Hi, I got this MB https://theretroweb.com/motherboards/s/qdi-p6 … -advance-5-133e from a scrap with PIII/550MHz and testing it. Visually it's in a good conditions, no scratches, etc.
But the 1st problem I had after power on was the PS/2 keyboard. For some reason it doesn't work with my Labtec Y-SAM64 PS/2 kbd that I usually use for testing (it rarely make some keypress so I can enter SETUP but cannot set much more). This MB has 2 onboard PS/2 connectors for KB and mouse. I measured all PS/2 pins for shorts/connections, checked 5V voltage, all components in signal path (LC filters, 4k7 pull-up R-network and serial 0R-network) and it was OK so I decided that it could be partially damaged SuperIO chip W83977EF-AW e.g. by ESD. I had some spare one so I replaced it but still the same!

So I tried another keyboard Chicony KB-2971 and it worked. WTF? but well maybe some strange incompatability...
But wait, not done yet. Futher madness started when I plugged PS/2 mouse (1st I tried MS intellimouse optical USB in PS/2 adapter) in PS/2 mouse port. It caused that keyboard became nearly irresponsive. ~95% key hits did nothing. Even problems in SETUP or plain DOS before loading any mouse driver. When I pulled mouse out KB started work normally. I also tried to load CTMOUSE driver in autoexec, then the keyboard worked until I made 1st move with mouse. The mouse doesn't work normally - programs that use mouse on start just show mouse cursor running to corner to the right or down or cursor jumps like crazy - totally incrontrollabe with mouse...

Then I took out the scope to look at KBCLD and KBDATA waveforms. Without mouse attached it looks normally. Both in H level at idle with short data/clk passed when a key pressed. But with attached mouse a strange thing happen. KBCLK went L with only a short pulses to H at frequency 22,2Hz and H duration ~140us. (it was noted to be a inhibit state - superIO can do this by pulling CLK low when it cannot read out incoming data) This happened quite soon in POST stage of HDD detection and remains under plain DOS. When CTMOUSE was loaded the KBCLK went idle (H) but only until the 1st mouse move - then again KBCLK went L with only a short pulses to H but now with frequency 200Hz. I found a correlation with pulsed on IRQ12 line on ISA slot that should be linked with PS/2 port, the same frequency but H duration was only ~40us.

I consulted with AI but it gave a lot of misleading statements. It belives that it's a problem of VT82C596B southbridge that was unable properly read out both KB and mouse at high data rate. It suggested me some shot code that is capable to reset keyboard in working state (it makes KBCLK going idle H) until mouse moves again.

1;PS/2 keyboard reset
2.MODEL TINY
3.CODE
4       ORG 100h
5       STARTUPCODE
6       CLI
7
8WAIT1: IN AL, 64h
9       TEST AL, 02h
10       JNZ WAIT1
11
12       MOV AL, 60h
13       OUT 64h, AL
14
15WAIT2: IN AL, 64h
16       TEST AL, 02h
17       JNZ WAIT2
18
19       MOV AL,61h
20       OUT 60h, AL
21       STI
22       INT 20h
23END

It told me that USB intellimouse is hardcoded to spit data at high 200Hz rate that VT82C596B can't handle and suggested to use some older native PS/2 mouse that can be programmed to lower data rate. So I found and tested Genius Mouse NetScroll Optical PS/2. There was really some change in behavior. When power on the MB with KB and mouse attached the mouse light up immediate after power on and I can see pulses at KBCLK with 200Hz rate but strangely the keyboard was relative responsive so I can type with only rare misses. I loaded mouse driver and it swithed data rate to 80Hz and keeps KBCLK idle (H) so keyboard typing was OK but mouse still doesn't work properly - cursor flew instantly to corner or sceen edge. After some messing with mouse it switched data rate to 200Hz and pulses at KBCLK appeared again. (I checked multiple times but there's no any electrical short or connections between KBDATA, KBCLK, MDATA, MCLK).

I never seen such behavior before but I mostly deal with intel chipset MBs. I know that VIA and Ali had a lot of weird quirks, so can't say this would be normal on some crappy chipsets... Or maybe there could be some broken connection between Super IO and Southbridge or Southbridge and Northbrige that could cause this?

As a last try I booted Hiren's CD and tried MiniXP and Linux/Gparted. Under MiniXP the mouse cursor was frozen in the middle of sceen, in control panel PS/2 mouse was detected. Under Linux I had jumpy cursor like under DOS and received lost sync messages in console, see below. I also tried to plug USB mouse in USB port but even USB enabled in SETUP anb UHCI is detected the USB mouse was not detected neither in MiniXP nor Linux. USB ports provide at least 5V power. So maybe also USB ports are dead...