PDA

View Full Version : XT clone memory chip error



romanon
July 11th, 2014, 08:26 PM
Hi, i have my Czechoslovakian XT clone with memory error message, which one chip is bad? Thanks.

P.S. parity chips are on middle line, bank 0 and 1 are 256k, bank 2 and 3, 64k

modem7
July 11th, 2014, 11:23 PM
Bank 0: (256 Kbit chips) addresses 0 to 255 KB.
Bank 1: (256 Kbit chips) addresses 256 to 511 KB.
Bank 2: (64 Kbit chips) addresses 512 to 575 KB.
Bank 3: (64 Kbit chips) addresses 576 to 639 KB.

The "3C000" on screen will be the address of the failure. 3C000 hex equates to address 240 KB, which matches the "240 KB OK" (RAM tested sucessfully so far).

The "04" is the error bit pattern. 04 (00000100 binary) indicates a single bit in error, bit 2.

So the faulty chip should be the MD2 one in bank 0.

romanon
August 19th, 2014, 05:26 AM
Thanks!
Chip removed and changed. Works fine now ;)

romanon
August 20th, 2014, 11:27 AM
Ehm, one problem here, i have enabled all 4 banks but system reports only 512k of memory (without any errors). If i enable only bank 0, reports correctly 256k, if 0-1, 512k. But when aj enable 0-2 or 0-3 still have only 512k, where can be problem? Thanks

bobba84
August 20th, 2014, 02:40 PM
From what I've heard some BIOS'es run an initial check for RAM size by scanning the end of each block (I think) and if it doesn't find it it assumes it's not there so doesn't test it. Chances are you have some bad RAM in at least bank 2.

By the way, nice looking machine!

romanon
August 20th, 2014, 10:03 PM
Thanks :)
But how can i know, which chip is bad?

cr1901
August 21st, 2014, 02:42 AM
Thanks :)
But how can i know, which chip is bad?

Consider taking a known good DRAM chip (preferably a 41256, since it can safely substitute a 4164) from another board, and stacking it on top of each RAM chip in bank 2 (and 3), resetting the board each time. When stacked on top of a bad chip, the good RAM chip should accept the writes and reads from/to the CPU thanks to a closed circuit between the pins of both chips.

Whenever the error disappears (i.e. all RAM is counted properly), you've found the bad chip :D, at which point you should swap the two.

Also, if you haven't done so already, make sure to double-check the system board switches. Banks 0 and 1 are most likely fine if they pass POST as is.

romanon
August 21st, 2014, 04:00 AM
Consider taking a known good DRAM chip (preferably a 41256, since it can safely substitute a 4164) from another board, and stacking it on top of each RAM chip in bank 2 (and 3), resetting the board each time. When stacked on top of a bad chip, the good RAM chip should accept the writes and reads from/to the CPU thanks to a closed circuit between the pins of both chips.

Problem is, that all ram chips are soldred...



Also, if you haven't done so already, make sure to double-check the system board switches. Banks 0 and 1 are most likely fine if they pass POST as is.


I checked it, maybe is someting wrong with switches on memory board, but i dont have scheme for it :(

bobba84
August 21st, 2014, 04:30 AM
You don't need to remove the chips to stack a second one on top. Just make sure it is the right way around and place it on top. It will in effect 'help' the bad chip to work.

romanon
August 21st, 2014, 05:09 AM
ok i tried change dip switch setting on memory board, i dont have any scheme so i only guess combinations..
01001111 - 512k
11001111 - 256k
11101111 - 256k
11111111 - nothing
00101111 - 128k

ehh it can be so many combinations...

3pcedev
August 22nd, 2014, 01:32 AM
256 combinations to be precise