Dell laptop memory issue

Hi folks,

A friend of mine has an old Dell Inspiron 4000 laptop, which has been
running fine for several years since he bought it new. It came with only
128MB RAM, and he asked me about upgrading it. I found him two 256MB
Kingston modules on eBay, ordered them and fitted them for him. Ran
memtest86+ for about sixteen hours (something like 7 or 8 full passes) - no
errors whatsoever.

He took the machine away happy. Everything he used it for worked fine, and
much faster with 4x the RAM ... except one thing. He runs some DJing
software (I don't know the exact name), and uses a Roland PCMCIA sound card
to provide better quality sound than the on-board hardware.

This DJing software was shutting down the laptop after an indeterminate
period (never less than two hours, sometimes up to seven or eight). I don't
mean BSOD, nor a normal shutdown, but a complete and instant power-off. Same
symptom every time, the only variable was the lead time.

We took the RAM back out and reverted to his previously 128MB, and the DJing
software works fine again, indefinitely. Yet I've tested that RAM to pieces
with memtest86+, and there is nothing wrong with it, so I can't send it
back.

Does anyone with more knowledge than me have any idea what's causing the
trouble with this one piece of software? My only theory is that it's somehow
power- or heat- related. Maybe the PCMCIA sound card is taking the machine's
power consumption to its limit, and the extra stick of RAM was just pushing
it over the edge, causing it to shut down due to ... overheating? It's not
much of a theory, but I can't explain why the machine does an instant
power-off if the RAM is bad, instead of a normal BSOD-type crash.

Grateful for any suggestions,

CC

Magnate <> wrote:
> Hi folks,

> A friend of mine has an old Dell Inspiron 4000 laptop, which has been
> running fine for several years since he bought it new. It came with only
> 128MB RAM, and he asked me about upgrading it. I found him two 256MB
> Kingston modules on eBay, ordered them and fitted them for him. Ran
> memtest86+ for about sixteen hours (something like 7 or 8 full passes) - no
> errors whatsoever.

> He took the machine away happy. Everything he used it for worked fine, and
> much faster with 4x the RAM ... except one thing. He runs some DJing
> software (I don't know the exact name), and uses a Roland PCMCIA sound card
> to provide better quality sound than the on-board hardware.

> This DJing software was shutting down the laptop after an indeterminate
> period (never less than two hours, sometimes up to seven or eight). I don't
> mean BSOD, nor a normal shutdown, but a complete and instant power-off. Same
> symptom every time, the only variable was the lead time.

> We took the RAM back out and reverted to his previously 128MB, and the DJing
> software works fine again, indefinitely. Yet I've tested that RAM to pieces
> with memtest86+, and there is nothing wrong with it, so I can't send it
> back.

I once had a module with a weak bit that took about 3 days to be detected
by memtest86+. Of course, this is a worst-case scenario.

> Does anyone with more knowledge than me have any idea what's causing the
> trouble with this one piece of software? My only theory is that it's somehow
> power- or heat- related. Maybe the PCMCIA sound card is taking the machine's
> power consumption to its limit, and the extra stick of RAM was just pushing
> it over the edge, causing it to shut down due to ... overheating? It's not
> much of a theory, but I can't explain why the machine does an instant
> power-off if the RAM is bad, instead of a normal BSOD-type crash.

> Grateful for any suggestions,

A complete, instant poweroff does not sound like a memory issue to
me at all. However the heat-theory does have merit. I had this on
a sony Vaio sr11k (defective by design, had 2 self-destruct
devices in there), that had inadequate chipset cooling. It eventually
died after 2.5 years of light usage (1 year warranty).

The north-bridge has to drive twice as many lines to the RAM now.
Maybe the RAM is also faster? That would add even more thermal load.
And it is possible this will kill the northbridge relatively fast.

As to diagnosis, try with one of the new modules. If it takes noticeably
longer to crash or does not crash at all, that would be a strong
indicator. You could also reduce RAM timing as much as possible.

Also, it would be a good idea to have a look into the laptop. Often
these things get so clogged up by dust when they get older, that they
overheat pretty fast. A good cleaning will sometimes do wonders.
You can use a vaccuum cleaner, but make sure it is anti-static.
AFAIK most modern ones are.

Arno

"Arno" <> wrote
> A complete, instant poweroff does not sound like a memory issue to
> me at all. However the heat-theory does have merit. I had this on
> a sony Vaio sr11k (defective by design, had 2 self-destruct
> devices in there), that had inadequate chipset cooling. It eventually
> died after 2.5 years of light usage (1 year warranty).

I caught up with my friend yesterday, and he confirmed that it has now
happened once with only the 128MB RAM, so the overheating theory is growing
on me.

> The north-bridge has to drive twice as many lines to the RAM now.
> Maybe the RAM is also faster? That would add even more thermal load.
> And it is possible this will kill the northbridge relatively fast.
>
> As to diagnosis, try with one of the new modules. If it takes noticeably
> longer to crash or does not crash at all, that would be a strong
> indicator.

Genius! Why didn't I think of that? Many thanks for the suggestion - will
try one module.

> Also, it would be a good idea to have a look into the laptop. Often
> these things get so clogged up by dust when they get older, that they
> overheat pretty fast. A good cleaning will sometimes do wonders.
> You can use a vaccuum cleaner, but make sure it is anti-static.
> AFAIK most modern ones are.

Yes, I suspect that the innards are gunked up with
dust-that-is-now-compacted.

Thanks again Arno,

CC