An Amkey SNK-58 keyboard.There were a number of sales like this on eBay earlier this year. I am not clear whether the seller had several of them or the one sale kept falling through after closing. Anyway, we got this one.
It appears to have been part of some kind of T-Bar brand serial test equipment. I am not sure what exactly, though they were known for similar things, such as an earlier monitor described here.
It has Amkey foam and foil switches. The PCB has the same general appearance as the terminal keyboard that Haata has.
The June 1982 date matches the date stamps on the chips.
The keys feel quite smooth. They all function perfectly, so if the foil had degraded it isn't in a way that affects operation.
Also note how the space bar stabilizer extends out horizontally from the mounting plate, again like Haata's.
The EIA Interface daughter board, with machined jumper sockets and the old-school EIA circuit names for the RS-232 pins, has no connection to the keyboard other than sharing the mounting plate. It has a separate 34-pin connector.
The controller does not have a microprocessor, though it does have 256 bits of RAM. Since these are capacitive switches, there are lots of unusual (for a keyboard) components, like an analog demux, voltage comparator, and an op-amp. The digital side is a mix of 74-series TTL and 4000-series CMOS.
Not surprisingly, given that the keys have the ASCII CTRL character names on them and there is a 20-pin connector, it's parallel ASCII with strobe and works fine with the existing program.
The character data comes directly from the two 74S287 256x4 PROM chips, one for each nibble. (The ones with labels TBar 2A/3A. There are indeed 58 keys.) These are socketed, so one could reprogram it without damage, either to get the raw scan code + shifts or to make lowercase unshifted, as right now uppercase is unshifted and lowercase is shifted. But there still wouldn't be any break codes and there aren't really enough shift keys for general use.