Let's be serious (ly ridiculous)

[AKmalamute]

 There's a GB building for newly constructed F62s & F77s coming up, and since they'll be kits anyway, you can request just a few pieces. Like newly minted F-style barrels.

 I love buckling spring as much as the next guy but not quite as much as the guy down from him. But I have enough 60% boards (err, in the wings. two ALPs and an MX, but I don't have an PCBs and only two plates so ...) that I don't want to spend $350 on yet another keyboard. My ergodox is already housed in stainless steel, after all.

 There's the thing ... the ergodox uses column stagger, and has two separate keyboards allowing for infinite ergonomic adjustability.  That then, is my plan ... to build a buckling spring ergodox.

 Oh, I'm not the fanboy some are, but those two things I mentioned ... column stagger and separate halves, are what I want from my next high-dollar investment. BUT!

 If I'm capacitively sensing keystrokes, how can I make sure the right hand knows what the left hand was typing? I assume that kind of thing doesn't travel across modern serial I/O expanders. If I use a ribbon cable to bring all the rows & columns across, the capacitance is going to be very sensitive to the particulars of that ribbon cable, right? And could change as the cable ages, so I'd have to rewrite my firmware every six months to make it do the same thing -- each time measuring what the controller thinks is happening.

 Obviously I could just make them two separate keyboards. But then I'd need a Fn button on each, that only affected that keyboard. I might be able to put a controller on each side, and just port the decrypted data from one hand to the other, but that's the same expense just a different way of using the extra controller.

 Or if I'm willing to add diodes everywhere (oh! the expense! The time soldering!) can Model-F switches be conductively sensed, like MX et. al, are? I'm going to have to one-off my PCB either way, right? At which point a simple Teensy and 23108 combo with phone cord / audio cable / extra-USB cord will work dandy.

 I'm on the fence about this, honestly. Strictly because I can't code my way out of paper bag so the firmware has to be already available for the hardware combo I acquire. But having said that, let's say I have set myself a course. How would you recommend I connect two separate halves of an IBM-never-dreamed Model F to each other?

[CPTBadAss]

Hm, I wonder how they got the M15 keyboard to work. Maybe you can start your project by looking into that board?

[AKmalamute]

initial search implies that no, no one wants to do a tear-down of a $600 keyboard.

But from the pictures, it's a cable of stout, and specified length. I've read that the m15 didn't have a backplate, just the membrane, but I assume it's still capacitive.

 I had forgotten it existed; thanks Cpt. Of course, maybe there's a good reason it was only made for a couple of years! (But I just want to make one so ...)

[esoomenona]

.

[AKmalamute]

So, in this article, somebody uses a QProx QT160 to interpret capacitive signals, with plans to make a controller "that outputs to an i2c bus" ... so, like the 23018 does.

 This sounds like, now that touchscreens are all the rage (and believe me, I'm raging against touch screens), there are some DIY capacitive chips out there. So maybe the trick is to have a small sub-controller interpret switch state, and send the encoded data (from each hand, separately) over i2c to the teensy, which reinterprets that into scancodes to be sent down the USB wires.

 Is this idea any worse than what other folks think of when they imagine making a custom BS keyboard?

[AKmalamute]

After a bit more reading about what the XWhatsit controller does, I think the solution might be, rather than make a second keyboard (two 32u2 needed, one for each hand) a 328P could be embedded in the left hand to interpret analog data into keypress events, then pass that over i2c like our beloved ergodox, to the right hand which has a full 32u4, or Teensy 3.0 or MChck or whatever I wind up with when I get around to that part, to both interpret right hand data into keypress events, and pass all of it along the USB cord to the computer.

 I've downloaded the controller's source, and will attempt to make sense of what it's doing so I can borrow all the fancy features of TMK while capacitively sensing keystrokes.

 So, I've never seen a model-F in person, I don't think. Are the newly-made barrels going to be significantly different than old-timey barrels? I've recently seen on ebay a model-F 122 with RJ45, that shipped would be less than $100, and should by all rights provide more keyswitches than that much money sunk into new barrels. I suppose I should really plan on that, just so I can see what the inside "was supposed to look like" before recreating the wheel.

As always, thoughts? Programmers who'd like to volunteer? I can read C, kinda, but can't do much more than copy & paste and hope I get the things into the right places.