For the past five months, I’ve had an old IBM Model M sitting on my desk. Specifically, a 1390572, a 122-key monstrosity. Ever since spending the summer typing on one of these old keyboards I have wanted to get this keyboard working, because it is very loud and very fun to type on.

I found a converter for about $40 online, but that doesn’t seem worth it. The keyboard has this weird DIN-5 connector, where the pins sweep 240 degrees; I’ve never seen any connector quite like it. It looks a bit like a PS/2 connector, but with a pin missing, and it’s a lot bigger. Anyway, I found a mating connector and soldered some wires to it. Then, I found the pinout for this “Terminal DIN-5” connection online. It has 5v power, ground, data, and clock, which isn’t suprising at all.

I hooked up an MSP430 launchpad to it, giving the keyboard 5v power directly. Since the inputs to the MSP430 aren’t 5v tolerant, and I only planned on reading from the keyboard, I just threw two voltage dividers on for the data and clock before hooking them up to two inputs on the launchpad. I wrote some quick code to poll the clock line and whenever it went high I had the launchpad send the value of the data line over serial.

Well, both lines ended up being high by default, so I swapped my code to send the status on falling edges instead of rising edges. I pressed a key, but nothing happened. I was a little disappointed, but then I had the idea to try bringing the clock line low for a bit. When I pressed a key, my serial connection was flooded with 11 bits.

I pressed A. I was met with 10000111000. I pressed it again, and the same number repeated. I looked up a list of PS/2 scancodes. The scancode for A is 00011100. Hey, that lines up with the middle of the number. Pressing a bunch of other keys, they all begin with 1 and end with 0. The second bit seems to be a parity bit, as it is set to make the scancode have an odd number of ones in the dozen or so cases I checked.

Wait a minute. This is just PS/2! The bit order is flipped, but that’s just the way I was printing out the data. I wrote a quick bit of code for typing the corresponding character for each keypress, and then ordered a PS/2 to USB converter from Amazon, as I feel leaving a $10 dev board in the project when a $5 converter would do the job isn’t worth it.

So, I wired up the PS/2 converter to the keyboard, plug it into my laptop, and … nothing… What gives? I tried diagnosing the issue. I checked the wiring, everything looked right. A friend suggested that maybe the Model M was drawing too much power, so I hooked it up to the power from the board I used to read its signal initially. Still Nothing. Maybe, I thought, the converter doesn’t do anything if there is no load. So I put a resistor across the power pins from the converter. Still nothing!

Maybe the host is expecting some sort of response from the keyboard that it isn’t providing. So, I hooked up an Adafruit Trinket in parallel to the PS/2 to USB converter to read the signals on the line. I also hooked up a multimeter to the CLK line. I noticed something - the keyboard only pulls CLK up to 2.5 volts. I quickly tested the keyboard at a lower voltage - it ran perfectly fine at 3.3 volts. Is there a chance it doesn’t produce a voltage high enough for the converter to recognize as logical high? I checked the logic level on the converter - its CLK line is held at 5 volts. So, I ordered a level shifter and decided to try it out. Unfortunately, I could not get it to work and then the semester ended, so I’ll have to try again in the fall.

To be continued… (hopefully)

Tags: Keyboard IBM Legacy
Part of a series on IBM.