measure 5V rail instead of guessing
this video might be helpful if you have a scope https://hackaday.com/2023/02/07/connecting-a- … t-the-hard-way/

What tipe of psu are you using AT or ATX ? If it is a ATX psu check how the jumper for power of the keyboard is set up ( you have 2 pos: 1. is taking power from the 5v main rail 2. power from the 5V standby rail (this one can't supply a lot of power/amps and in cheap psu's is poorly regulated). It should be near the keyboard connector.

I am using an AT power supply, which might be getting flaky. It's one of those really old ones, with the power switch on the side. I liked the case it was in, but it doesn't have a power button on the front, so I can't change the PSU for a newer one. I might have to use a different case / power supply. I do have a vague memory of trying out the motherboard with an ATX power supply and it didn't work (or at least, it didn't react to shorting the power button pins on the motherboard). But starting by measuring the 5V is a good idea, and will be my next step. I do have a scope, so I'll also check what is going on with the signals.

I didn't even know there was a jumper for the keyboard power! Very interesting.

Quick observation (I haven't measured anything yet), but I just discovered the keyboard clock speed in the BIOS settings, which can be set to 8mhz, 12mhz or 16mhz. When I set the clock to 16mhz, the keyboard works fine, _except_ that it gives a keyboard error during POST. When I press F1 to continue, it's ok after that. Not ideal, but a step closer, and maybe another hint as to what is going on.

I don't know what keyboard clocks your new keyboard supports , but usually 8 mhz should be just fine. You have the jumper in the pic, If you use a AT power supply it should be on the 2-3 pos (disabled).

wkjagt wrote on 2023-02-17, 21:51:

Quick observation (I haven't measured anything yet), but I just discovered the keyboard clock speed in the BIOS settings, which can be set to 8mhz, 12mhz or 16mhz. When I set the clock to 16mhz, the keyboard works fine, _except_ that it gives a keyboard error during POST. When I press F1 to continue, it's ok after that. Not ideal, but a step closer, and maybe another hint as to what is going on.

this clock setting is for _keyboard Controller_, not the keyboard itself, and Im very surprised it exists in the bios of P5A-B at all as that board doesnt have external keyboard controller chip - its all integrated into ALi M1533 South Bridge https://bitsavers.org/components/ali/M1541_M1 … orth_Bridge.pdf

I knew that was the speed for the keyboard controller and that 8 mhz should fine (that is the later adopted ISA bus clock ) but somehow i was always under the impression that when you change the MCU/keyboard controller clock that signal is somehow tied to the keyboard clock also. You learn new things all the time. Thank you.

Have you contacted the keyboard manufacturer's support about this issue?

I tried with a different power supply, but that gives the same problematic behaviour.

I just tested this same keyboard on my 486, and it works totally perfect, no problems at all. So the problem seems to be with the P5A-B, or the combination of the motherboard and the keyboard.

By the way, this is the same motherboard I have another problem with (described in PC "waits" for 15 seconds on boot). That problem disappeared while I was trying to figure it out, but since came back. These two problems may even be related. I am saying that because during that 15 second or so pause (as described in the linked post), the keyboard seems to turn itself off (or the PC sends a command to disable it?) or at least, that is when the backlight turns off.

Next step will be measuring voltages (instead of guessing). And I was thinking hooking up an Arduino in parallel with the keyboard and use the PS/2 keyboard library to see if I can see what scan codes and commands are being sent between the motherboard and the keyboard, and see if maybe that could give some hints. If that works, it'll be easier than interpreting clock and data pulses on an oscilloscope.

Doornkaat wrote on 2023-02-18, 07:12:

Have you contacted the keyboard manufacturer's support about this issue?

Nope, but good idea! Although in modern keyboards, PS/2 support seems to be implemented for a completely different use case (NKRO) than mine.