be as centered as possible. Ring magnets can make this easier but they come in not so great variety. So, the clicker basically goes right under the slider, in a way that the whole switch can be used without one coliding with the other, at the expense of that weird shape and not ideal stacking, neither the worst as the switch is sightly above 21mm. I think the tradeoff is better at the end, who knows.
Housing:
Do I need to explain how a housing works? Well, in this case yes. I have a serious problem with nowadays and even vintage switches, not with what the housing does, but how it does it.
A)The mounting methods suck in most cases, square hole plates are harder to make, unless you mass produce plates by press cutting. Circular plate holes are way, WAY easier and cheaper. You can either CNC a plate in fewer steps or go with the good old marking tip and a very careful drill press job. If you already got one, you can stack them and make another plate out of it.
B)I dislike this whole new concept of hot swappable switches, but there is a smart and absolutely fair point in them. Applying heat to desolder/solder/reflow/whatever a switch is basically applying permantent damage to your PCB. You have a limited quantity of swaps until the traces burn and die. If this thing was resistive like a normal contact based switch, I would still use an open housing, simply because is better. Yes, you have to be a bit more creative, make a sandwich assembly and ensure to have a PCB, but come on, you can deep clean this thing as many times as you want.
C)I get the point with make switches that are shorter, but, why thinner? Beamsprings are the kings of stabilization, maybe because of the crazy amount of surface they cover, on the other side of the coin, Mitsumi miniature mechanicals suck because of their microscopic size. Come on, engineers, make your switches THICC.
(joke)