Page 1 of 1


Posted: 04 Jul 2020, 18:37
by MMcM
I got this 1976 board even though it has an obvious problem.
VT50-top.jpg (399.94 KiB) Viewed 1342 times
Now I'm on the lookout for a Hi-Tek E key or else having to use un corps noir tranchant un flamant as a typing test phrase. Other than that, it cleaned up nicely: only a few scratches and a couple switches with broken corners, as you'd expect.

Of course, the first thing that stands out are the iconic red and blue key caps. On this one, the blues have noticeably yellowed and the space bar is pretty much all the way to teal.

But there are other noteworthy things about these keyboards.

The VT52 family of terminals had an integrated keyboard. That gives the designer some latitude as to how the main logic board and the keyboard communicate. At one extreme, they can just do the same as if there were an external cable and only send keys when pressed, maybe even using a UART. At the other extreme, the keyboard module can house just the switch matrix and the terminal can do all the scanning.

The VT50 takes an intermediate approach. The matrix itself is abstracted away, but the terminal still does the scanning, which makes sense, since it's all discrete logic with the timing driven off the vertical retrace. To do this, there are a decimal decoder and a selector on opposing sides of the switches. To the terminal, it appears that a single key code number is entered into a register and the state of that key read back. Loop a counter doing that to scan for key presses. Three bits select column and the rest select row. To avoid ghosting, one side goes through octal NAND gates rather than being shorted together, though the terminal logic never tracked more than two keys down.

On the VT50, the BREAK and CTRL keys are pulled out as separate active-low signals. The VT50H added a number pad and so needed another couple rows. The CTRL signal was therefore repurposed as a seventh key code bit, and the key itself moved into the matrix in the ordinary way. The key codes are all changed as well, even for the common keys, but the decoding principle is the same, just with a 16-way 74150 instead of an 8-way 74151. This design was kept for the better-known VT52 and the VT78 (which adds a PDP-8 inside the case). Some jumpers were also added to make BREAK part of the matrix and CAPS LOCK separate; these were then configured that way on the VT61 and VT62, which also had a wider (16 instead of 11) connector.

The blank keys don't do anything on this keyboard. The three on the right aren't connected on the PCB. The switches are still there and soldered in, because the whole grid was OEM'ed as a unit. But there aren't any traces from them.
VT50-bottom.jpg (494.59 KiB) Viewed 1342 times
The trace on the locking key just put it in line with the two SHIFT keys, so it was a shift- and not caps-lock. But the copper has been cut. I thought maybe this was peculiar to this example, but nope, that mod is shown in the print set.
VT50-drawing.jpg (589.01 KiB) Viewed 1342 times
VT50-drawing-zoom.jpg (27.75 KiB) Viewed 1342 times
This maybe makes more sense when you remember that the VT50 terminal did not have lowercase.

The prints also reveal some renaming. The SCROLL key is called PAGE and the COPY key PRINT. Also, apparently the optional, and now coveted, "gold key" on the VT78 was called the "super key."

The QMK code should support a VT50H / VT52, though I don't have one to test. Likewise, the converter board, which I gave a separate connector for whether the eleventh signal was a second extra key or a seventh key code bit. Also note how the jumper colors have the same RMA digit order on the old DEC board as some I had from Adafruit.
VT50-converter.jpg (490.04 KiB) Viewed 1342 times