Any Market For A Modern Capacitive Buckling Spring?

[bhtooefr]

And a PCB for capsense, you'd just have to connect the controller, no need to solder every switch down.

[robo]

I looked into the patents covering beamspring and buckling spring capacitive, as well as membrane, and they either were never approved, have lapsed, or expired. The beamspring design by IBM was rejected for patent here in the US. The patent that explicitly covers buckling spring membrane keyboards lapsed in 1997 with no hope of being resurrected. The patent that covers the capacitive buckling spring is more than 20 years old, which according to patent reform laws that were instituted in 1995, is the maximum term for a patent of that type that were granted before 1980. So, technically, anyone can make any of those types of keyboard and not pay anyone anything. I was quite happy when I saw that.

Very interesting. I guess what Unicomp owns, then, is the tooling for IBM/Lexmark's particular implementation of the membrane buckling spring mechanism, and for the Model M style keyboard itself. It occurs to me that it might be non-trivial to replicate the feel with an original buckling spring design using the same principle, although perhaps with 3D printing allowing rapid prototyping, this isn't such an obstacle anymore.

I think we need to suggest CM look into this, they seem pretty open and innovative toward the keybeard community

*edit: Sent off a pm to CM Bram to see what he thinks of the idea

I'm all for that, although it occurs to me that here we're talking about designing and manufacturing a switch, which is not something that CM has ever done AFAIK. They make plastic cases and do design and marketing, but they buy their switches (and boards?) from 3rd parties. Perhaps a more likely choice is a company like Kalih with experience with switch design experience who may want to differentiate themselves in the market.

This leads to the idea of a modular buckling spring switch, discussed before. I'm wondering if again 3D printing changes the game here. If the actuators and springs were readily available, then the main other physical component is the correctly shaped 'tube'. Instead of having these as discrete components, a single piece of plastic could be printed with the keys in custom locations. A custom circuit board would also be needed but that's also true of existing modular keyswitch keyboards. So just possibly the need for a modular buckling spring keyswitch is a bit anachronistic?

[bhtooefr]

Really, I think Brother buckling spring has the right approach for this, although it does complicate the PCB manufacturing (strange drill shapes through it).

The Brother switch clips through the membrane (being membrane BS and all) onto the backplate:



That's a self-contained BS module there.

[Muirium]

Ah, discrete switches! Anachronism be damned, there's a lot of things they can do so much more elegantly than non-self-contained mechanisms. Just for starters:

http://deskthority.net/workshop-f7/my-d ... t2534.html
http://deskthority.net/workshop-f7/alek ... t7090.html

Ergo customs live and breathe MX.

Besides modularity, buckling spring's other weakness is cap selection. IBM's originals are quite nice of course (I'm typing, as usual, on a set right now) but there's fantastic caps out there for MX and more coming all the time. For those of us into building our own custom keyboards, and exploring new layouts, cap selection isn't just aesthetic: it's fundamental.

I wonder if there is some way we could design around the height of a full length buckling spring and MX mount. Both of them take so much space. In fact, they are one another's antithesis. MX allows the cap stem right inside the switch, while buckling springs live right inside their cap bodies! Bridging the two with an adapter makes a hellishly tall switch.

[bhtooefr]

We really need to get low-volume cap manufacturing going, to solve that problem. Alps is another thing that needs custom caps, and it can benefit from more than just what Signature Plastics can make, too - 18 mm pitch caps are a thing that is practical.

[Muirium]

Yes, I have high hopes for CNC. Caps are right in the sweet spot for manufacturing ourselves, with the right entry-level equipment. (It's still too expensive and technical for me, but we have the right people!) They aren't as tolerance challenged as switch components, and in fact can be made from a range of materials, whatever piques our interest.

Metal caps would be a great way to start off. Let alone Alps mount, Topre mount, or IBM ones!

[jacobolus]

I’m about to sign up for TechShop, a (memberhip-gym style) “public workshop” with a bunch of machines for making stuff. The first few weeks I want to spend trying to CNC cut some keyboards out of wood, but I’m slightly curious whether it would be feasible to CNC mill some aluminum keycap molds and use their injection molding machine to make Alps/MX/whatever keycaps.

[Muirium]

Or CNC mill some aluminium caps directly! Have a wee looksie at the price, just for curiosity's sake. Because metal caps are a specialty you may well be able to sell fairly swiftly.

Alps mount is a bit tricky: it's always a bit tight, just by nature. In fact, even plain old MX takes some trial and error, as others have found. You got some switches to test against?

[bhtooefr]

The Apple //c that I have, the Alps caps are actually almost Cherry MX loose, so loose Alps do exist. And that's a ~1986 vintage keyboard that was actually made by Alps (not sure about the caps, Apple could have had someone else make them).

[dorkvader]

Hall effect, like capsense, doesn't even need diodes as I understand. So hand wiring one of those would be a cakewalk!

Custom PCBs are cool (and much easier to source than custom membranes, let alone membranes that have a chance in hell of lasting) but there's a certain solder fume induced magic to hand wired matrices. Let me tell you…

like your previous post (hall effect better than capsense in all positive respects) I feel this to be a bit... optimistic.

You see, of the 9 or 10 (or more?) hall effect keyboards I have, there are at least six that I want to convert to USB. I've spent some time considering how best to to it.

The hall sensors do have diodes in them (they have an amplifier in each sensor, as well as some other dedicated circuitry), but the best diodeless NKRO scheme is to wire the switches in rows for power and wire the output sensors in columns. Then you just power one row at a time and then sense each column for 5V to see which key (if any) is pressed. Since they "turn on" more or less instantly, you can do this. Honeywell advertises 100 KHZ scanrate in their whitepaper, which is at least 3 orders of magnitude better than cherry.

But because each switch requires 3 connections, only two of which are in parallel, wiring it would be... difficult. I have still considered it of course.

Another thing you seem to have not considered is holding the PCB on. You see, despite your aesthetic elegance, having a non-discrete switch system makes more sense for reliability reasons. I have one keyboard with a "spare" switch unit that can be resoldered in place if another fails for a 0 day turnaround in the field. The Hall sensors are removable from the switches and indeed are not held in place by anything. This allows someone to swap out a faulty one quickly.

Because of this "high reliability" design, the PCB can come off the switches. In all hall effects the PCB is either riveted to the plate (in older dual magnet designs) or screwed to the switches and/or the plate (single magnet designs). In one case of a kb posted at GH, there is no plate at all and the switches are screwed directly to the PCB, though I wouldn't recommend this. (it is, of course, only possible with the newer "single magnet" designs)

In all areas, there are tradeoffs. A hall keyboard would be really expensive, and you'd have to have a powerful microcontroller reading it, to add hysteresis (only a little is necessary, but a little is indeed necessary to prevent bouncing) and deal with the "analog" nature of hall sensors, then you need to design a switch on top of them. I estimate this will take much design, prototyping, testing, and still will cost a few dollars a switch. (or more) The downside of using vintage hall switches and sensors is that they are no longer made, and keycap availability is limited.

Since I want to have a few hall keyboards, I have bought up all the "reasonably priced" ones I could find, so that I could make a few working ones.

we both extol the virtues of hall effect switches --rightly so, but it's not all gravy. It never is.

---

Now about the gaming viability of buckling spring switches: you can "tap" them pretty quick with some practise, but they do take longer to return and reset the hysteresis loop than linear switches. On the other hand they likely won't need debouncing, since all the switch bouncing (in theory) will occur past the capacitance threshold of the switch.

Sadly, they will require some expensive tooling to reproduce the pivot plates, where there are some very critical dimensions that will prevent the spring from buckling if even a little off. And, like all vintage switches, the supply is limited.

[Muirium]

Optimism, guilty as charged!

Seeing now just how little circuitry Xwhatsit needs to make his sterling capsense controllers, and how little more is on the IBM PCB, I've become a real fan of indirect switching. Time was, its prime advantage was to make cleaner signals for simpler controllers to debounce. Nowadays, that's such a non issue that we value it instead for NKRO and smoother feel.

Hall effect intrigues me for the same reasons. But if it is more of a pain in the arse, then capsense is the way to go. There's something quite magic about a PCB which contains everything needed in itself, with the switches acting as physical mediators instead of dirty electric contacts. If Hall effect switches aren't as conducive to such abstraction on the board, then no loss, as there's no apparent downside to capsense that I know.

As for discrete switches, I think they're doable with Xwhatsit's design. He's built in enough tolerance that even his gnarly Model F test rig with "green wires" worked without a hitch.



I don't know if the same is inherently true for all capsense controllers, but I wouldn't be surprised. They straddle the space between analogue and digital, and do so well!

[quantalume]

Sadly, they will require some expensive tooling to reproduce the pivot plates, where there are some very critical dimensions that will prevent the spring from buckling if even a little off.

That's interesting and probably explains why, on most Model Ms that I've used, there is at least one switch which is not quite as clicky as the rest. I've tried switching the springs to no avail, even though nothing seemed different about the "bad" spring. Likely, there was some small difference in the pivot plate or how it sits on the rubber sheet underneath.

[stoner_cough]

We really need to get low-volume cap manufacturing going, to solve that problem. Alps is another thing that needs custom caps, and it can benefit from more than just what Signature Plastics can make, too - 18 mm pitch caps are a thing that is practical.

I believe that there should be an industry standard for the switch stems. I don't know if every company would be able to jus agree on one, or how it would happen, but it would make the customers happy, which is their goal.

[Muirium]

Topre is moving in that direction. Where "standard" = MX compatible, of course. Matias could do it too, as the only remaining maker of Alps. But buckling spring is a strange mechanism, which reaches right inside the cap. The only place to put an MX mount would be right on the top, making a very tall keyboard indeed, and likely a wobbly one.

[bhtooefr]

Matias has the same problem re: needing a very tall slider to use MX caps. Well, you could probably do an MX mount linear Alps clone without too much trouble, but the tactile leaf takes up the space necessary.

I'll note that Alps mount buckling spring has been done before (albeit shorter travel than IBM buckling spring), and Alps mount Cherry MX (even blue/green, the most complex Cherry MX other than MX Lock) has also been done before, so why not standardize on Alps mount?

[Hypersphere]

IBM had a great design with the first buckling spring keys and then made it even better with the two-piece stem and cap. Their design provided inherent standardization on all buckling spring boards because the profiles of all the keys are the same, so that the keycaps were interchangeable. Alas, only Unicomp still makes the two-piece IBM BS keys, but every time I use one of my IBM BS boards that I have reconfigured and fitted with caps to match the custom layout, I am thankful for the IBM stem and cap design.

Meanwhile, 3D printing technology is becoming more widely available. At some stage, it will be feasible to have such a gadget on every desk, and people will be able to print out a new set of keys or keycaps whenever they like. It will be interesting to see if this technology will result in more or less standardization.