STPC Elite and Winbond Super I/O W83977

moedeker2 · ‎2003-07-29

Posted on July 29, 2003 at 11:24

moedeker2 · ‎2003-07-03

Posted on July 04, 2003 at 05:41

I use the W83977 internal keyboard controller. All accesses to I/O port 61h (internal in STPC) are also putted onto ISA bus at the STPC Elite. The problem is that the W83977 could NOT distinguish between port 60h and port 61h. Every write to port 61h is taken from W83977 as a write to port 60h. The result is a horrible amount of accesses from keyboard controller to keyboard. It's not possible to for the keyboard controller to recognisize a pressed key because of the traffic at KBCLK and KBDAT.

This seems to be a problem of the W839 But there is no support from Winbond due to this chip.

Does anyone know if it is possible to configure the STPC that the accesses to internal regs are not putted onto ISA bus?

Bernd

thierry239955_st · ‎2003-07-03

Posted on July 04, 2003 at 06:57

Maybe I didn't understood well your problem but can't you modify the address decoding in your board by connecting the A0 line of the ISA to the Chip Select (AEN) of the Super I/O in such a way the odd addresses (then the 0x61) are not taken into account by the chip?

Farfalla

moedeker2 · ‎2003-07-03

Posted on July 04, 2003 at 08:39

Also is the RTC at address 70/71h in the super I/O. So I have to decode A0-A9 and then disable the chip by AEN or IOW. But I think it's much easier if there is a chance with a software fix. If there no software fix possible I will test it with an additional wired CPLD.

Bernd

moedeker2 · ‎2003-07-04

Posted on July 04, 2003 at 13:07

I just disabled IOWn to the W83977 if it is Port 61h with an additional CPLD. Now the problem is corrected. I could'nt see no more KBCLK and KBDAT activities. Keyboard is now fine.

But this is not the wanted solution because of the hardware expense.

I am still waiting for a hint (from you, from Winbond, from ST, ...)

Bernd

moedeker2 · ‎2003-07-29

Posted on July 29, 2003 at 11:24

I got a new Revision ''Rev. B'' of this chip. The old ''Rev. A'' seems to have the port 61h bug.

Everything is fine

Bernd