• Please review our updated Terms and Rules here

IBM 5170 motherboard issue

romanon said:
ok, my POST card finally arrived.
Click to expand...
You have the same card as me. Mine was made six months earlier.

romanon said:
shouldnt be 3,3V light off?
Click to expand...
The 3.3V LED on my card also lights when the card is in an ISA slot. Design flaw?

romanon said:
There is result
Click to expand...
Your photo shows that code 0C was the last code output by the BIOS. 0B (displayed as '0b') was the code output before the 0C.

The codes output by the IBM 5170 BIOS/POST are shown in the 'IBM AT BIOS' section of the web page at [here].

Based on the hypothesis that the CPU was not returning from protected mode (based on what the SuperSoft/Landmark diagnostics showed), the last code that I expected you to see was 20. Instead, you are seeing the earlier generated code of 0C, a code associated with the keyboard controller chip on the motherboard.

The keyboard controller chip is normally an 8042 (i.e. ROM type), but on some motherboards, such as yours, it is an 8742 (i.e. EPROM type). A photo of an 8742 on a 5170 motherboard is [here].

Back to the POST code of 0C.
From my reading of the POST programming in the IBM BIOS chips, the sequence is:

...
1. Output POST code 0B
2. Send command of AA to 8042 keyboard controller chip, a command to perform a self test, then wait for response.
3. Output POST code 0C
4. If response is not 55 (55 = self test passed), halt the CPU, otherwise continue.
5. Get state of display switch from 8042 keyboard controller chip.
6. Send command to 8042 keyboard controller chip to "WRITE BYTE 0"
7. If command not accepted, output POST code 0D, then halt the CPU.
8. Output POST code 0E
...

So, the fact that the last code you see is 0C, suggests to me that the keyboard controller chip is either:
* responding to the 'perform a self test' command, but the response indicates that the self test failed; or
* not responding at all.

The 8742 chip on your motherboard appears to have an uncovered window. That is not normal.


SUPERSOFT/LANDMARK ROM

Looking at the manual for the Supersoft/Landmark ROM, I see that (unlike the PC/XT version) the AT version of it does NOT have a 'keyboard controller test'.

Maybe during its PROTECTED MODE CPU test, it is issuing a 'take line P20 low' to the keyboard controller chip (see [here]) in order to reset the CPU (to bring the CPU out of protected mode), BUT in doing so, is assuming that the keyboard controller chip is fully functional.
 
ok thanks, sooo what is next step what should i do?
This is my keyboard controller chip.
fotka.jpg
 
romanon said:
ok thanks, sooo what is next step what should i do?
Click to expand...
It is not straightforward. What we have is a hypothesis. If it was my motherboard, I would be using a logic state analyser to prove/disprove the hypothesis, because I do not like to replace chips unless I am very confident that they are faulty.

Some people who have read the last few posts may be saying to themselves, "The combined facts of a 0C error code and an uncovered window on the 8742, suggests that over many years, some/all of the EPROM contents of the 8742 have been lost." The 8742 may be the problem cause, but maybe it is not.
 
Ok i know. But where i can buy a new chip? On ebay are only erased D8742 chips and i dont have a programming machine for this IC's...
 
Isn't there supposed to be a sticker covering the glass window on that 8742? Maybe the contents were flushed over the years from exposure to daylight.
 
archeocomp said:
Willem EPROM Programmer can handle 8742, but it is not cheap
Click to expand...
And for the 8742, the Willem units require an MCS 48 adapter (see [here]).

The 8742 consumes quite a lot of power. This means that the Willem will need to powered via an external power supply, not via USB power.

The Willem's jumpers need to be set so that the programming voltage is 21V. Also, to compensate for a diode in the MCS 48 adapter, the Willem's 5V/5.6V/6.2V jumper set needs to be set to 5.6V (or possibly 6.2V).

An 'IBM 5170 keyboard controller chip' image to use for programming is at http://www.minuszerodegrees.net/rom/rom.htm
 
The NEEDHAM EMP-20 EPROM burner seems to be capable of programming the 8742 using the 02A plugin module.......
 
modem7 said:
An 'IBM 5170 keyboard controller chip' image to use for programming is at http://www.minuszerodegrees.net/rom/rom.htm
Click to expand...
archeocomp said:
:) Excellent!!
Click to expand...
The image there (thank you to member Per) is of the keyboard controller on a type 1 motherboard. It should work on a type 3 motherboard, but for a while I have been planning on adding the equivalent type 3 image. I will do that very soon.

modem7 said:
And for the 8742, the Willem units require an MCS 48 adapter
Click to expand...
archeocomp said:
I have one (the adapter), but not the programmer, so I could be of help.
Click to expand...
If a Europe based 8742 solution does not eventuate, then I could post a programmed 8742 from Australia (I have lots of blank 8742s).
But maybe the OP does not have soldering skills/tools.
 
modem7 said:
If a Europe based 8742 solution does not eventuate, then I could post a programmed 8742 from Australia (I have lots of blank 8742s).
But maybe the OP does not have soldering skills/tools.
Click to expand...

I have some soldering skills but only little. This is not problem, because i know people with this skills. Problem is that i dont have any 8742s and i dont know anyone with programming machine.
It would be a problem for you make one 8742 for me and ship it? Thanks
 
modem7 said:
The image there (thank you to member Per) is of the keyboard controller on a type 1 motherboard. It should work on a type 3 motherboard, but for a while I have been planning on adding the equivalent type 3 image. I will do that very soon.
Click to expand...
I removed the keyboard controller chip from a 5170 motherboard of type 3, and then used my Willem EPROM programmer to read the ROM contents of the chip. The code turns out to be the same code as what Per supplied to me as code for the keyboard controller on a type 1 motherboard. I think it fair to assume that the same code will be found in the keyboard controller on a type 2 motherboard.

My type 3 motherboard now has a socketed keyboard controller chip. With that chip out, and a POST card plugged into the motherboard, a power-on of the motherboard results in 0C being the last POST card shown.

Allowing for possible future demand for replacement chips, I programmed the code into four 8742s (testing them all in my motherboard).

romanon said:
I have some soldering skills but only little. This is not problem, because i know people with this skills.
Click to expand...
That is good, because I have found that compared to modern PCBs, old PCBs are more easily damaged by poor soldering technique. I recommend that the replacement chip goes into a socket.

romanon said:
Problem is that i dont have any 8742s and i dont know anyone with programming machine. It would be a problem for you make one 8742 for me and ship it? Thanks
Click to expand...
Yes, I can do that.

I need to remind you though, that although the keyboard controller chip on your motherboard is highly suspected of being faulty, there is no guarantee that it is.

I will contact you via PM.
 
Do you happen to have a 286, 386, 486 clone mainboard with socketed keyboard controller? Might be worth to try such a chip first (in a socket of course).
 
Tadaaa, problem solved, BIG THANKS to Modem7 for his right diagnosis and sending keyboard controller chip to me ;)

Possible Cause of failture of that chip was removed sticker from the top, and possible exposure to UV light from daylight...who knows...
 
You must log in or register to reply here.
Back
Top