First experiments in CNC

[Muirium]

Mmm… homemade switches! On the horizon, maybe. They'd be expensive, but they'd be whatever we want them to be.

Same for caps.

[cookie]

Switches would be the ultimate goal, but I'd like them to be something noone have done before.
Just overall right and nice

One day maybe...

[Muirium]

Playing with beam spring switches lately, I'd like to shrink them a little (if possible) and put an MX mount on top. Believe it or not: the PCB is already possible and Xwhatsit's controller has us covered. Now that would be a dream board…

[macmakkara]

lets start planning then

[matt3o]

I sold all my Model Ms, if someone provides one I'll work on the BS to MX slider

[cookie]

Beam Springs are too huge for my taste. we need something totally new!

[andrewjoy]

If i was to design a switch i would have a capsence controller like ibm but i would have a slider nice and wide so its stable

it would move inside a barrel similar again to ibm but larger and with the top interchangeable so you can use an adapter for most styles of cap

i would find a way to detect the change in capacitance not from a pcb below but from the barrel, and get it so the further you press it down the greater the change in capacitance, that way you would be able to set the actuation point yourself

there would have to be a spring under the slider to return the key but as its not part of the switch it would be easy to change

if the only thing that moves is the slider and there is no physical switching mechanism it would be more reliable

don't know what i would to about a tactile bump or click but i guess i could use a solenoid clacking on nice thick metal case

[matt3o]

the click should activate on actuation point though. more complexity if actuation point is variable

[shamanas]

I'm surprised there are no Buckling Spring clones (or similar switches) out there, I only know of the Ya Sheng ones.

As for the actuation point, it should be ralatively simple to make it variable, no?
I mean, by measuring the electrical charge due to capacitance for many distances you should be able to plot a pretty good graph and then use an approximate formula in the controller software, allowing you to change the actuation distance from there (of course you add a key combo to do this on the Fn layer, or maybe a volume wheel type thingy).

The Noppoo EC108-Pro actually lets you chose between 2mm and 3mm actuation distance (I think with a +-0.2mm error margin) but of course setting it to an exact value would be more difficult.

[andrewjoy]

the click should activate on actuation point though. more complexity if actuation point is variable

that the beauty of the solenoid it can be fired by controller when key actuates if it hits the plate the sliders are in you get tactile feedback too

[rklm]

and this is a keyboard case, not yet 100% finished, but I like it.

Spoiler:

It's acrylic, but the plate is 3mm with little pockets for the switch clips

Spoiler:

Some edges seem rough but that's just some waste material to be removed and acrylic dust/powder.

(bottom is glossy, top is matte)

Spoiler:

the plate is 3mm thick but it goes 1.5mm deep into the sides.

Alright Matt... You have me wanting to buy a CNC machine now!

I love that 3mm plate design!

What do you think of this socket design?

https://tinkercad.com/things/1cBShmYHabs






It is about as good a negative of Cherry MX as I can conceive of.

Also, if you don't mind me asking, what kind of software are you using to generate the gcode and what is the size of the cutting head of your CNC?

[Muirium]

Impressive! Send Matt your files…

the click should activate on actuation point though. more complexity if actuation point is variable

that the beauty of the solenoid it can be fired by controller when key actuates if it hits the plate the sliders are in you get tactile feedback too

Yup. The only downsides from a solenoid are: extra power draw (not great for using with tablets and such) and limited fire rate, so if you mash a lot of keys the illusion breaks down as the solenoid can't keep up. Well, unless you have several working in parallel!

Believe it or not: linear capsense is a piece of cake. Capsense is definitely the best system for custom switches as it's inherently NKRO without diodes, and inherently analog so we can dial in our own activation point or even MEASURE VELOCITY FOR OH LETS SAY ANGRY CAPS! No need for a PCB (and the fixed layout it brings) as the sense pads can go inside an integrated switch, too. Even the power draw is close or equal to a matrix of MX switches and the diodes they need.

But we want Topre-like tactility, right? Or buckling spring-like click? Or even beam spring-like both of the above! That will take some inventiveness in the mechanism, for sure. But I'm pretty serious we could make our own beam-spring inspired switch. I'd say it's easier than Topre. More modular too. We'd need to cut springy metal sheets and master the art of manufacturing the plastic shells with sliders and internal mounts. All of which I'm ignorant about! Seems quite feasible though, with enough effort.

[matt3o]

Alright Matt... You have me wanting to buy a CNC machine now!

I love that 3mm plate design!

What do you think of this socket design?

the design is very nice, but a bit too complex.

Consider that the bigger the endmill you use, the faster and deeper you can cut. Similarly, the smaller the bit the slower you cut. So, the trick is to find a way to use big endmills as much as possible. For my design I use 3mm and 1.5mm end mills, and I can have a plate in 15 minutes.

Your design probably requires smaller bits and more machine time. Not that it cannot be done, but it definitely costs more (in time, electricity and bits wear off). You should try to simplify it as much as possible.

Also, if you don't mind me asking, what kind of software are you using to generate the gcode and what is the size of the cutting head of your CNC?

I use madcam.

But we want Topre-like tactility, right? Or buckling spring-like click? Or even beam spring-like both of the above! That will take some inventiveness in the mechanism, for sure. But I'm pretty serious we could make our own beam-spring inspired switch. I'd say it's easier than Topre. More modular too. We'd need to cut springy metal sheets and master the art of manufacturing the plastic shells with sliders and internal mounts. All of which I'm ignorant about! Seems quite feasible though, with enough effort.

I would be perfectly fine with a cherry-MX compatible buckling spring that doesn't need those nasty PCBs.

[Muirium]

So long as it feels like Model F. Even if we cheat!

Just imagining DSA or SA caps on buckling spring. Ho goodness!

[jacobolus]

Why don’t you just make spherical replacement buckling spring keycap tops? Trying to make adapters for MX caps seems like a fool’s errand, as does trying to engineer your own switches from scratch.

[matt3o]

we did adapters for alps and topre, don't see why we shouldn't try for buckling spring.

[rklm]

the design is very nice, but a bit too complex.

Consider that the bigger the endmill you use, the faster and deeper you can cut. Similarly, the smaller the bit the slower you cut. So, the trick is to find a way to use big endmills as much as possible. For my design I use 3mm and 1.5mm end mills, and I can have a plate in 15 minutes.

Your design probably requires smaller bits and more machine time. Not that it cannot be done, but it definitely costs more (in time, electricity and bits wear off). You should try to simplify it as much as possible.

I revised my design (getting rid of the angles around the bottom and rounding it out so a 1.5mm bit would fit):

Link to project


I will try recreating it in the legit software later.

I noticed the corners of the mounting holes in your plate are round, are the squares still 14x14mm? If so, do the round corners pinch the switch?

I don't like how the corners look in my design (I tried to give it clearance in every way )

[jacobolus]

we did adapters for alps and topre, don't see why we shouldn't try for buckling spring.

How well do the Alps adapters work? From just looking at pictures, my guess is that they’re nowhere near as nice as just using native Alps caps.

The Topre adapters you’re talking about are just brand new sliders, right? How well do those work? Hypersphere’s review of the CM Novatouch suggests that this is really hard to get right, even with a real company behind it who can afford tooling etc.: http://deskthority.net/keyboards-f2/cm- ... t8687.html

[matt3o]

I revised my design (getting rid of the angles around the bottom and rounding it out so a 1.5mm bit would fit):

Definitely better, but it can be probably improved even more. The rounded corners are not a big deal, so you can go straight on the corners.

The Topre adapters you’re talking about are just brand new sliders, right?

http://deskthority.net/workshop-f7/topr ... t7437.html

[jacobolus]

The Topre adapters you’re talking about are just brand new sliders, right?

http://deskthority.net/workshop-f7/topr ... t7437.html

Yeah, I’ve seen that. The important bit is Muirium’s unanswered question: “The video suggests the new slider isn't as slippy and free moving as the originals. Does this affect key feel much?”

Again, I’m sure it’s possible (the Novatouch exists after all). It’s just pretty non-trivial to get right.

I think a buckling spring adapter is going to be even more difficult. (And probably less useful.)

[matt3o]

the slider went through various revisions. the last one is basically perfect (better than novatouch), I should just make it 0.5mm shorter, but I'm out of funds Also remember that topre sliders should be lubed (the originals are lubed).

[matt3o]

Radioactive keycap, glows in the dark!

[macmakkara]

Wow nice! How did you do that?

[matt3o]

I made the prototype with the CNC, then silicone mold. Then cast in resin (with glow in the dark powder). Painted black and finally scratched the surface of the logo with CNC again. Pretty complex process actually.

[snoopy]

You should make nice novelty caps for the M or beamspring. That would be awesome.

[Muirium]

I'd love to see that! But he's an MX man these days…

[Laser]

Looks nice!
Is it possible to have a top cover in HHKB style ?
(to cover bottom-left switch hole and the last two bottom-right switch holes)

[matt3o]

Looks nice!
Is it possible to have a top cover in HHKB style ?
(to cover bottom-left switch hole and the last two bottom-right switch holes)

any layout would be supported.

[Laser]

Then, this is great!
Only two questions remain: when - and - how much?

[Halvar]

I made the prototype with the CNC, then silicone mold. Then cast in resin (with glow in the dark powder). Painted black and finally scratched the surface of the logo with CNC again. Pretty complex process actually.

Awesome stuff. The fact that you can scratch out a curved logo looking that smooth is amazing.

Can you make a working key cap from your silicone mold, I mean a cap with a working mount?