Why don't we re-create the buckling-spring switch?

[trauring]

People always talk about how great the buckling-spring switches that IBM designed decades ago were so great, and about how little Unicomp, who bought IBM's keyboard factory from Lexmark after Lexmark bought out IBM's PC business, has done with IBM's designs. The patents surrounding the keyswitches have long since expired (Model F - http://www.google.com/patents/US4118611 and Model M - http://www.google.com/patents/US4528431).

Why not just create new buckling-spring switches? The DataHand project has had to re-create its switches from the ground up. These keyswitches were originally designed in the 1970s and early 1980s. How complicated can the technology be?

One problem, of course, would be keycaps. Unicomp sells whole sets of keycaps, but that would be limiting. We would need to either develop a new source of keycaps that fit the switches, or design the switch in such a way that it fit existing keycaps – a buckling-MX hybrid switch. That's a lot more complicated because from what I see of the design, the spring reached all the way to the top of the keycap in the original design (and it was connected on the sides, not in the middle). That could mean it would impossible to replicate the full feeling of the buckling-spring switch, although it doesn't mean what we end up with would be bad. Obviously if it isn't any better than an MX-style switch, it isn't really worth it, but it would still be interesting.

If we could create such a hybrid switch, I imagine it wouldn't be too hard to find an OEM to manufacture them.

Probably just a crazy idea. Anyone interested?

[E TwentyNine]

https://geekhack.org/index.php?topic=67646.0

I'm sure there are other threads on it, sooner or later someone will do it.

[ideus]

There are a few threads on the subject already. It would be good to search for them for the interested parties.

[fohat.digs]

It might help a lot if you disassembled a buckling spring keyboard and a keyboard with discreet switches to truly understand what they are about.

I have put forth the argument that buckling spring assemblies are not switches at all, but rather mechanisms to activate switches.

[regack]

wcass just posted this a few days ago... go vote for it! https://geekhack.org/index.php?topic=69530.msg1666924#msg1666924

[dorkvader]

If we could create such a hybrid switch, I imagine it wouldn't be too hard to find an OEM to manufacture them.

This is actually quite difficult.

Now it is possible to make ahybrid switch but since the MX stem is so narrow, and the buckling spring barrel must be pretty wide (even ya sheng are long) the whole assembly would have to be quite tall, about double the height of a standard switch. This would also incur some issues with wobble. (hmm alouthgh if you made an "mx platform" style slider it wouldn't have too much wobble issue and wouldn't be too hard to make a mould for.

one critical measurement to get right is the degree to which the spring is eccentrically loaded. Keeping this value at the ideal is critical for the spring to buckle precisely and reliably.

after that, you have to check on your actual switch mechanism. As fohat says, the buckling spring mechanism is just a complex actuator. The "switch mechanism" of membranes has some difficulties. IT is possible to make it NKRO by breaking out all the switch contacts to a huge edge connector and recreate the matrix on a separate place, like a lot of vintage keyboards did. Making a capacitive KB has issues as well with making the conductive flippies. A proposed solution is to adhere metal to the printed or moulded flip plate, though a better solution for industrial production is spray deposition metal similar to how mylar is made.

Another option is to do like what bloody did with it's "mx like" optical KB is to just have a flip plate with a vertical wall that blocks an optical sensor. This would probably be more expensive, but would be easier to design in other ways. It'd also have a pretty decent debounce time.

Overall there are a number of challenges that need be overcome and it'd take time, testing, and money to solve them. Then even if we do have a design made, we'd have to get a $50,000 injection moulding machine and some $5000 moulds, then we'd have to make thousands of these kbs to recoup that loss. Even convincing a chinese oem to do it for us would take a huge commitment.