Strange Problem with a C128

Hello,

today I tried to fix th C128 of a friend. The system is unstable and 
when checked with a diagnostics module (C128/128D diagnotic Rev 785260) 
it reported 7 chips in the second RAM bank (U46-U52) as bad.

To make sure the module works, I used it with one of my C128 and there 
it reported no dead RAMs, all banks came back OK.

The same system had 2 dead RAMs (MT4264) in the first bank already 
replaced, so I thought that it's just more of the same and they just 
used a bad batch of RAM back then when making it. Most of the time the 
MT4264 shows a strange fault when going bad, you get one stuck bit every 
256 Bytes, as if one column died. The rest works fine.

Well, after replacing all 8 RAMs (U46-U53), the module still reports 
dead RAM (now U46, U48, U52 and U53). Since I used sockets I can swap 
the chips around. The fault sometimes moves, sometimes it doesn't but 
any given combination of RAMs (KM4164-12 and U2164) will always give the 
same fault pattern, no matter how often I run the test. Using any of the 
RAMs in the lower bank never results in any fault so I think all the 
chips are good.

Then I decided to go for broke and remove one of the RAMs from the upper 
bank and try again. The RAM test didn't show it as bad. Even removing 
all the RAM from the upper bank will not show all of them as bad.

Does anyone here have any idea about what could cause this type of 
fault? By now I'm pretty sure it's not the RAM.

I would also like to do my own tests from the machine language monitor. 
Can anyone here give me the address space used by the second RAM bank? 
I'd like to fill it with a few patterns of my own.

  Gerrit


       Message was sent through the cbm-hackers mailing list