First thing to do: Find out about the original protocol.
The vintage PC pages state that the keyboard talks with a USART, at 1200 baud with 8 data bits, 2 stop bits and even parity.
By looking at the PCB and using the pinout of the 7407 chip, the connector pins can be identified (the only line connected to the controller IC should be the data line)
(looking at the plug):
'
4 DAT o o 1 VCC
3 GND o o 2 reset
+ o
Luckily, I had a FTDI serial-to-usb breakout board as well as a DIN connector lying around, so I hooked up the data line to the "RXD" pin of the ftdi chip:
And voilą! Connecting with minicom, I can receive single ASCII characters as I type. Wait... characters?
It seems this keyboard just sends individual characters via serial interface. Repetition is done by the keyboard itself. The CTRL-key doesn't seem to do anything, whereas the shift keys work, but don't cause any ascii or keycode to be sent.
The LOCK and GRAPH key do send some non-ascii keycodes though...
Anyway, this interface is unusable. It does not provide a pressed/released state of keys and so writing a converter would involve simulating key releases etc. Therefore, it seems better to repleace the controller altogether!