Hi, I believe that's a common issue. The intel i8042 / i8048 aka MCS 48 (along with i8051/52) series are one of the earliest microcontroller chips.
For example, the i8052 also had a variant called 8052AH-BASIC, which was a stand-alone computer-in-a-chip running BASIC.
It could be controlled by a serial terminal and supported interrupts, even! 😁

Programming the old intel microcontrollers isn't exactly easy, but also not very hard.
To program something like a P8042AH, it just requires a supported programmer, such as the Willem.
Or an adapter board. Have a look for "MCS-48 adapter". Maybe there's one for your programmer, too.

You also need the Keyboard controller's ROM, where can we find that nowadays ?

Well I see this is the bigger problem than I thought - if there is no 'common' knowledge how to program them cheaply (Willem programmes are costly, and TL866 doesn't support for MCS-48 for now) and there are no ROM images with keyboard BIOSes (although It should be rather easy to reproduce) then in a few years we will have trouble finding replacement KBC 8042s..

Well now that you mention reproduce the code, maybe reproducing its behavior with another readily available controller could be made ?
You mentioned atmel microcontrollers, maybe the behavior can be reproduced with one of these and then a adapter PCB could be made in order to fit this into the i8042's place ?

This document may be useful: http://www.halicery.com/8042/8042_1503033.TXT
Also, google "1503033.bin" for download raw binary dump
Home-made programmer: https://www.eevblog.com/forum/microcontroller … icrocontroller/

Deksor wrote:

Well now that you mention reproduce the code, maybe reproducing its behavior with another readily available controller could be made ?
You mentioned atmel microcontrollers, maybe the behavior can be reproduced with one of these and then a adapter PCB could be made in order to fit this into the i8042's place ?

Atmel's 8051 is the closest relative, but their pinout is completely different, except for VCC and GND. Another microcontrollers from Atmel - AVR dosn't have parallel slave bus, plus have different pins for VCC and GND. PIC microcontrollers from Microchip have "parrallel slave port", but also pinout is complectly different from 8042.

Tronix wrote:

This document may be useful: http://www.halicery.com/8042/8042_1503033.TXT

Highly interesting!
By the way, the inner workings and the technical issues that appear with "Legacy Keyboard support" are in-depth discussed on OSDEV wiki article, which can be useful to in-depth understand whats happening there.

VIA has a hardwired version of 8042 KBC direct replacement - VT82C42, such that one no longer has to worry about the OTP ROM code. It claims to be compatible with IBM/Phoenix/AMI KBC protocols including the PS/2 standard.

Tronix wrote:

This document may be useful: http://www.halicery.com/8042/8042_1503033.TXT
Also, google "1503033.bin" for download raw binary dump
Home-made programmer: https://www.eevblog.com/forum/microcontroller … icrocontroller/

I've seen this thread, somehow I missed the bin though… that's definitely an option.

kjliew wrote:

VIA has a hardwired version of 8042 KBC direct replacement - VT82C42, such that one no longer has to worry about the OTP ROM code. It claims to be compatible with IBM/Phoenix/AMI KBC protocols including the PS/2 standard.

So it's a direct replacement without need of programming (already programmed as KBC)? Great news.
I will order some from utsource then and test it in my 286/386 mobos… Perhaps I will even find some motherboard with this chip installed. I think I've seen them on MVP3 Super Socket 7 motherboards and maybe some later as well.

Home-made programmer: https://www.eevblog.com/forum/microcontroller … icrocontroller/

interesting. Just please note that some user down the blog said that the transistors were drawn upside-down, so better double check the circuit. 😀

The MCS-48 family also supported external EPROMs, so maybe an alternate design could use an 27Cxxx EPROM instead of internal storage.

I've also found an intersting document, explaining the history of the chip family. 😀

https://www.dos4ever.com/8031board/history8051.pdf
Source: https://www.dos4ever.com/8031board/8031board.html

Now that's a surprise! At some point, the old Genius mice also had got an 8042 derivative inside. Haven't expected that.

By the way, some of the modern 805x/875x are pin compatible with their MCS-51 ancestors and contain internal EEPROM/Flash-ROM.
These can be used with USB programmers such as the G540, for example.

- I'm currently trying to program one of these (AT89C51RB2 series w/ 1KiB XRAM, 16-32KiB Flash, 256 Bytes Scratch Pad RAM),
so it could act as a replacement for my 8052AH-BASIC chip (these are really rare nowadays).

So maybe there's also a chance to "port" an 8042 Keyboard-Controller to the
later generations, not sure. I'm just starting to understand this platform. 😅

PS: Also interesting:
http://www.theresistornetwork.com/2013/07/pro … tel-mcs-48.html
http://www.cpushack.com/Historyofthe8051.html

I wouldn't worry about pin compatibility. Just make an adapter PCB with a surface-mount micro that has the necessary IO.

It's kind of a long-term project of mine to re-implement everything on an XT / AT motherboard from scratch, up to the CPU itself. No idea how far I'll get before I lose interest, but I'm almost done with the PS2 interface. Hardware layer is done, but as yet untested. KB protocol layer is in the works now - a day or two probably. Next up: XT. I'm writing it in C with user-provided function pointers for the hardware interface (IO port manipulation, interrupt management, etc.) so I can easily compile it for different platforms, but AVR is my platform du jour, or ARM Cortex M when I need more brawn.

Hi there

Resuming this old post. Any updates on this interesting subject?

I think I have a partly faulty 8042 - it works for a while then the keyboard stops working - and it seems that the MB's BIOS only likes that controller.
Is there any changes in 2021 to extract the programming - reprogram another chip? I do have a TL866.

Thanks!