Failed sound memory test...

[lubb]

I have tested the EIIIX's upgraded sound memory (now 32MB) with the EIIIX's internal diagnostic tools... The first five cycles passed, but the next stage, when the memory was filled with combined values, the verification process always stops at 74 %, then displays some addresses... When canceling this, the memory is reported as "bad"... I tried to re-seat all the ZIP modules, but without success..

I wonder if someone could point me how to identify the bad DRAM ZIP memory module (each one is 512k bytes, there are 64 of them, in total.)

Many thanks.

[dr.c]

Here you, and note this, all of you.

Swap a quarter of the memory to another quarter.
Lauch test.
If the diagnostics is still the same, this part of the memories is good (a half).
Do the same with the other two quarters.
If the diagnostics is still the same, this part of the memories is good (a half). If this is the case, go and f.. yourself..  (just jocking).
If the diag is changed, it means that one of the remaining quarter of the memory is bad. Divide this quarter in two, remake the test, etc....
At the end, you will be with two chips, not knowing wich one is bad. Swap one to another anywhere. If the diag is the same the bad one is the remaining.
Take the bad one and swap it tyo any other one. If the diag changed, its really the bad one. If not, go back to step 5
 

[lubb]

... but tell me, are there chances that the poor ZIP modules will keep any little legs after completing this tormenting diagnostic procedure..? I mean, the modules are so firmly sitting in their sockets, and their legs are so fragile... I wonder what tool should be used to extract them..? some usual IC extraction tool or some specialized tool for ZIP modules..?  

I extracted the modules with surgical tool called pean, but it is  somehow less than ideal...

I have also tried to discover, how is the "bad memory" actually manifesting... I loaded the whole memory with sound banks, then listened to it, if there are any artefacts in sound... I also filed the memory with one long continuous steady sample (sine wave)...

The sound in both cases seems to be clean, without any artefacts...


Thank you for you help.

[dr.c]

I can't believe this...

Is you machine worling fine, or not ?
If YES, forget the test, its maybe the test prog wich has a bug !
If your machine is OK, why don't you just make music ?

[lubb]

I tried the test because I have done the upgrade of the memory from 8MB to 32MB... I fancied the test is capable to show if the procedure was done rightly (to show bad contact with a socket, for example) ...

That is all.

Of course, the main goal is to make music or noise, not to run tests, naturally, naturally.

Thanks for reminding it.

[dr.c]

I believe these idiots at E-Mu didn't upgrade the memory test wich gives back an error.

IF IT WORKS, DON'T FIX IT !

[lubb]

The interesting thing is that when the same memory test is run on my second EIIIX machine (both of them with the OS 2.10), it passes...

But, for sure, I am no going to fix it if it works...

Thank you !