B2H reloaded: Dual Magnet Clicky MX style switch

[Lalaland124]

The "why":

By now some of you might know that I'm a huge fan of Omron's B2H switch Chyros showed in his video (https://www.youtube.com/watch?v=q4Gvj_2TZOI&t=144s). Over the last year I tried finding these switches myself - However, having contacted almost all major Omron HQ around the world multiple times and having dug through dozens of Japanese Google Book files together with Beardsmore, I have to admit that I greatly underestimated the hustle of locating a 40+ year old product. So a few months ago I decided to deal with my frustration by thinking of a way to bring this ingenious old switch mechanism back into the modern keyboard world.


The "how":

Here's what I came up with so far.

The 3D CAD model:

Spoiler:

B2H_MX2.png (111.31 KiB) Viewed 10131 times

B2H_MX3.png (71.27 KiB) Viewed 10131 times

B2H_MX1.png (130.12 KiB) Viewed 10131 times

B2H_MXV1.gif (441.36 KiB) Viewed 10131 times

B2H_MXV2.gif (1.06 MiB) Viewed 10131 times

And the first prototypes (huge thanks to Andreas for printing these first versions and to the entire Matt3o discord server. You guys helped a lot !)

Spoiler:

IMG_5114.jpg (1.87 MiB) Viewed 10131 times

IMG_5119.jpg (2.87 MiB) Viewed 10131 times

First sound test:

https://youtube.com/shorts/FmeM5_U31jc?feature=share
https://youtube.com/shorts/0Wnihkx0xhE?feature=share

As you can see it's really just a very early proof of concept and there is a lot of stuff to fix. But it works! Kinda at least.. For now I have to use pliers because the magnets I ordered are a bit too much for the weak cherry mx red spring to handle . Nonetheless, the click sound and switch feel is very promising for a first version. I put this project on hold for the last month or so due to having more on my plate than I can chew, but I figured I might motivate myself to continue working on it by sharing the progress here. I'd be also happy to upload some of the files if anyone wants to join the project!

[thefarside]

Awesome work! Very interesting project. Please keep us updated on your progress!

[Bjerrk]

So ... This is brilliant.

[mmm]

Very elegant design - and nice illustrations!

I know this is just a proto-type, but I'm a bit curious about 3d printed switches - how are the friction between the slider and housing?

[Muirium]

Why not have the magnets on the chunky side of the shell, where MX has its switch contacts, so you can retain the LED mount?

I guess this is not a contact (and therefore PCB compatible) switch anyway, but I’d put the moving parts at the larger side of the shell, if imitating MX’s esthetic.

[browncow]

hey, this looks really interesting, but am i correct if i see contact plates here, or do you plan or making this a hall design?

[Muirium]

This pic appears to have an MX black shaped peg on the left to actuate standard MX contacts (which are otherwise not shown):



You’d feel those scuffy contacts, but. Either go non-contact or go home!

Magnet activated reed switch would be more appropriate I think, and still compatible with MX PCBs. Actuating with the fly magnet would be interesting.

[zrrion]

>Either go non-contact or go home
feature bloat
getting the simplest switch out the door as soon as possible is way better than never getting out the perfect switch made. (looking at you silo switches)

[Muirium]

Still at the concept stage, though. Now’s the time to be smart about not painting yourself into a corner later.

As there’s already magnets in this switch: the reed is calling!

[Lalaland124]

Thanks for the awesome feedback guys! Regarding the switch contacts: I forgot to add them in the animated video. However, I honestly haven't thought about it too much. I wanted to keep it simple which is why I went with the standard mx red contacts. Quite literally you could use the exact same ones cherry uses and the click should still happen at the same time the contacts close. So yeah, like zrrion mentioned the goal was to make them as easy to manufacture as possible.

Having said that, I do get your point Mu Although I hardly ever use the led's on my keyboard I know people love em so why not try the Hall IC version as well and thus leave some room for led support.

Concerning the friction between the slider and the housing in the 3d printed model: It's honestly not as bad as I expected. I mean it's not even close to what we're used to, but the prototype definitely holds up pretty well against ultra dirty alps and smith corona leaf springs. Though I have to say that I lubed these things up pretty well .

The main problems I have to fix now are:

a) How strong should the magnets be? (the current version feels quite literally like flicking a light switch on and off)

b) Can you make standard mx springs non-magnetic?

c) Can you have led support with the current design?

d) Some of the walls inside the switch are too thin and are prone to breaking - gotta fix that.

and that's pretty much it. I probably have to invest in a good 3d printer, as this would make things much easier and would allow me to have better feedback loops. Also, if anyone is an expert in physics (I'm looking at you Bjeerk ) or to be more precise - in everything that has something to do with magnets and knows how to determine the type of magnets needed to ensure the spring is still strong enough to separate them again - please let me know . I tried to weaken them by putting them in my oven but that didn't go too well lol.

One thing I wanted to add is that the (45SH NiCuNi-) magnets I used were only 5 x 4 x 1mm and 3 x 3 x 1mm in size and their total travel inside the housing was only ~1-3mm! This is why I was so surprised at how audible that click was.

[Bjerrk]

Completely disagree with Mu, don't let some notion of perfection ruin something great. Also, it's your thing, don't let Mu or myself tell you what to do.

[Muirium]

For I am the Devil!

Actually, I'd like to keep this switch *simple* with less moving parts than MX. A reed switch is a profoundly simple thing. Here's a picture of a great big one to make it easier to see:



They're a sealed package. A magnet moves close, attracts the contacts, and the switch actuates, as if by magic, without needing pushed. No contact noise—no Mattttias chatttter—and no scuffing in the key feel. Best of all: they're MX PCB compatible if you put the pins in the right place. A reed switch needs no fancy schmancy controller: they're just a regular dumb contact switch, as far as the keyboard's concerned. But with greater reliability and better keyfeel.

Magnets are the core of this design. Use them to trigger a neat little reed switch and you can ditch the contacts entirely. Simpler and better!

[Muirium]

a) How strong should the magnets be? (the current version feels quite literally like flicking a light switch on and off)
…
d) Some of the walls inside the switch are too thin and are prone to breaking - gotta fix that.

I have an idea which may address both of these: that wall between your magnets? How about making it a detachable metal plate? Just thinking aloud here, but maybe a regular ferrous metal could be a good way to 'dial in' a desired tactility? I'm guessing steel would really weaken those magnets, while something non-magnetic like aluminium might have a much subtler effect.

Well, assuming the magnets don't just stick to the damn thing!

Anyway, your prototype's tactility is way too high: so those magnets need to go further apart. Flip them around to the fat side of the switch shell, and maybe a thicker plastic divide will give you the tactility you desire… as well as a good spot to place a reed switch.

[Lalaland124]

Does a hall ic sensor with cherry pcb compatibility exist already? Because the good part about the other design is that I wouldn't have to worry too much about manufacturing as all the complex parts are already being made in that exact same way I want them to be. I guess it would be incredibly helpful to get somebody on board who has some experience in switch manufacturing/design. Didn't someone on DT design a switch before and then sent it to Kailh or similar to make it? If so, please reach out to me !

I totally get your point Mu and if it turns out that it not only would be more sensible but also cheaper and easier to manufacture the switch like you proposed, I'll definitely go for the reed design!

Completely disagree with Mu, don't let some notion of perfection ruin something great. Also, it's your thing, don't let Mu or myself tell you what to do.

And don't worry Bjerrk - perfection won't bring this project down. Like a smart designer once said: I embrace imperfection as much as I embrace the pursuit of perfection.

[Muirium]

I'm wary of Hall Effect because, as far as I understand, it has many of the same costs as capsense: it's inherently analogue, so it needs a whole different kind of controller. It does let you do fancy stuff like dynamically varying the actuation point, but at cost of complexity and indeed delays.

Keep it simple! Reed switch is Hall Effect's dumber, cheaper, simpler, binary brother. It just needs a magnet, and you've already got that.

Hataa and XMIT are the two people I can think of who've worked with manufacturers on their own switch designs.

[Lalaland124]

I'm wary of Hall Effect because, as far as I understand, it has many of the same costs as capsense: it's inherently analogue, so it needs a whole different kind of controller. It does let you do fancy stuff like dynamically varying the actuation point, but at cost of complexity and indeed delays.

Keep it simple! Reed switch is Hall Effect's dumber, cheaper, simpler, binary brother. It just needs a magnet, and you've already got that.

Hataa and XMIT are the two people I can think of who've worked with manufacturers on their own switch designs.

Ahh you're right - it was XMIT's project I was thinking about.

Gotta give it to you Mu - you make it sound very simple (and compelling) to go with reeds. So if I got this right you propose putting the magnet and the reed to the side where currently the traditional cherry contacts are and thus creating room for the led to fit, right?

[Muirium]

Yes. I'd like to ditch the bulky Cherry contacts entirely, and put the magnets on that larger side of the switch; and a reed switch to detect their motion. But it's your baby. There's probably consequences I'm not thinking of: like where you're sourcing / how you're modding the slider for example.

Electrically speaking: all you need is a reed switch. Hook it up to the pins and you're done, as far as the electrons know.

[xxhellfirexx]

Just my two cents, but maybe the sliding magnet can be moved into the stem to make space for LEDs on both sides of the switch. Having the magnet in the middle of the switch will give a more even tactile feel. The hall sensor can be placed in front or behind the slider or underneath.

[Muirium]

Nice idea to use that central channel for the freely moving magnet. Can’t quite visualise how the other magnet is moving it around, though.

[Lalaland124]

Just my two cents, but maybe the sliding magnet can be moved into the stem to make space for LEDs on both sides of the switch. Having the magnet in the middle of the switch will give a more even tactile feel. The hall sensor can be placed in front or behind the slider or underneath.

That was one of my first ideas as well - it would create a lot of space which would be awesome. Only problem is that this part of a cherry mx slider is really really tiny. I can'T imagine a way to fit a reasonably sized magnet in there that would also generate a nice sound. The other thing is that this would certainly manipulate the spring of the switch as well, but that's a problem you'll have with every design as long as you use the current springs from cherry..

[Patrick28_]

Has any progress been made with these switches after your last post? It seems to me that this project has a ton of potential, as contactless clickies could be exceptionally durable and smooth switches, if the magnetic reeds were to be included in the final product, that is. Although, even if the reeds weren't part of the design, I still think it would provide a super unique tactile experience that's different to what many modern designs have to offer.

Oh and another thing: how flexible do you think this design is for modifications down the road? For example, maybe restricting the movement of the magnet to make it tactile, and maybe even making the bump more of less rounded by making the now restricted magnet taller or shorter. (Full disclosure: I have absolutely no experience with magnets or magnetic switches, so these are only some potentially possible ideas)