froller wrote on 2021-02-13, 09:35:

maxtherabbit wrote on 2021-02-13, 03:24:

froller wrote on 2021-02-12, 20:58:

Does your one have 4 72-pin SIMM slots and VT82C41N soldered between two lower ISA slots along with some 74-series logic chip?

Indeed

It looks fully packed and not suitable for modding. At least on photos.

I've got it handled, going to desolder the VT82C41N and spin up a little PCB that mounts to a slot bracket with a DIP-40 KBC, 7406 and associated passives. Then connect that to the DIP-24 socket with a ribbon cable.

I was hoping that the mini-KBC might have the ability to drive a mouse but it appears it does not. I'm waiting for a 28C010 to arrive from china so I can enable ps/2 support on my own board and probe the 4 unidentified pins on it with a scope, as a last resort.

Done

maxtherabbit wrote on 2021-02-13, 20:38:

Done

According to layout you are planning on using AT keyboard connector located on MB.
Why don't you just place both PS/2 mouse and keyboard connectors on your board?

froller wrote on 2021-02-14, 10:54:

maxtherabbit wrote on 2021-02-13, 20:38:

Done

According to layout you are planning on using AT keyboard connector located on MB.
Why don't you just place both PS/2 mouse and keyboard connectors on your board?

I guess I didn't see a reason to bother. I don't like having jacks that don't work either, since the one one the motherboard would be disabled by your suggestion - plus it would increase BOM cost with another set of inductors/pullups/filter caps.

All that extra crap would cost as much or more than a DIN to mini-DIN adapter dongle

maxtherabbit wrote on 2021-02-14, 15:50:

All that extra crap would cost as much or more than a DIN to mini-DIN adapter dongle

I got sick and tired searching for fancy slick DIN-miniDIN adapter so I made one myself from 10cm of ethernet cable, miniDIN socket salvaged from dead mobo and DIN plug (the only thing I had to buy). Works much better than looks.

UTSOURCE finally got over the hangover after Chinese New Year and managed to send me 2 pcs HT6542B along with 2 pcs SiS85C496 for my burnt Rev:D

I'm planning to solder both DIP-24 and DIP-40 sockets one inside another to be able to switch back to original VT83C41 in no time. Hope they fit with just cutting and no grinding at all.

I swapped VT83C41 to AMIKEY-2 I salvaged from dead 4DPS board and it almost works.

- Planting 74LS06 renders old VT83C41 completely unusable either with pull-up resistor pack (no key presses detected) or without it (some keys stuck forever).
- Replacing VT83C41 with AMIKEY-2 brings everything to life again.
- You have to update your BIOS to enable PS/2 mouse support
- You have to replace JP17 soldered jumper wires with fancy pin headers and jumper caps (see attached image) in order to be able to update BIOS in-circuit.
- With updated BIOS PS/2 mouse is detected and driver loads ok.
- UNFORTUNATELY every little move of mouse freezes the KBC

It is either some software problems or missing some pull-up resistors like R51.
R51 is a pull-up resistor on open collector output of 74LS06. This output is connected to Pin3 of 83C42 (AMIKEY-2, HT6542B, VT83C42, whatever) that is OSCO. Corresponding input is connected to SYSCLK (ISA B20). OSCI (Pin 2) in turn is tied to the ground. This is one of typical clock connections for 83C42. So improper clocking may lead to freezing of whole KBC.

Will check this evening.

UPD: I also noticed open JP18 and JP19. First one ties Pin 33 (GPIO P16) of 83C42 to ground. Second one connects Pin 36 (MS-OBFO) to some mobo internals, probably the interrupt controller. Seems I have to close JP19 to make mouse work.

froller wrote on 2021-03-16, 08:34:

- UNFORTUNATELY every little move of mouse freezes the KBC

Second one connects Pin 36 (MS-OBFO) to some mobo internals, probably the interrupt controller. Seems I have to close JP19 to make mouse work.

Not having the interrupt line connected causes exactly what your described. MS-OBFO should be directly connected to IRQ12

maxtherabbit wrote on 2021-03-16, 12:37:

Not having the interrupt line connected causes exactly what your described. MS-OBFO should be directly connected to IRQ12

Thanks, man! Will solder JP19 closed.

According to Holtek documentation MS-OBFO is pulled low if PS/2 mouse support is off. I think it's safe to leave this jumper closed forever.

Yes it works! 😀)
Gonna do some more checks and photos before writing my "Lucky Star LS-486e Rev:D Ultimate Upgrade Guide" 😀

Looks like this topic is done.
Summarized info is here