DEC 3000/500 - Is my memory borked?

[camerongray]

I recently (as of yesterday) got a DEC 3000 Model 500 that had been sat without being switched on for at least 15 years. This is my first ever DEC system (most of my experience before this was with Sun gear so it's all pretty alien to me!) I was finally able to get console out of it however when running the "test" command I get an error as shown below. Does this indicate a problem with one of the RAM sticks (I have 8x8mb for a total of 64mb) or could the issue lie somewhere else in the system?

Thanks,
Cameron

Code: Select all

>>> test
T-STS-ASIC - OK                                 
? T-ERR-MEM - MCHK reading Addr = 03FFFFFCd AAAAAAAA Wr AAAAAAAA               
? T-ERR-MEM - bad page = 1FFF
T-STS-MEM - Uncorrected Error count = 1                                       
T-STS-MEM - Corrected Error count = 00000001
LAS_biu_stat  = 00000000.00003340  LAS_biu_addr  = 00000001.D4000018
LAS_fill_synd = 00000000.00000026  LAS_fill_addr = 00000000.014502E0
LAS_fadr      = 00000000.02000010  LAS_tc_err    = 00000000.0F000000
T-STS-MEM - Total Pages = 2000   Tested Pages = 2000   Bad Pages = 2
? T-ERR-MEM - FAILED status = 50 test Cell addr = 001081e0
?? 808       MEM 0x0050


  84 FAIL

>>>


[robespierre]

MCHK means Machine Check-a low level hardware fault. Check to see that your RAM risers and the I/O daughterboard are properly seated.

[camerongray]

Thanks!

I reseated the I/O daughterboard and interestingly in the initial self tests I now get "MEM ?? 808 0050 64MB" whereas before I simply got "MEM OK 64MB". I then reseated all the memory riser cards (putting them all back in different slots to see if it would change the error message) - The message is still the same.

Interestingly now when I run "test mem" I don't always experience the full "MCHK reading Addr... Uncorrected Error count" stuff". Sometimes it simply results in the "Corrected Error count" message and the lines afterwards.

What parts should I be suspicious of here? Is there any way to narrow it down as I'm currently suspecting the SIMMs, MMBs, System board and I/O board? - Can't really just replace all of those!

[urbancamo]

Try running with the minimum memory. You need 8 simms for this in the lowest numbered slots

[camerongray]

Unfortunately that's all I have, 8x8mb sticks, all in the lowest slots.

I don't have the documentation for the Model 500 however I found the Model 400 "System Service Information" which states that the "808" code means that the issue is with both SIMMs 0 and 1 in Bank 0. I swapped all the MMBs (including their respective SIMMs) to different slots and the memory test still gives "808" - Does this give any insight into the issue?

That said, if anyone has a copy of the service/hardware information for the Model 500 (or I suppose any pedestal DEC 3000) then that would be greatly appreciated, all I can find online for them is the user guides which don't give much in the way of diagnostic information.

[urbancamo]

I think all the 3000 series have the same memory apart from the 400. I should be able to sort you out with some memory if you need some

[camerongray]

Thanks, that would be awesome! I've dropped you a PM.

[smj]

Just in case anybody pulls this up later -- the memory for the 3000/300__ models is 72 pin parity SIMMs, compatible with the Indigo2 and other systems of a similar vintage. That is not the case for the other 3000 models, IIRC they all use proprietary modules. (For corroboration of that memory see viewtopic.php?f=4&t=16731097)