At some point everyone was talking about / showing off their mech keyboard in developer scene. I don’t think I’ve seen much in recent years.
I myself went deep into that for a while. Got a couple of keyboards and now I have two Apple Magic Keyboards. Don’t even know where I stashed my mechanicals.
I want a keyboard switch with a weight on the end of a lever, typewriter or piano style. Or some other mechanism whereby the resistance would be constant or even reverse-linear-ish (from gravity and momentum), not linear (from a spring). But as far as I know no such thing exists. :(
I don't know how their switches worked, but the Wang 2200 terminals¹ that my father worked with had an interesting angle on tactile feedback; on each keypress a single chunky solenoid attached to it physically moved to give a satisfying "chunk" noise and vibration.
The idea presumably was to give solid mechanical feedback to professional typists used to the same from electromechanical typewriters throwing the type arm onto the platten.
Note this was late 70s/early 80s so I may be confusing/conflating it with other machines.
Does anyone know why alanlog optical switches never gained traction, but analog magnetic ones did? Sounds like optical ones should be cheaper to manufacture, with all the same benefits?
Minor physics nitpick in the article "...and a gradually widening conductor lets the inductance increase proportionally with key travel instead of jumping abruptly as the metal traverses the field. As the key is pressed, more of the cone moves into the field produced by the coil, inducing eddy currents in the metal. Following Lenz’s law, those currents oppose the original field, which increases the coil’s effective inductance..."
L, the inductance, is reduced, not increased due to insertion of a conductor, unless the conductor is ferromagnetic. A non-ferromagnetic conductor will expel flux due to generated eddy currents, lowering flux-linkage, therefore L, assuming driving coil current is held at a steady rms magnitude.
My general experience with Cherry style mechanical key switches is disappointment. In my almost 40 years of computing they are the only type of key switch that have consistently given me issue. I've owned at least five boards at this point with them, and they all eventually have issues.
They're like owning a sports car, you have to get used to opening them up and cleaning the contacts, desoldering switches, oiling stems. They're just too high maintenance.
I gave that life up when the P key stopped working on my WhiteFox mid outage and I had to frantically switch keyboards.
My daily driver for the last five years has been a rubber dome Sun Type 7. It has given me zero problems, no one complains about the noise, it's got that so ugly it's cool "retro chic" thing going even though I bought it new direct from Oracle.
I still have multiple IBM buckling spring boards from when I was a kid and none of them have ever given me an issue.
This was a fun read. Hall effect switches especially feel like they came out of nowhere in the last couple years, but seeing them broken down like this makes the hype make more sense.
That's mostly copycats of Ben Heck approach from 2012 - some technical improvements, but like without understanding, further thinking about function - or imagination why someone needed it first, then what would be next. So all seems stuck for me, and not much than a bit a closer since then to what (and how) I want, starting from smoothly increasing the repeat rate with pressure for cursor keys or scrolling (but not to have few levels only to macro-simulate that, then OS's are used to not get such things right - High Resolution Scrolling https://www.youtube.com/watch?v=FSy9G6bNuKA 2024).
The only thing I don't like about this article is it didn't talk about the actuation tech. It basically only showed switches with springs. My Void Switch design doesn't have springs at all!
I'm typing on my AHEK-95 right now which has 95 Void Switches and a custom hall effect PCB (and custom firmware). It also has two custom analog rotary encoders (which is just a ring of six magnets over two hall effect sensors; the firmware is what figures out what direction the ring moved).
I sent one of my AHEK-95 keyboards to Chyrosran22 (known for his brutal keyboard reviews) and he reviewed it:
One other thing missing in the article: Hall Effect sensors are temperature sensitive! My AHEK-95 has a number of features in the firmware to work around this (constant re-calibration) but anyone with other Hall Effect keyboards might have experienced it: Some keys seem to stop responding reliably after a while (usually a week or two in winter). This is because the temperature changed (enough), causing sensor drift. The fix is to just reset the keyboard (<1 second) which forces a recalibration but it can be annoying (my father in law has a Wooting keyboard that suffers from this after a few weeks if there's big temperature fluctuations at his desk which is right next to an exterior door).
Anyone with a 3D printer should print my Void Switches! It's a teeny tiny amount of filament (use PETG for best results) and only requires some 4x2mm magnets. I uploaded a model to Printables that has all the parts in one 3mf file: https://www.printables.com/model/233699-void-switch-fidget
Aside: Last time I ordered 4x2mm magnets, I ordered six thousand from AliExpress. That was after my last order of five thousand that I blew through quite quickly :D. I highly recommend ordering them in bulk like this because they end up costing about a penny each and they're useful in so many 3D printed things.
We've printed your switches at my university, they feel really great. Now we need to build a full keyboard in order to try out the typing experience, when used inside a keyboard.
Do you think the PCB and typing assembly could be made curved, like the assembly of a Model F or M keyboard?
I’m making my own switches for a musical instrument (the dimensions don’t work well for mechanical keyboard switches) and decided on capacitive sensing because it’s cheap on a per-switch basis. Making things with copper tape is annoying, though. Using the spring as the capacitive element sounds promising.
14 comments
[ 2.7 ms ] story [ 35.2 ms ] threadAt some point everyone was talking about / showing off their mech keyboard in developer scene. I don’t think I’ve seen much in recent years.
I myself went deep into that for a while. Got a couple of keyboards and now I have two Apple Magic Keyboards. Don’t even know where I stashed my mechanicals.
The idea presumably was to give solid mechanical feedback to professional typists used to the same from electromechanical typewriters throwing the type arm onto the platten.
Note this was late 70s/early 80s so I may be confusing/conflating it with other machines.
¹ https://en.wikipedia.org/wiki/Wang_2200#/media/File%3AWang22...
L, the inductance, is reduced, not increased due to insertion of a conductor, unless the conductor is ferromagnetic. A non-ferromagnetic conductor will expel flux due to generated eddy currents, lowering flux-linkage, therefore L, assuming driving coil current is held at a steady rms magnitude.
They're like owning a sports car, you have to get used to opening them up and cleaning the contacts, desoldering switches, oiling stems. They're just too high maintenance.
I gave that life up when the P key stopped working on my WhiteFox mid outage and I had to frantically switch keyboards.
My daily driver for the last five years has been a rubber dome Sun Type 7. It has given me zero problems, no one complains about the noise, it's got that so ugly it's cool "retro chic" thing going even though I bought it new direct from Oracle.
I still have multiple IBM buckling spring boards from when I was a kid and none of them have ever given me an issue.
(* XBOX Controller Mods: Analog WASD Gaming Keyboard https://www.youtube.com/watch?v=gEwDImE0DU4 2012 *)
https://github.com/riskable/void_switch
I'm typing on my AHEK-95 right now which has 95 Void Switches and a custom hall effect PCB (and custom firmware). It also has two custom analog rotary encoders (which is just a ring of six magnets over two hall effect sensors; the firmware is what figures out what direction the ring moved).
I sent one of my AHEK-95 keyboards to Chyrosran22 (known for his brutal keyboard reviews) and he reviewed it:
https://youtu.be/iv6Rh8UNWlI?si=9xGNm4jIDLFzx80c
One other thing missing in the article: Hall Effect sensors are temperature sensitive! My AHEK-95 has a number of features in the firmware to work around this (constant re-calibration) but anyone with other Hall Effect keyboards might have experienced it: Some keys seem to stop responding reliably after a while (usually a week or two in winter). This is because the temperature changed (enough), causing sensor drift. The fix is to just reset the keyboard (<1 second) which forces a recalibration but it can be annoying (my father in law has a Wooting keyboard that suffers from this after a few weeks if there's big temperature fluctuations at his desk which is right next to an exterior door).
Anyone with a 3D printer should print my Void Switches! It's a teeny tiny amount of filament (use PETG for best results) and only requires some 4x2mm magnets. I uploaded a model to Printables that has all the parts in one 3mf file: https://www.printables.com/model/233699-void-switch-fidget
Aside: Last time I ordered 4x2mm magnets, I ordered six thousand from AliExpress. That was after my last order of five thousand that I blew through quite quickly :D. I highly recommend ordering them in bulk like this because they end up costing about a penny each and they're useful in so many 3D printed things.
Do you think the PCB and typing assembly could be made curved, like the assembly of a Model F or M keyboard?
Has anyone else tried this?