rasz_pl wrote on 2022-08-18, 12:04:

scruit wrote on 2022-08-17, 11:25:

- There are ps/2 keyboards that can work with XT adapters because, although they natively use ps/2 protocol, they can detect and communicate on XT protocol

never seen one able to switch automagically

People claim that at least early series of the IBM Model M keyboard that came with a DIN-plug cable could autodetect XT/AT. As the keyboard cable is detachable, it is likely that they didn't redesign the keyboard when they started shipping them with PS/2-plug cables. So I expect that early PS/2-type model M keyboard (like ones shipped with the IBM PS/2 Model 30) do support the XT protocol.

I've also had issues where certain PS2 keyboards did not work with AT motherboards using passive PS2 to AT adapters no matter what I did. They work fine on other boards. *shrug*

I think it's either that the board does not have enough power on the Vcc line, or it is an issue with the clock signal or data signal level. Sometimes you get them light up and detected at POST but the keys do not register, sometimes they just spit out keyboard not detected errors. Some boards tolerate them fine, some do not. Such is life.

Solo761 wrote on 2022-08-17, 08:54:

I had similar issue with one of my Socket 7 motherboards. Some cheapo PS/2 keyboard I had, relatively modern, about 5 years old, […]
Show full quote

I had similar issue with one of my Socket 7 motherboards. Some cheapo PS/2 keyboard I had, relatively modern, about 5 years old, worked fine. I found two older PS/2 keyboards from 15-20 years ago and it was no go with them.

One appeared to work fine when powering on (LEDs turning on), but when you pressed any key it would just blink it's LEDs, unless I turned caps lock LED off on boot, then it worked fine, until I tried to press any of the keys that turn LEDs on.
The other one was just blinking with it's LEDs. Didn't recognize any key press.

Turned out it was motherboard's fault (btw. it was QDI Titanium IB+), it had transistor to controll voltage on AT keyboard (and I guess PS/2 mouse port) pins and it was faulty. Without anything connected it held at 5V, but when keyboard was connected it was sub 4V. And when you actually started to use the keyboard it went below 3V.
Modern keyboard was least power hungry so even below 3V was enough for it to work. For these others it wasn't, first one managed to keep a bit above 3V when LEDs were off so that's why it worked if caps lock was off. But turning any LEDs on would dip it below 3V and it couldn't recover.

After I replaced that transistor all of these keyboards work without issues.

Interesting, this. Which IC on my board would likely be the culprit for such a similar issue then?

afaik build into main chipset on this one

appiah4 wrote on 2022-08-18, 13:30:

Solo761 wrote on 2022-08-17, 08:54:

Turned out it was motherboard's fault (btw. it was QDI Titanium IB+), it had transistor to controll voltage on AT keyboard (and I guess PS/2 mouse port) pins and it was faulty. Without anything connected it held at 5V, but when keyboard was connected it was sub 4V. And when you actually started to use the keyboard it went below 3V.

Interesting, this. Which IC on my board would likely be the culprit for such a similar issue then?
[... picture of a Bioster MB-8433-UUD ...]

None. The idea of transistors to control the power to AT keyboards was introduced to optionally cut keyboard power in a suspend-to-RAM situation. The power management on the Biostar 486 board is not that sophisticated. One of the five interference suppression chokes at the top left corner of your board is in the +5V line, and it might possibly have too high impedance, but no transistors are involved in powering the keyboard.

mkarcher wrote on 2022-08-18, 13:56:

appiah4 wrote on 2022-08-18, 13:30:

Solo761 wrote on 2022-08-17, 08:54:

Turned out it was motherboard's fault (btw. it was QDI Titanium IB+), it had transistor to controll voltage on AT keyboard (and I guess PS/2 mouse port) pins and it was faulty. Without anything connected it held at 5V, but when keyboard was connected it was sub 4V. And when you actually started to use the keyboard it went below 3V.

Interesting, this. Which IC on my board would likely be the culprit for such a similar issue then?
[... picture of a Bioster MB-8433-UUD ...]

None. The idea of transistors to control the power to AT keyboards was introduced to optionally cut keyboard power in a suspend-to-RAM situation. The power management on the Biostar 486 board is not that sophisticated. One of the five interference suppression chokes at the top left corner of your board is in the +5V line, and it might possibly have too high impedance, but no transistors are involved in powering the keyboard.

Ah, I see. So as long as the PSU +5V line is fine, the power delivery should also be fine.. I suppose I could find whichever choke is on the +5V line (they might all be, I think?) and mess around with different values but I really don't see the point to be honest. It works fine with any AT keyboard I throw at it..

appiah4 wrote on 2022-08-18, 14:11:

Ah, I see. So as long as the PSU +5V line is fine, the power delivery should also be fine.. I suppose I could find whichever choke is on the +5V line (they might all be, I think?) and mess around with different values but I really don't see the point to be honest. It works fine with any AT keyboard I throw at it..

Four of the chokes are for keyboard and mouse clock and data respectively, the fifth one is for Vcc. I don't have that board at hand at the moment, so you would need to check that yourself.

I agree that replacing the choke is unlikely to help, though. If you want to try it, start with just shorting the +5V choke, but after doing so, do not plug or unplug the keyboard while the computer is powered on (you shouldn't do that anyway). If shorting doesn't help, using a different choke won't help, either.

dr_st wrote on 2022-08-18, 12:24:

[...] […]
Show full quote

[...]

I have a couple of keyboards that came with such a purple USB-->PS/2 adapter and work with it. It looks passive, and I don't think it works with all keyboards.

Furthermore, Startech sells such a PS/2-->USB adapter, which they claim works with the keyboards that originally shipped with a similar one. I have never seen such a keyboard myself, though.

What do you think?

Hmm yes, that does indeed look like a completely passive one for a dual-protocol board. Never seen those in real life, but they certainly sound like they would work like the similar mice do.

appiah4 wrote on 2022-08-18, 13:30:

Interesting, this. Which IC on my board would likely be the culprit for such a similar issue then?

In case of my board it was regular three legged transistor near AT socket. No such part here.

Here's a bit better image of your board, as mkarcher says, only chokes are near AT socket on your board.
https://theretroweb.com/motherboards/s/biostar-mb-8425uud-a

If you have it somewhere accessible you could try checking 5V line of AT connector and what exact voltage you have there with each keyboard. That's how I noticed what was the cause of problem and how I traced it to this transistor. It was the only thing in path from 5V rail from power supply to AT connector and before transistor 5V was steady, after it it would drop. So it had to be where the issue was.

rasz_pl wrote on 2022-08-18, 12:04:

Rincewind42 I would try few things -measure voltage at keyboard connector, static and with some kind of a scope during long key […]
Show full quote

Rincewind42 I would try few things
-measure voltage at keyboard connector, static and with some kind of a scope during long keypresses generating errors
-put a scope on data lines and check frequency, should be in 10-16.7 kHz range, compare to working keyboards
-put a scope on data line and see how clean it is

Would be interesting, for sure, but I don't have a scope, just a simple multimeter. In any case, even if I figure out what the problem is, it's most likely unfixable.

Btw, I picked up another "new old stock" noname PS/2 keyboard today (Optima, never heard of it). It works flawlessly and it has the classic look, plus it's non-clicky, so I'm happy 😀