The Alphameric keyboard in the wiki is Type 140-1595 and not particularly similar outside or inside. So that 140 type must have been fairly broad.
Other similar codes on it, like 142-4171, 142-4172, 142-4173, might indicate subassemblies. Or they might indicate more specifically was it was known as. This particular keyboard was on eBay for some time. I am fairly certain that this find is the same one, based on the broken spacer. The price finally came down enough that even with shipping it was worth grabbing.
The immediately obvious interest is the layout. No doubt it made perfect sense with the system for which it was intended. But out of context it appears particularly quirky.
- Huge top bezel with room for some kind of branding.
- No typewriter digits.
- No Tab.
- No Caps Lock.
- No Backspace / Rub Out.
- 1.75u semicolon.
- Spacer in the typewriter section.
- Shift and Enter in the bottom row.
- Space so squeezed that it needs a legend.
- Aggressive removal of vowels in legend abbreviations (yes, there are examples for all five).
The controller is an 8749, which is consistent with low-volume manufacture. The sense circuit seems a little bit different from other foam and foil, though I didn't work out the details. I imagine the multivibrator generates the sense pulses. I could not find out what a 03G511E1 is. There seems to be a couple of Shenzen vendors with some but no datasheet that I can locate. It may be part of it, too.
The keyboard connected with a DIN-5, wired up more or less like an XT: +5 on pin 5, ground on pin 4, reset (after a 7407 buffer) on 3, 2 and 1 with a pair of open collector circuits going to INT and T0 and coming from data ports.
So the converter can be wired up just like for Soarer's or an existing one temporarily reprogrammed. The protocol is more or less XT (or clones) as well. Two start bits, eight data bits, data idle is hi-Z. The clock pulse timing is uniform.
The scan codes are completely different though, being, it seems, the switch position, since adjacent keys tend to have adjacent codes.
Since there are no LEDs, it's hard to tell whether it really supports the to-keyboard part of the protocol. It may have only needed the CTS function for when the computer was busy. The initialization sequence does not appear to be necessary, plus there is a separate hard reset signal. It might be fully supported or it might just be ignored enough not to do any harm.
It's just possible that some version of Soarer's could be configured to work with this board.
More practically, the protocol is also compatible with the TMK XT protcol driver. And the copy in QMK, which is what I used. I might wish that it were a bit more robust in the face of lost clock pulses: it's happy to wait for all ten for an arbitrary amount of time and so can get permanently out of sync. Since soft reset apparently isn't needed, it might be that a separate interrupt handler (like this) would be simpler. But given that the layout makes it kind of challenging to use as a real keyboard, I'm not sure it's worth the effort.