• Please review our updated Terms and Rules here

XT Not Access Extended Memory Board; Parity Error F000:F9B9

S

SailorEd

Member
Joined
Feb 17, 2015
Messages
20
Location
Recently reloacted to AZ from Thailand
After 25 years in storage, now trying to restore my clone XT. Finally got monitor and keyboard issues resolved and NOW need your help with memory problem. Can not access 384K on extended memory board. When cold booting goes through 256K ram check (on-motherboard) and boots to DOS. "MEM" shows only 256K available. When soft booting; shows error: "I/O CARD PARITY NMI@F000:F9B9". I am at a loss what to do next.
I have removed and replaced 1st bank of chips and first row. Removed and replaced all chips in sockets on board.
When last used years ago, all was working OK. Battery on the board is of course dead.

Not enough memory to run PC tools which is on the harddisk. I do not have 5.25 utility disks.

Motherboard has 256K. Drives consist of HDD and two 5.25 floppy. 384K extended memory board is of unknown manufacturer. Might be Digigraphics.
Both Motherboard and extended card utilize 64K chips; four rows and six rows respectively.

BIOS is Phoneix ver 2.27 Can not get into BIOS. Have tried all common key combinations. Based on name Yangtech, Inc. showing on Boot screen, obvious that builder of unit did there own modifications to the Phoenix BIOS.

Would be most appreciative of advice on how to determine what is problem in accessing the extended memory card. WHAT TO DO?
 
An XT BIOS probably isn't going to have a user menu. Have you tried booting and soft-booting without the memory card? I'm under the impression most XT systems attempt to autodetect the amount of memory, so removing the card shouldn't cause issues.

Based on the information provided, NMI@F000:F9B9 is where I believe the BIOS jumps to on NMI (curious that it's not a standard IBM Entry Point). It would be nice if the PARITY ERROR handler tried to detect the bad bits, but maybe it tried/couldn't in this case.
 
SailorEd said:
Can not access 384K on extended memory board.
Click to expand...
Best not to use "extended memory" because extended memory is memory past the 1 MB address.

SailorEd said:
When soft booting; shows error: "I/O CARD PARITY NMI@F000:F9B9".
Click to expand...
That will be the mutifunction card signaling to the motherboard that a RAM parity error occured, which in turns generates an NMI (non maskable interrupt).

SailorEd said:
I have removed and replaced 1st bank of chips and first row. Removed and replaced all chips in sockets on board.
Click to expand...
I assume that your "replace" is replace-with-same, rather than replace-with-new, i.e. you have re-seated all RAM chips on the card.

SailorEd said:
Both Motherboard and extended card utilize 64K chips; four rows and six rows respectively.
Click to expand...
So the card layout will be similar to the AST SixPakPlus, the one pictured at [here]. Six banks of 64 KB, although the bank and bit order could be different.

SailorEd said:
Not enough memory to run PC tools which is on the harddisk. I do not have 5.25 utility disks.
Click to expand...
If you connect a 1.44M diskette drive up to the floppy controller, there is a good chance that it will read/write 720K diskettes.

The use of ChecKit software per the procedure at [here] may inform you of the address at fault. An address would narrow down the possible RAM chips. I do not know if 256K of RAM is enough for ChecKit.

Because your XT-class machine is counting as far as 256K, I expect that any faulty RAM chip (and there could be more than one) is in RAM bank 0 of the mutifunction card.

Have you tried swapping banks about? After doing so, What you are looking for is for your machine to count up to something more than 256K. Once you have identified the faulty bank, you then start swapping chips about in order to identify the chip at fault.

By the way, there is always the possibility that the RAM chips on the mutifunction card are good, and that some other type of card failure (in the RAM component) is the cause of the symptom.
 
Thanks to all for the comments. I found two ram chip sockets that are damage from battery leakage. Not real bad, but the inside metal contact for a couple of chip pins came out when I was trying to clean the contact area in the socket. Sooo, I ordered a SixPackPlus card from e-bay and installed.
And of course keeping with Murphy's Law; installed board killed system; nothing; no boot.
From searching the site posts, seems the SixPackPlus is know to conflict with NEC V20 which is in the machine. Sooooo, now ordered 8088 intel chip to see what will happen next.
Thanks again for the help.
Ed
 
Moved Memory Chips to New 384K Card; Total Memory only upto 320K

Moved Memory Chips to New 384K Card; Total Memory only upto 320K

Still need your help. Could not just do nothing waiting for 8088 chip to try to get sixpackplus solution to expanding memory to 640K.

Looked through my old boards and found a Digigraphic Systems Extend 384M card; s/n 102793 Assy 30000024B. Look to be in very good condition. Board has one block switch with five (5) on-off toggles. So this is what I did:

1). Removed all 64K chips from old board that I determined had damage from battery that was in machine for last 25 years.
2). Installed all 64K memory chips into Digigraphics card and installed into mainboard.
3). Tried about all possible combination of settings of the five toggles in the one block switch. Most combinations resulted in only the mainboard 256K of ram being recognized.
4). Appears that when only the number three toggle is in the "off" position is any of the card memory recognized. When in this position, a total memory of 320K is shown by the "mem" command. No errors indicated upon boot. Soft boot showed F000:F9B9 same as shown with old board that had battery damage.
5). When #1 position is "off" an error "FF RAM BAD 20833" was shown.
6) When the #2 position is "off" the error "FD Ram BAD 308BA
In both cases 5 and 6 the system hung-up.
In some other trials with 2 and 3 off the following errors where experienced:???:2 off-192K shown on boot; error FF Ram Bad 3000B
3 off-283K shown on boot; errors FF Ram BAD 46F10, FF Ram Bad 42A18, FF Ram Bad 41A13

Is it correct to think the block switch is referencing the mainboard memory installed and the 3off relates to the 256K on the mainboard.

Is it correct to think that the #3off is only recognizing 320K due to bad memory chips in the subsequently accessed bank.

Are banks of chips accessed from the lowest numbered bank to the highest?

Are the errors referring to bank and chip locations that are bad?

Is there a practical way to check memory other than moving between sockets; banks etc.?

Thanks again for your help. I underestimated the problems associated with getting the old girl up and running.
 
SailorEd said:
From searching the site posts, seems the SixPackPlus is know to conflict with NEC V20 which is in the machine. Sooooo, now ordered 8088 intel chip to see what will happen next.
Click to expand...

I think the only way a board would not be compatible with NEC V20 CPU is if board's firmware relies on an instruction that works differently on NEC V20 vs. 8088. Indeed there are a few instructions that work a bit differently (e.g. AAD, or REP with double segment override prefixes). But: (1) AST Six Pack Plus doesn't have a firmware; (2) I doubt anyone is using REP with double prefixes, and there is a very low probability of using AAD instruction with argument != 0Ah.

So I think your SixPackPlus might be damaged and causing a short circuit.
 
sergey said:
I think the only way a board would not be compatible with NEC V20 CPU is if board's firmware relies on an instruction that works differently on NEC V20 vs. 8088. Indeed there are a few instructions that work a bit differently (e.g. AAD, or REP with double segment override prefixes). But: (1) AST Six Pack Plus doesn't have a firmware; (2) I doubt anyone is using REP with double prefixes, and there is a very low probability of using AAD instruction with argument != 0Ah.

So I think your SixPackPlus might be damaged and causing a short circuit.
Click to expand...
I think that he found my web page at [here]. The power supply runs (producing POWER GOOD and voltages) but the 5160 motherboard appears dead.

It is probably related to the cause described at [here]. Essentially, the POST in the 5160 is being run faster than IBM anticipated.

The problem disappeared when I upgraded the 5160's BIOS to the third revision (01/10/86), a BIOS in which IBM added a delay:

LODSW ; ALLOW RAM CHARGE TIME
LODSW
LOOSW
LODSW
 
SailorEd said:
and found a Digigraphic Systems Extend 384M card;
Click to expand...
As you are discovering, a lack of documentation is a curse.

SailorEd said:
Is it correct to think the block switch is referencing the mainboard memory installed and the 3off relates to the 256K on the mainboard.
Click to expand...
The RAM related switches on the card set at least two things:
* The address at which the card's RAM is to start at;
* How much RAM is on the card (because maybe not all banks are populated)

Some RAM cards may also have switches to:
* Turn parity checking on or off
* Bank size (for cards that allow different sized RAM chips to be used)

And of course, if the card is providing additional functionality beyond RAM, then there could be switches that affect the additional functionality.

The documentation for your card would reveal the switch settings start the RAM at 256K, and the switch settings to inform the card that all 384K is fitted.

SailorEd said:
Is it correct to think that the #3off is only recognizing 320K due to bad memory chips in the subsequently accessed bank.
Click to expand...
That is one possibility.
Another is that the card's switches are not set properly (e.g. card configured to provide only one bank's worth of RAM [64K]).

SailorEd said:
Are banks of chips accessed from the lowest numbered bank to the highest?
Click to expand...
Usually, but I remember that someone on this forum had a card that was opposite.

SailorEd said:
Are the errors referring to bank and chip locations that are bad?
Click to expand...
If a RAM error is reported, then usually the two facts that are provided are:
* Address of failure (address as the CPU sees it)
* Bits in that address that are not as expected
 
Thanks for inputs. My BIOS is a Phoenix ver 2.27 per on-line data I found is suppose to be date 29 May 86. I have not seen name on the motherbaord; it is a XT configuration with one switch block; 8 ROM sockets and 8 bus expansion slots. When I figure out how to dump the BIOS I will look for the delay statements you referenced. Then will get back to the SixPackPlus card. Seller told me it was tested and working.

On another note, FYI. I found new 64K chips in another main board I had and put them in the Digigraphics card (first two banks) and it recognized the memory OK; main and card. Sooo, I obviously had bad chips causing all the problems with the Digigraphics card memory not being accessed.
Regards
Ed
 
SailorEd said:
On another note, FYI. I found new 64K chips in another main board I had and put them in the Digigraphics card (first two banks) and it recognized the memory OK; main and card. Sooo, I obviously had bad chips causing all the problems with the Digigraphics card memory not being accessed.
Click to expand...
Good to hear you sorted that out.

Although you do not have an IBM XT (IBM 5160), the web page at [here] shows that faulty RAM and faulty tantalum capacitors are common in the IBM PC family of computers.

SailorEd said:
Then will get back to the SixPackPlus card. Seller told me it was tested and working.
Click to expand...
I presume that is the SixPackPlus that you write of at [here]. I added comments.
 
modem7 said:
The use of ChecKit software per the procedure at [here] may inform you of the address at fault. An address would narrow down the possible RAM chips. I do not know if 256K of RAM is enough for ChecKit.
Click to expand...
Stone said:
I've always needed more than 256K to run CheckIt.
Click to expand...
Because of the high RAM requirement of CheckIt, I have updated my procedure to include the optional use of DEBUG instead of CheckIt.
 
Thanks again for all the help. Glad to report that finally got everything running including the Backpack parallel drive for 1.44 disks. One of the forum members provided me with driver on 5.25 disks so I could install it.
Regards,
Ed
 
You must log in or register to reply here.
Back
Top