I have a 5162 board in similar condition. It physically looks okay, but I think one of the ICs is dead. I recommend starting off by buying a POST card to see what kind of error codes the board throws up.

I read that at least in case of 5170 any dead memory chip in bank 0 may cause the board to look dead. I think I will start with trying to replace it and see what happens.

I have a postcard in my another place. Will try to do some tests - just afraid to get 'no reading' at all, as in case of most of my dead boards.

if a dead chip in bank one does this, switch the ram around. Thats what I would do first.

Luckybob, I think until I get a set of chips that are all known to work properly, swapping them will not work here - If I get it right, it could serve to locate the bad chip if you would be getting any feedback from the motherboard whatsoever. And that's not the case.

Here's what I found about 5170. Please note that this is a different motherboard.

RAM failure in motherboard bank 0 […]
Show full quote

RAM failure in motherboard bank 0

Vintage RAM chips have a relatively high failure rate.
The failure of any chip in the 5170's first bank of RAM (bank 0) results in what appears to be a 'dead' motherboard.

Note that rather than a RAM chip actually failing, what might have instead happened is that a RAM chip has developed poor electrical contact with its socket.
And so the first thing to try is to simply to reseat all of the chips in bank 0 (wiggle each chip in its socket).

For the 5170, one way of diagosing a faulty RAM chip in bank 0 is by the use of a POST card (set to port 80h). If there is a failure in bank 0, the POST code of DD will be shown on the POST card. A breakdown of POST code DD is shown here.

If your 5170 motherboard is of type 1, then it has two banks of RAM, and therefore what you could try is swapping the bank 0 and bank 1 chips. If the motherboard then 'comes to life' and generates a 201 error on-screen, then a faulty bank 0 chip was the problem.
If your 5170 motherboard is of type 2 or 3, then it has only one bank of RAM. Swapping chips within the bank will not identify a faulty chip (though the removal/insertion of chips may fix a poor electrical contact problem).

In my case I can't swap memory between bank 0 and 1 - they are different types of memory. Bank 0 is 16/18 pin DIP and Bank 1 accomodates SIMMs.

BTW - I have posted a similar thread on VCFED, as I know they specialize in this kind of lagacy hardware. This is my fourth thread there. The thing is that they never seem to answer to me. I guess you need to be some kind of "elite" to even get noticed. Or am I doing something wrong?

That site is a bit slower. most of my stuff goes unanswered sometimes.

I've considered using Cunningham's law to get some traction, but I'm too lazy.

edit: https://meta.wikimedia.org/wiki/Cunningham%27s_Law

luckybob wrote:

That site is a bit slower. most of my stuff goes unanswered sometimes.

Yes, I guess I'll just stop asking there.

I've ordered some memory chips.

There are two types that fill Bank 0:

HM50464P-15
HM50256P-15 (parity bit)

I found exactly the same parity bit chips locally. Finding HM50464P-15 is much more difficult. I see those are simply 64K/4 DRAM modules. And some 5162 boards use D41464C-15 (NEC) instead (photos on the web).

I found another make: P51C259L-15 by Intel. From the specs I see they are also 64k/4 DRMA modules with 150ns. Would you assume they're pin compatible with former chips?

I don't know for sure, but I would bet any chip with the same pinout and configuration would be the same. If you can find spec sheets it will be made obvious. I do remember the 41464 being a very common number in different chips from different manufacturers.

I still did not reseat the cpu itself. Is this worth trying? We'll see. I have tried another set of DRAM but to no avail.

This board makes me a bit upset. I really wanted this to be working.

Anonymous Coward wrote:

I have a 5162 board in similar condition. It physically looks okay, but I think one of the ICs is dead. I recommend starting off by buying a POST card to see what kind of error codes the board throws up.

You said your board i's dead. Was it throwing any post codes at all? Did you tried fixing it? Any hypothesis you have?

I have the following thoughts:

- need to try another set of 256kb 9 chip simm modules while disabling the second bank (128kb dip-based) with jumper 10.

- is that possible that this particular board NEEDS 287 to even work? I doubt that, but could not find any answer in the technical reference manual.

- is that possible that the board will only work with the original IBM XT 286 PSU for some reason? I read that it has some "Power Good" signal that is being sent to the board logic. Is this something exclusive about this PSU? I have one in the original case but I was performing my tests with some generic at PSU. Can this be the reason for my failures to get it going?

If the POST card displays nothing I would start with the BIOS chip. Is the sticker on it still intact?

I did not test it yet with the post card as I need to travel about 90km to get it from my house in the country-side where I keep all my hardware. Will do that over weekend.

The stickers on the BIOS chips (there are two in this model - even and odd) are intact.

There is something about the PSU - I will continue all my tests using the original one as it seems to be sending some signal to the board. The technical reference manual states that it somehow interferes with onboard logic via mentioned "power good" signal. Can't verify other PSU don't do exactly the same thing though (UPDATE: it rather seems that power good signal (orange wire on J8 plug) is typial for all at PSUs).

I have yet another question - there are some alternative BIOS roms said to be fully compatible with IBM 5170 (AWARD/AMI/QUADTEL):

http://www.minuszerodegrees.net/bios/bios.htm#5170

The site does not say if those ROMS will work with 5162 - has anyone tried?

I'm still trying to figure out what may be the problem with the system board.

Yesterday I tried the following:

- I have burnt two 27256 eproms with Supersoft/Landmark diagnostic ROM. It is not mentioned if it works with 5162, but as it does with 5170 and even some 386/486, why wouldn't it. I connected the speaker and the video card (I only have VGA around - so I wonder if this may be the problem as the ROMs were not designed for VGA and VGA does not work - https://ibm.retropc.se/supersoft_landmark/515 … 64KB-256KB.html). These attempt turned out to be futile. There was nothing on the screen and there were no beeps at all.

- I have tried substituting SIMM modules with some different ones. Nothing changed.

As it's close to the weekend, tomorrow I will have opportunity to use my post card. I will also try to hook it up to my CGA 5153 monitor and try different RAM modules.

If this brings no results (--.--) reading on the post cards which seems most probable, I guess I will send the board to someone with more knowledge.

I really wouldn't want to give up on this particular board.