I have this fantastic 1985 IBM 6770 System/40 Quietwriter. The keyboard is a proto-model-M with an actual screen on it. I would like to build a converter to use it as a modern keyboard (with proper screen support, of course!).
(I did this massive teardown and demo here:
)
Poking at the keyboard protocol with a logic analyser, it's... weird. I was expecting something like bidirectional AT, but it's not. All the data is expressed on a single wire (plus a reset line). It looks like the base station sends a high-frequency signal which either triggers the keyboard into reporting its state or announces to the keyboard that the base station is about to send. There's a constant stream of short packets, which suggests the protocol is polled. Here's a sample trace of one such packet:
Just to make life even worse... none of the chips on the keyboard PCB are identifiable! I wonder if IBM has anonymised them? The fact that two chips have consecutive numbers despite being from different manufacturers suggests this.
Before I start writing trying to figure out how this works from base principles, which I'm honestly not really looking forward to... does this look at all familiar to anyone?
(BTW: there's some more details on a superuser.com question here.
https://electronics.stackexchange.com/questions/651458/what-is-this-single-wire-bidirectional-communications-protocol They weren't able to help.)