ahyeadude wrote on 2023-10-27, 20:20:

So, I followed the instructions to modify the AMIBIOS in this thread... […]
Show full quote

mockingbird wrote on 2023-10-27, 19:51:

ahyeadude wrote on 2023-10-27, 19:21:

To close this loop. I gave up on this particular 386sx motherboard and am using another. The interposer works fine, but just have problems with using ps2suppc.com and the 5434 as referenced above.

I am building a GD5434 ISA and will report back if I have this issue with my 386. I flashed it to Mr. BIOS, so I don't need ps2suppc, but nevertheless.

So, I followed the instructions to modify the AMIBIOS in this thread...

AMI Color BIOS (1993 and earlier) modification in hex editor

Involves flipping a couple of bits. Easy. Works great. No more ps2suppc.com and it works perfectly fine with a 5434!

Thank you for making and sharing that python script to calculate the AMI bios checksum value, I was able to do the same on my 080893 AMIBIOS to enable PS/2 mouse support and unhide the option for it. Now I have working PS/2 mouse support on my DataExpert Opti495SX without needing to use PS2SUPPC anymore 😀

Although it's great for testing with, has anyone else found that PS2SUPPC can fail if run on a fast enough system? Running it on a 486 or a 386DX with cache I've found that quite often PS2SUPPC will fail to detect the mod board and the keyboard will then be non-functional until the computer is restarted. I found though that disabling cache, slowing the system or going into de-turbo state allows PS2SUPPC to work correctly and detect that the KBC mod is in place and a mouse is connected. Going back to turbo after that is fine and the mouse then works nicely.

I had thought that my build of this board was wrong since it's a version from a couple of years back using the 3904 inverters, so I made another but wired up a 7406 like feipoa's pdf schematic from many years back - that works but exhibits the same issue which allowed me to determine that it's not the circuit. Once I worked out it's PC performance that causes the issue, I checked both boards and they work the same.
This is all using two AMERICAN MEGATRENDS MEGA-KB-H-WP keyboard controllers, one of which was borrowed from a Dell LPX system that uses PS/2 mouse as standard. I'm not sure if the KBC is a factor though.

I don't think I ran into the issues that you are having, but it has been a long time since I was playing with these units. I did have an issue getting the discrete transistor-based model working. For the most part, I noticed that some keyboard controllers work better than others in certain motherboards. But if the solution is to just hack the BIOS to enable PS/2 rather than relying on PS2SUPPC, isn't this sufficient? Or are you saying it is too much effort to hack the BIOS on every board we want to test with a PS/2 mouse?

feipoa wrote on 2023-11-08, 22:22:

I don't think I ran into the issues that you are having, but it has been a long time since I was playing with these units. I did have an issue getting the discrete transistor-based model working. For the most part, I noticed that some keyboard controllers work better than others in certain motherboards. But if the solution is to just hack the BIOS to enable PS/2 rather than relying on PS2SUPPC, isn't this sufficient? Or are you saying it is too much effort to hack the BIOS on every board we want to test with a PS/2 mouse?

Really I'm just wondering if anyone else came across issues with PS2SUPPC causing the keyboard to stop if the computer is too fast. It stumped me for a while but made sense after a while since the only motherboards I was really ever able to get the PS/2 mouse modification working on were 386SX type systems that are rather sluggish, it always seemed to fail using the enabler on the faster systems. To my surprise, today I found that slowing down the system helped the success rate.
In the last few days I've learned a lot about the 8042 implementation and have enough understanding of the circuit to know it should work with a compatible KBC and the extra inverters, somehow never spotted your old hand drawn schematic 😀

I'm fine with using the enabler program but having to toggle turbo to use it does make a bios modification less hassle in the long run.

I don't have many mouse capable keyboard controllers to test with so can't confirm whether it's down to which type of KBC is being used. This board was purchased specifically because it came with this KBC, I don't have any of those Via KBCs yet. Way too many of my boards that have the KBC socketed and in a good location have either Jet keyboard BIOS or the Winbond one, neither likes the PS/2 mouse interposer PCB.

Thermalwrong wrote on 2023-11-08, 23:04:

Really I'm just wondering if anyone else came across issues with PS2SUPPC causing the keyboard to stop if the computer is too fast. It stumped me for a while but made sense after a while since the only motherboards I was really ever able to get the PS/2 mouse modification working on were 386SX type systems that are rather sluggish, it always seemed to fail using the enabler on the faster systems. To my surprise, today I found that slowing down the system helped the success rate.
In the last few days I've learned a lot about the 8042 implementation and have enough understanding of the circuit to know it should work with a compatible KBC and the extra inverters, somehow never spotted your old hand drawn schematic 😀

You're supposed to carefully watch bits 0 and 1 of the status at port 64h to pace yourself when accessing the KBC, maybe ps2suppc doesn't? https://www.plantation-productions.com/Webste … h20/CH20-2.html

Thermalwrong wrote on 2023-11-08, 22:01:

Thank you for making and sharing that python script to calculate the AMI bios checksum value, I was able to do the same on my 08 […]
Show full quote

ahyeadude wrote on 2023-10-27, 20:20:

So, I followed the instructions to modify the AMIBIOS in this thread... […]
Show full quote

mockingbird wrote on 2023-10-27, 19:51:

I am building a GD5434 ISA and will report back if I have this issue with my 386. I flashed it to Mr. BIOS, so I don't need ps2suppc, but nevertheless.

So, I followed the instructions to modify the AMIBIOS in this thread...

AMI Color BIOS (1993 and earlier) modification in hex editor

Involves flipping a couple of bits. Easy. Works great. No more ps2suppc.com and it works perfectly fine with a 5434!

Thank you for making and sharing that python script to calculate the AMI bios checksum value, I was able to do the same on my 080893 AMIBIOS to enable PS/2 mouse support and unhide the option for it. Now I have working PS/2 mouse support on my DataExpert Opti495SX without needing to use PS2SUPPC anymore 😀

Although it's great for testing with, has anyone else found that PS2SUPPC can fail if run on a fast enough system? Running it on a 486 or a 386DX with cache I've found that quite often PS2SUPPC will fail to detect the mod board and the keyboard will then be non-functional until the computer is restarted. I found though that disabling cache, slowing the system or going into de-turbo state allows PS2SUPPC to work correctly and detect that the KBC mod is in place and a mouse is connected. Going back to turbo after that is fine and the mouse then works nicely.

I had thought that my build of this board was wrong since it's a version from a couple of years back using the 3904 inverters, so I made another but wired up a 7406 like feipoa's pdf schematic from many years back - that works but exhibits the same issue which allowed me to determine that it's not the circuit. Once I worked out it's PC performance that causes the issue, I checked both boards and they work the same.
This is all using two AMERICAN MEGATRENDS MEGA-KB-H-WP keyboard controllers, one of which was borrowed from a Dell LPX system that uses PS/2 mouse as standard. I'm not sure if the KBC is a factor though.

If you have a key pressed at the time of PS2SUPPC's execution it will cause it to fail.

Are you running it from command line or in AUTOEXEC.BAT? My suspicion is what you are seeing may be a result of the faster machine loading and executing the binary so quickly that it still sees the key press from you hitting return to run it.

I've got a Lucky Star LS-486E rev F board, it has the same small DIP24 keyboard controller (vt82c41n) that maxtherabbit mentioned in the LuckyStar LS-486e Rev:D Hardware Upgrade Guide tucked between 2 ISA slots.
I would love to have a proper PS/2 mouse port on this board. Has anyone successfully managed replacing a vt82c41n with mouse-capable controller, like a vt82c42 or AMIKEY-2?

deltapi wrote on 2024-02-02, 19:42:

I've got a Lucky Star LS-486E rev F board, it has the same small DIP24 keyboard controller (vt82c41n) that maxtherabbit mentioned in the LuckyStar LS-486e Rev:D Hardware Upgrade Guide tucked between 2 ISA slots.
I would love to have a proper PS/2 mouse port on this board. Has anyone successfully managed replacing a vt82c41n with mouse-capable controller, like a vt82c42 or AMIKEY-2?

I've only upgraded two revision D boards, not rev F.

deltapi wrote on 2024-02-02, 19:42:

I've got a Lucky Star LS-486E rev F board, it has the same small DIP24 keyboard controller (vt82c41n) that maxtherabbit mentioned in the LuckyStar LS-486e Rev:D Hardware Upgrade Guide tucked between 2 ISA slots.
I would love to have a proper PS/2 mouse port on this board. Has anyone successfully managed replacing a vt82c41n with mouse-capable controller, like a vt82c42 or AMIKEY-2?

Yes I even made a PCB for it

maxtherabbit wrote on 2024-02-03, 21:10:

Yes I even made a PCB for it

That's awesome, have you posted it anywhere? I found your interposer files, I didn't see anything for this one. If you haven't, would you be willing to share it?

deltapi wrote on 2024-02-05, 01:27:

maxtherabbit wrote on 2024-02-03, 21:10:

Yes I even made a PCB for it

That's awesome, have you posted it anywhere? I found your interposer files, I didn't see anything for this one. If you haven't, would you be willing to share it?

Yes but there is a problem. I still have the gerber files (and a few unassembled PCBs) but I lost the original design files in a hard drive crash.

The PCBs and gerbers I have have a flaw - the pins on the ps/2 mini-DIN footprint are backwards because I foolishly trusted the pin numbering in kicad without verifying it myself. The boards can still be made to work but you have to solder the ps/2 plug on the wrong side of the PCB to invert the pin order. The original design intent was to build the board with a slot bracket in "ISA card" orientation but because of the backwards connector you end up having to build it with the components on the "solder side" which block the adjacent slot.

If you're in the USA I'd be happy to give you a board, if not I would just want shipping reimbursement.

maxtherabbit wrote on 2024-02-05, 17:33:

Yes but there is a problem. I still have the gerber files (and a few unassembled PCBs) but I lost the original design files in a hard drive crash.

The PCBs and gerbers I have have a flaw - the pins on the ps/2 mini-DIN footprint are backwards because I foolishly trusted the pin numbering in kicad without verifying it myself. The boards can still be made to work but you have to solder the ps/2 plug on the wrong side of the PCB to invert the pin order. The original design intent was to build the board with a slot bracket in "ISA card" orientation but because of the backwards connector you end up having to build it with the components on the "solder side" which block the adjacent slot.

If you're in the USA I'd be happy to give you a board, if not I would just want shipping reimbursement.

Out of curiosity, what is the difference between this and your v3 revision? Thanks

mockingbird wrote on 2024-02-05, 21:35:

maxtherabbit wrote on 2024-02-05, 17:33:

Yes but there is a problem. I still have the gerber files (and a few unassembled PCBs) but I lost the original design files in a hard drive crash.

The PCBs and gerbers I have have a flaw - the pins on the ps/2 mini-DIN footprint are backwards because I foolishly trusted the pin numbering in kicad without verifying it myself. The boards can still be made to work but you have to solder the ps/2 plug on the wrong side of the PCB to invert the pin order. The original design intent was to build the board with a slot bracket in "ISA card" orientation but because of the backwards connector you end up having to build it with the components on the "solder side" which block the adjacent slot.

If you're in the USA I'd be happy to give you a board, if not I would just want shipping reimbursement.

Out of curiosity, what is the difference between this and your v3 revision? Thanks

It's a completely different thing. This board is solely for the lucky star ls486e boards with the DIP24 KBC

Pico rp2040 has IO quick enough to replace whole KBC providing both native PS2 and USB keyboard/mouse, maybe even with transparent joypad mapping. One plus would be ability to change board compatibility (am I correct in assuming its about how spare IO are routed to things like Turbo/A20?) in software.
Whole thing would fit in DIP 40 footprint pcb and be orderable preassembled from jlcpcb for ~$10-15 in parts and assembly (minus DIP pins as those are hilariously expensive to solder at jlc).

rasz_pl wrote on 2024-02-06, 03:27:

Pico rp2040 has IO quick enough to replace whole KBC providing both native PS2 and USB keyboard/mouse, maybe even with transparent joypad mapping. One plus would be ability to change board compatibility (am I correct in assuming its about how spare IO are routed to things like Turbo/A20?) in software.
Whole thing would fit in DIP 40 footprint pcb and be orderable preassembled from jlcpcb for ~$10-15 in parts and assembly (minus DIP pins as those are hilariously expensive to solder at jlc).

The DIP24 KBC in the lucky star board has no GPIO whatsoever. It's a bare bones KBC and nothing else.

I have zero interest in using modern microcontrollers for retro PC development personally

maxtherabbit wrote on 2024-02-06, 02:53:

It's a completely different thing. This board is solely for the lucky star ls486e boards with the DIP24 KBC

Thanks. If you share the gerber, I think I can create a KiCad file out of it -- if you're interested.

mockingbird wrote on 2024-02-06, 04:09:

maxtherabbit wrote on 2024-02-06, 02:53:

It's a completely different thing. This board is solely for the lucky star ls486e boards with the DIP24 KBC

Thanks. If you share the gerber, I think I can create a KiCad file out of it -- if you're interested.

Including a schematic file?? That would be a game changer

maxtherabbit wrote on 2024-02-05, 17:33:

If you're in the USA I'd be happy to give you a board, if not I would just want shipping reimbursement.

I'm in Canada, but I'd gladly pay shipping expenses, thank you!

@maxtherabbit

May I ask the value of chokes L1 and L2 in your "ps2interposer v3"? There is description which states 470uH, but I wanted to make sure.
Shouldn't these be ferrite beads / EMI filters? If so - what value?

On the 486 boards I looked at, I saw these inductors in an axial can or SMD and of value between 2 uH to 10 uH. I use 2.2 uH on my adaptors. I noticed that the Lucky Star LS-486E rev.D upgrade guide uses 470 uH, which I assume maxtherabbit adopted. It was not clear to me why this value was selected. Is there any benefit in using the higher inductance?

I did not observe any benefit of the higher value when I used it on my LSD board for the PS/2 upgrade. I did notice some keyboard controllers not working with the mouse on the LSD, but I haven't bothered removing the 470 uH inductors in exchange for 2.2 uH inductors to see if the inductor value was a factor in this.

It seems that for EMI Filters/ferrite beads - inductance is not main factor/unit (what confuses me a little)...
It is impedance at 100MHz frequency !

This is interesting read (well, at least for me 😀 https://www.alliedcomponents.com/blog/ferrite-bead-inductor
Also this: https://www.allaboutcircuits.com/technical-ar … -ferrite-beads/
And this: https://www.youtube.com/watch?v=MHqWu8b5VD4

So, as I understand it (at this point): in our case ferrite beads should cut everything above normal working frequency for PS/2 protocol (10-16.7 kHz) ?

Edit.
It seems that generally all ferrite beads are meant to cut very high frequency signals - in MHz range. So it doesn't really matter which one we choose 😉
My point is that the PS/2 signal is safe regardless of which one we choose, it is more important to guess (or maybe measure) what kind of noise is expected here.