Unfortunately I have no board of rev. D. but the guide's awesome - props!

I'll be releasing a PS/2 add-on board for rev C shortly. Will post here when it's fully tested. It will require desoldering the 24-pin KBC from the motherboard same as rev D, but will then connect the add-on board to that footprint using a 24pin ribbon cable.

On my board I just used modbin to enable ps/2 support in the original lucky star BIOS, but I like the idea of using the BIOS from the UMC board instead. I'll give that a try

maxtherabbit wrote on 2021-03-18, 17:43:

I'll be releasing a PS/2 add-on board for rev C shortly. Will post here when it's fully tested. It will require desoldering the 24-pin KBC from the motherboard same as rev D, but will then connect the add-on board to that footprint using a 24pin ribbon cable.

What is your mobo exact revision? I saw some photos of upgrade capabe LS486e's besides Rev:D.

maxtherabbit wrote on 2021-03-18, 17:43:

On my board I just used modbin to enable ps/2 support in the original lucky star BIOS, but I like the idea of using the BIOS from the UMC board instead. I'll give that a try

This BIOS is from SiS496/497 motherboard. UMC8669 is some kind of peripheral controller. Probably for floppy drive or parallel port (or both).
I was also confused when I saw BIOS string containing "UMC" on a pure SiS motherboard.
Looks like there are motherboards with different chips instead of UMC8669 and they have different BIOSes. But both Rev:D's I have are exactly the same except battery holder.
The BIOS in guide was downloaded from original ROM chip and then changed by modbin to enable PS/2 mouse support. I also added "8042" to BIOS string to distinct new BIOS from the original one.

Generally it is a bad idea to use BIOS from different chipset. It won't POST with 146.5% probability.

I'll have to look at the board again to get you the exact revision. It's the one where the DIP24 KBC is tucked in between the ISA slots.

Now I'm tracking about the BIOS, guess I'll stick with what I have since my board used the Winbond super IO

maxtherabbit wrote on 2021-03-18, 18:30:

I'll have to look at the board again to get you the exact revision. It's the one where the DIP24 KBC is tucked in between the ISA slots.
Now I'm tracking about the BIOS, guess I'll stick with what I have since my board used the Winbond super IO

Your BIOS string should be like "486E 96/10/24 WB3787 PLUG & PLAY BIOS"

As far as I can see Rev:A (a.k.a. no revision) had AMIKEY-2 by design. Thus there should be more revisions capable of PS/2 mouse upgrade.
Moreover some boards have 4 x 64KB DIP-32 cache that means it can be extended to 512KB. Unfortunately my Rev:D's can't 'cause they have 8 x 32KB SO-28 and no spare footprints.

this is my board - no obvious revision markings I can see

Took some pics after I desoldered the DIP24 KBC and also of a via I found for IRQ12. I scraped off the solder resist so I can solder a wire to the via.

maxtherabbit wrote on 2021-03-19, 00:47:

this is my board - no obvious revision markings I can see

Looks pretty much like C2.

Have you tried to trace back CN1 and CN5 and resistor packs RN4, RN5 and RP9?

RP4 RP5 and RN9 do not connect to the DIP24 footprint. CN5 is an IRDA header. I don't see anything called CN1

Oh, my bad. CN1 is DIN-5 keyboard connector.

BTW, check this out The AT-PS/2 Keyboard Interface by Adam Chapweske.

Document revision 3 is out.
Changes: R51 value changed from 10k to 4.7k to avoid mouse glitches.

Document revision 4 is out.
Changes: jumper wire tying pin 26 of Holtek HT6542B to +5V.

W/o jumper wire it worked fine with AMIKEY-2 but failed to initialize keyboard controller with POST code 0x05 with HT6542B.

It's better to install a 40-pin socket instead of soldering a 80C42 compatible controller directly to the board.

The 74LS06 inverter is basically for PS/2 mouse support. 80C42 can run in either AT or PS/2 mode while this VT80C41 can do AT only. Although once you've got a 80C42 working, there's no point in going back anyway. C5/C7 and L4/L5 values are not strictly specific though 100pF and 1uH are most commonly seen. Up to 1nF and 10uH should do.

All three TH_* pins (5, 25, 26) should be pulled up to +5V for normal operation. Otherwise it's really a matter of chance if a particular chip works well or not.

Thanks for putting all this info in one place anyway. This VT82C41 sucks. I guess the ROM inside isn't masked. OTP EPROM probably. May get corrupt over time. It's a pain in the arse to find a compatible 24-pin direct replacement. I'm not even sure a data sheet for it even exists.

Thanks for the write-up. It will save me (and others) a lot of probing time. Is there any consequence in using 470 uH for the inductors? I usually see these around 1 to 4 uH on socket 3/4/5/7 boards, so I keep a stock 2.2 uH.

I agree that a DIP-40 socket should be added to the BOM. Perhaps ICS-640-T or ED40DT for the DIP-40 socket.

Some of the BOM items are non-stock, e.g. CAY16-103J4. CAT16-1002F4LF may be a suitable replacement. Part numbers missing for some items, so perhaps SG73S2ATTD472J for the 4.7K and C0805C470J5GACTU for the 47 pF cap.

I have performed the PS/2 mouse mod, both 2.2 uH or 470 uH inductors are functional for PS/2 mice on this board. The problem I had was with AMIKEY-2 - PS/2 mouse wouldn't function with it, but Holtek HT6542B and VIA VT82C42N both worked fine. I've decided to stick with the VIA. The hardest part of the mod is not to get burn marks on your ISA slots when soldering in the inverter IC.

feipoa wrote on 2021-08-02, 02:57:

both 2.2 uH or 470 uH inductors are functional for PS/2 mice on this board. The problem I had was with AMIKEY-2 - PS/2 mouse wouldn't function with it, but Holtek HT6542B and VIA VT82C42N both worked fine.

Initially I tested all these mods with AMIKEY-2 salvaged from Zida Tomato 4DPS which is initially capable of PS/2 mouse. This worked for me just fine.
You can just short-circuit these inductors for test puposes or use another with less inductance.

ph4nt0m wrote on 2021-06-21, 09:45:

It's better to install a 40-pin socket instead of soldering a 80C42 compatible controller directly to the board.

It isn't 'cause chip in socket is too high and prevents some ISA cards from being fully inserted in slots 1 and 2.
Anyway there's no way back to VT83C41 that doesn't require soldering operations.

froller wrote on 2022-01-21, 11:56:

ph4nt0m wrote on 2021-06-21, 09:45:

It's better to install a 40-pin socket instead of soldering a 80C42 compatible controller directly to the board.

It isn't 'cause chip in socket is too high and prevents some ISA cards from being fully inserted in slots 1 and 2.
Anyway there's no way back to VT83C41 that doesn't require soldering operations.

I have done it, no problem. The ISA slot is 15mm high, the socketed 80C42 is only 10mm high.

Depends on the card. I have a 386 mobo with socketed BIOS ROM chip in front of the ISA slots and some cards I own do not fully sit in the two slots this affects. A nearby KBC chip is not socketed and presents no such issues.

Latest revision is 5.
I've fixed mistake in BOM and also added links to Mouser to help getting proper parts or finding analogs.