Wanted: AT&T UNIX PC / PC 7300 / 3B1 mouse

[agentb]

Not like anyone has one of these kicking around! But I could use an AT&T UNIX PC mouse if anyone happens to have a spare.

My backup plan is to try to make a converter from a ps/2 mouse. I have a decent idea of the protocol the UNIX PC is expecting, but not exactly sure how the keyboard and mouse interact as the mouse passes through the keyboard which talks to a Motorola 6850 UART chip.

 

[agentb]

Followup.. I was able to figure out hooking up a ps2 mouse so now have an old dell mouse working with it! Running through an arduino which reads the ps2 mouse and then sends 3-byte packets to the 3b1 at 1200 baud

 

[agentb]

The original AT&T 3b1 mouse looks suspiciously like the Sun optical mouse from Mouse Systems -- but every Mouse Systems mouse that I could find (that looked like the 3b1 mouse) was optical, requiring that metallic grid mouse pad. Hmmm...

 

[shirsch]

I ended up with (3) AT&T 7300 / 3B1 machines and only (2) mice - bummer. Has the Arduino adapter continued to work well for you? Barring a lucky eBay score (which I seriously doubt will happen) I'll need to build up one for myself.

 

[agentb]

3 machines, wow! The mouse project was working very well for me, I had hoped to make a PCB for it but didn’t learn kicad. I fortunately was able to get a proper UNIX PC mouse.

If you want to try the project, I have the docs here:
https://github.com/agentbooth/3b1mouse

The biggest trick was realizing the serial comm needed to be inverted which I used the software serial support for that.

 

[shirsch]

The original AT&T 3b1 mouse looks suspiciously like the Sun optical mouse from Mouse Systems -- but every Mouse Systems mouse that I could find (that looked like the 3b1 mouse) was optical, requiring that metallic grid mouse pad. Hmmm...

I reached the same conclusion. I have a Mouse Systems branded optical mouse and (2) Sun variants in that form factor.

 

[shirsch]

3 machines, wow! The mouse project was working very well for me, I had hoped to make a PCB for it but didn’t learn kicad. I fortunately was able to get a proper UNIX PC mouse.

If you want to try the project, I have the docs here:
https://github.com/agentbooth/3b1mouse

The biggest trick was realizing the serial comm needed to be inverted which I used the software serial support for that.

Thanks - I have the project cloned and am picking through it as we speak. I'm going to try an implementation on RPi Pico, since I don't have an Arduino handy. May have questions for you regarding the protocol.

 

[agentb]

Sure, I’m pretty deep in UNIX PC internals right now (working on MAME support) if you have questions on the protocol. I haven’t used a Pico yet but seems like a pretty cool device. You will need inversion on the serial comm.

 

[shirsch]

Sure, I’m pretty deep in UNIX PC internals right now (working on MAME support) if you have questions on the protocol. I haven’t used a Pico yet but seems like a pretty cool device. You will need inversion on the serial comm.

By inversion, you mean high logic level = 0 and low = 1?

 

[agentb]

The inversion is specifically about the serial Tx line that goes from the microcontroller to the keyboard port. I tried the regular serial class/line on the arduino and it didn’t work. No inversion supported. Apparently you can run the line through an inverter to get it to work. But the softwareserial class lets you turn on inversion so the logic is opposite (no external inverter needed). I’m not sure exactly what voltages rs232 uses in this case for high and low.

 

[shirsch]

Based on your experience, I'm going to infer that the AT&T serial interface is following RS232 standard wherein a '1' is positive going. TTL I/O typically uses open-collector outputs that go negative on '1'. See: https://www.sparkfun.com/tutorials/215

 

[agentb]

I found this link the other day where a guy hooks up a UNIX PC keyboard to USB. Looks like he uses a 7400 to invert the signal: https://github.com/jsheradin/ATT_3B1_KB_USB

I think there's some confusing information in this project -- like the keycode values he's seeing and the N71 (vs N81) -- which I believe are both wrong -- but at least it covers the inversion of the serial signal.