• Please review our updated Terms and Rules here

ISA 16 bit Memory Card - Help!

pauls640

Member
Joined
Sep 24, 2013
Messages
48
Location
Italy
Hello everyone,

I have this old ISA card sitting in my shelves for years now, and no idea how to make it work, despite my noble efforts :). The difficult thing is, the marks on the card don't help much, at least for me :(. In fact, this thing is unknown to the entire world, if you google it, you find a post of another guy asking for info, and nobody answered.

As you can see, the only thing written on it is "Parity ram card", plus some id numbers, like "W5 500". The card comes from 1987, it has four banks of 18 chips each, and a dip switch on the right.

My thought is that it might be an expansion card for an OEM PC, probably Olivetti, since it has quite a number of memory onboard and there are no marks on it except part numbers. Maybe belongs to a M290?

Obviously, the card doesnt work on its own, but the IC's heat up when running on a PC, so it's probably working. It's very well preserved, no oxidation etc, looks new.

Do you have any idea to id the card?

Needless to say, my goal or mission impossible (depens on how you see it :D) is to make it work on a 286 system undergoing restoration.

2013-10-10 15.14.52.jpg
 
Last edited:
Sure,

here are some closeups, I'm also writing down all I can see on the card:

On the far left, next to a bar code the following chars : "W5 500 013397"

2013-10-10 19.49.32.jpg

On the left, below bank 0: 6450343 A38431X 7777 PAINTED in white:

2013-10-10 19.50.10.jpg

Next to it, "Parity ram card"

2013-10-10 19.51.07.jpg

A general view of the IC's on the right, written in white paint '3487'

2013-10-10 19.50.34.jpg

On the back, hope is visibile, anyway: VA-HL 94-V0 P/N 6480183 297.

2013-10-10 19.54.03.jpg

Then, always on the back, but on far right, two letters, 'E.T.'

Hope this helps.
 
The 6450343 is the same number as the 512k - 2048k memory expansion for 5162 & 5170 from IBM.
That same number cross-references to 59x7294

The 6480183 seems IBM-like for a part number, many parts for the 5170 have a part number that starts with 648.

Can't locate a picture or jumper info of the IBM part number 6450343 or 59x7294.
 
Right, the p/n is slightly different, but the layout in the pdf and the dip switch are there!! Well done!! :D

But... what now? the dip switch sets the starting address of the card, what does that mean? That means if I have 4MB of Ram onboard, I set the switch at 4? Then the pc automagically counts the extra Ram at boot?

I tried configuring the dipsw on a 386 test box I've got. This box has 4MB of onboard ram, so I set the card at 4. Nothing happens apparently, BUT as in techref, Cmos byte 31 shows 1101, so 13.:confused:

This thing needs a driver to work?
 
This thing needs a driver to work?

If it needs a driver it will probably have been provided in IBM PC-DOS, they were sneaky like that.

Addendum: possibly XMA2EMS.SYS
 
Last edited:
I have an original copy of PC-DOS, it's version 5.02, though. Probably it's too new to work. I'll check the docs.
 
Here (hope that helps - no idea what version it is!).

Thank you, looking around, xma2ems.sys should be the one to read xma memory from the board, but I tried to boot up, and it says 'cannot find adapter'

Can someone explain the dip switch thing? I bet I did not configure it right. I have configured it at 4mb, because my onboard ram is 4mb. Is this correct? And besides, can this card work on a 386?
 
Can someone explain the dip switch thing? I bet I did not configure it right. I have configured it at 4mb, because my onboard ram is 4mb. Is this correct?
It might depend on how that 4 MB is configured. Is it contiguous through 4 MB or is there a hole between 640 KB and 1 MB? Some boards give an option on this. A simple test of this would be to configure the card to start at 4 MB plus 384 KB instead of 4 MB and see if that makes a difference.
 
Last edited:
This may be the issue, taken from the PDF Trickster referenced in Post #7

Code:
The address of the first expansion memory option must start at 
address space hex 100000.  If additional memory expansion 
options or IBM Personal Computer AT 512KB Memory Module 
Kits  are installed, no gaps between memory are allowed.  All 
expansion memory must be one contiguous block starting at 
address hex 100000.

Seems very specific where the initial starting address must be.
 
So, according to that PDF, this board which was specifically designed for the IBM AT can only be used in a machine that has 1 MB of onboard memory.
 
So, according to that PDF, this board which was specifically designed for the IBM AT can only be used in a machine that has 1 MB of onboard memory.
The IBM document has to be be put into context. IBM produced this RAM card for the later model IBM ATs, not for AT clones. IBM expects only IBM harware. In a late model IBM AT, IBM is expecting you to fit only this particular IBM RAM card (one or more of them) to get extended memory.

So, knowing that, then IBM's view is that the first of these cards must start at address 1 MB, because in IBM's view, no other type of RAM card would be in the AT providing extented memory.
 
This particular card provides 2 MB of memory, which can only be used as extended memory.
Extended memory starts at the 1 MB address mark and must be contiguous (i.e. no gaps).

Examples:
* If your AT presently has no extended memory, then this card would be configured to start at the 1 MB address mark.
* If your AT already has 2 MB of extended memory (which would occupy memory between the 1 MB mark and the 3 MB mark), then this card would be configured to start at the 3 MB address mark.

Your 386 has 4 MB of RAM. With 4 MB of motherboard RAM on a 386, I would be expecting 3 MB of that to be configured in the BIOS SETUP as extended memory, in which case, the IBM memory board would be configured to start at the address of the end of the existing extended memory (at address of 4 MB). But maybe your 4 MB of motherboard RAM is configured in a non-standard way.

No driver is required for the motherboard itself to recognise the RAM on the IBM card. Once all (BIOS SETUP plus RAM card switches) has been configured properly, I would expect you to see the memory count at power on time reaching 6 MB.

And besides, can this card work on a 386?
The RAM on the IBM card is probably speed rated at 150 ns. Read your 386 motherboard manual to see what it has on the subject of RAM speed.
 
Thank you all. To make things simple, I installed the board on my 286 box going restoration. It has 1MB onboard, makin 640 base + 384k extended (machine counts this at boot).

Following these instructions,

http://www.minuszerodegrees.net/5170/cards/5170_memory_board_512K_2048K_switches.htm

I configured the dipsw to 1mb, so 'on on on off on', but ram is not seen at boot. The MoBo(wd286-wdm20) has a Phoenix BIOS, that lets you specify manually the amount of RAM, so I set 640k base + 2432K Ext.

When machine boots with PC-Dos, only the first 384k are recognized by himem.sys, whereas xma2ems.sys still says it cant find adapter... but Msd reports correctly 2432k of ext memory, probably because it just reads it from bios...

Do you think my dipsw configuration is wrong?
 
If your 286 had 384 KB of extended memory to start with, then the IBM RAM card needs to be configured to start at the end of that existing extended memory, i.e. at the 1408 KB address (1024 + 384).

But it sounds like something more fundamental is going on. Maybe the IBM RAM card is faulty. Switch settings is another possibility. We have seen a few people set switches to on when in fact they were setting them off, and vice versa (see [here]).

Unfortunately, I do not have the subject RAM card. Otherwise I would have put it into one of my clone AT machines (286/386) just to verify what I've written.

but Msd reports correctly 2432k of ext memory, probably because it just reads it from bios...
Probably.
 
I tried an extended memory test through PC-check, interestingly when checking the starting address of the card (100000 in this case or 160000), it throws an error.

I think that this software actually "talks" to the card, because after throwing an error, it goes on checking the rest of the memory (set in the Bios), and this test goes slower than the onboard RAM.

If you remove the card, BUT keep the memory size unchanged, the test goes faster and without errors (dont know why it does not complain, anyway)

So I believe something is working out, but I'm missing something from the picture.. :confused:
 
Last edited:
Back
Top