Finally figured what the error with my XT was.

About a year ago, I told about some errors I got with my XT. The symphtoms was that there was no BEEP on startup, and keyboard-input would act strange.

After going through the BIOS listings, I figured what's wrong. Here is the parts of the code that's acting different:

This simply means that somehow (in cold weather), the keyboard data line gets grounded (maybe something is shorted or a pullup resistor is loose?), causing the BIOS to skip the keyboard initialazion and diagnostic tests.

Thus, the keyboard buffer isn't set up, and typing what you want becomes a real pain.

*Edit*
In adition, the BLINK_INT routine toggles the keyboard clock on and off every now and then (bit 7 of I/O port 0x61h). It is obivouls that IBM meant to use the keyobard plug for something else under the MFG_TST=1 condition.

per said:

Thus, the keyboard buffer isn't set up, and typing what you want becomes a real pain.

*Edit*
In adition, the BLINK_INT routine toggles the keyboard clock on and off every now and then (bit 7 of I/O port 0x61h). It is obivouls that IBM meant to use the keyobard plug for something else under the MFG_TST=1 condition.

Click to expand...

Sorry, my bad. The keyboard stack IS set up correctly even though the tests are skipped.

But, the keyboard clock still toggles between enabled and disabled about 18.2 times each second.

I'm happy that this only happens very rarely (long time since last time now).

*In Addition*
The reasons for this might be many:

1. Shorting wire(s)
2. Loose pullup resistor(s)
3. Minor failure in the keyboard logic
4. Minor failure in the keyboard interface
5. Unepected memory event
6. Much more

While I feel for your plight, I think that's one of the coolest hardware issues ever. Most hardware issues prevent the box from working at all. Yours just managed to piss you off

Trixter said:

While I feel for your plight, I think that's one of the coolest hardware issues ever. Most hardware issues prevent the box from working at all. Yours just managed to piss you off

Click to expand...

Yea, I know.
After my analysis, what's happening in hardware when the issue occours is that the byte at a certain location becomes 01 when it should have stayed 00.

The reasons for this might be many:
Most likely a grounded keyboard line = loose pull-up/down resistors, or simply a short.
Possible bad signal = A loose contact from/to the PIO, or the shift register.
Problably not a DMA issue = Expansion card error.

per said:

It is obivouls that IBM meant to use the keyobard plug for something else under the MFG_TST=1 condition.

Click to expand...

This was discussed in posts 21 through 24 in an earlier thread: http://www.vintage-computer.com/vcforum/showthread.php?t=12377