- ADDS-Viewpoint-case.jpg (272.9 KiB) Viewed 2842 times
- ADDS-Viewpoint-back.jpg (1.31 MiB) Viewed 2842 times
But looking closer at the PCB, I can see that one of the signals just goes to a screw, so it's a chassis ground. The keyboard can't be output-only, because it has a buzzer and the DIP switch settings listed there confirm that one could enable key clicks. Furthermore, the buzzer itself is 12V, so the power isn't trivial, either.
So, I got it for reasonably cheap, half shipping. And dealt with the foam and foil.
- ADDS-Viewpoint-foil.jpg (472.37 KiB) Viewed 2842 times
- ADDS-Viewpoint-top.jpg (1.74 MiB) Viewed 2842 times
- ADDS-Viewpoint-bottom.jpg (1.55 MiB) Viewed 2842 times
- ADDS-Viewpoint-converter.jpg (429.21 KiB) Viewed 2842 times
After fooling around with having the host pull it low for a while, I found that a single pulse would cause it to toggle the line a few times on its own. And that multiple pulses would cause it to make noise, which turned out to be a quite loud alarm kind of sound. It also makes this for a few seconds when 12V is cut, as some capacitors discharge.
The most noticeable thing about these pulses was that the low segments were sometimes short (26µs) and sometimes long (126µs) and there seemed to always be eight of them. The ESC key in the upper-left generated long-long-short-long-long-short-short-short. Which is ASCII ESC (1B) in LSB-first order.
And, likewise, the 1 key next to it generated 31. So, it was indeed ASCII, but with a funny bit-serial protocol.
So I whipped up a decoder for Pulseview.
- vpt-1-pulseview.png (46.53 KiB) Viewed 2842 times
- vpt-1-2-pulseview.png (47.69 KiB) Viewed 2842 times
2 3 4 6 Q W Y A Z M ` [ ] \ ; '
I don't think those are a single row in the matrix.
It's almost as though the keyboard were in some kind of raw code + shifts mode. But it's completely reliable for just those keys and no others. And there doesn't seem to be enough room to do all the keys that way. Moreover, C4 is used for LOCK and CD-CF for F1-F3.
There is a similar Viewpoint 60 and Bitsavers has the firmware. It seems to be roughly similar. Even though it's an 8035 instead of an 8021, it doesn't use the INT line but does everything with T1. Nor does it enable the timer interrupt. Without working it all through, I don't see anything in there peculiar about the mapping of some keys. All the ASCII characters, unshifted, shifted, and ctrl-shifted, are right there as expected.
This keyboard was clearly really used at some point. And it's a masked ROM not PROM.
It's easy enough to build a sketch to correct for the anomaly and get a consistent virtual serial port.
I feel like I'd really need to compare with another keyboard of this kind to further determine whether this weird behavior is specific to this board.
Also, I don't think this is related to the "ADDS Protocol" that ADDS ASCII terminals speak. I know everyone just sets those to XT-mode and uses Soarer's. But I can't find any information on what the alternative looks like. I had thought to get one from elecplus, but they've had to shut down temporarily due to the pandemic.