Mysterious RAM problem.

[sandman.150284]

I know there are a zillion other posts all around the web concerning
this, but I still felt that I had to post one myself. Apologies in
advance. =)

I have a Compaq Evo N160 with 256Mbs on RAM. I bought a 128Mb stick
from an internet auction. Though the stick I bought was from a HP
Omnibook XE3, the specs seemed to match mine.

The very first time I installed the new stick, the system wouldn't
boot, not even BIOS would load up and the screen stayed dead. I
shutdown the system "hard", then tried again. This time it started, but
then I got a couple of these weird system lockups (OS Win2000). Out of
the blue and with no apparent reason, the system would just freeze, so
that the display would stay on but the keyboard and the mouse would
become dead and I would have to boot the
computer.

At that point I thought that the new memory stick was simply broken. To
confirm that, I ran five passes of Memtest86+ on both sticks alone,
with no errors. Then I installed the new stick alone, and the system
worked fine, although a little sluggishly as one might expect. =)

So, it seemed that the new stick wasn't broken after all. Again I tried
to install both of the sticks. And again, I couldn't get the system to
boot. Like the first time, not even BIOS would load up and the screen
stayed dead. I tried several times to boot it, but nothing happened.

Every spec I can think of is the same on both sticks. Manufacturer,
memory type, speed, the width of the data path, voltage, even the damn
latency.

- Does anyone recognize the symptoms above? What's going on?
- Is this simply a compatibility problem, or is some setting in my BIOS
wrong?
- Can anyone think of a reason why the two memory sticks don't seem to
like each other, anymore?
- Can anything be done, or should I just throw the new stick at a
waterfowl? (a Finnish
proverb)

 

[kony]

I know there are a zillion other posts all around the web concerning
this, but I still felt that I had to post one myself. Apologies in
advance. =)

I have a Compaq Evo N160 with 256Mbs on RAM. I bought a 128Mb stick
from an internet auction.

Beware of auctions, this is a very significant additional
variable, that we can't even be sure the module works.
Since there are few if any "tricks" to get a module working,
the best course if possible is to return the module to the
seller.

Though the stick I bought was from a HP
Omnibook XE3, the specs seemed to match mine.

The very first time I installed the new stick, the system wouldn't
boot, not even BIOS would load up and the screen stayed dead. I
shutdown the system "hard", then tried again. This time it started, but
then I got a couple of these weird system lockups (OS Win2000). Out of
the blue and with no apparent reason, the system would just freeze, so
that the display would stay on but the keyboard and the mouse would
become dead and I would have to boot the
computer.

Don't boot a system to windows (or any other OS for that
matter) after changing memory aspects without first testing
the memory subsystem... else you may have data corruption,
including that written to the drive, that persists after the
memory instability is corrected. This can also cause
continual OS instability.

Run Memtest86+ for several hours before anything else.
Google will find it.

At that point I thought that the new memory stick was simply broken. To
confirm that, I ran five passes of Memtest86+ on both sticks alone,
with no errors. Then I installed the new stick alone, and the system
worked fine, although a little sluggishly as one might expect. =)

While running it on one (either) stick alone is a reasonable
first step, it doesn't tell you if BOTH of them running
together are stable. When both were installed did the bios
report the full capacity as expected?

So, it seemed that the new stick wasn't broken after all.

It is possible a memory module has very little stability
margin, that it could be stable alone in a system but if you
added two of the same module (or two different of course),
it wouldn't then be stable. Sometimes it is the mainboard's
fault, designed poorly or the bios is buggy, but other times
the modules weren't really stable enough that they should
have been spec'd as they were. On many desktop systems the
resolution is to manually set slower (higher numbered)
memory timings but apparently most notebook manufacturers
feel a notebook shouldn't be allowed this level of
adjustment. To be fair, many desktop OEM systems dont'
allow it either but to resolve issues you see, it is a good
feature to have and yet anothe reason why desktop OEM
systems are avoided by (some) experienced users.

Again I tried
to install both of the sticks. And again, I couldn't get the system to
boot. Like the first time, not even BIOS would load up and the screen
stayed dead. I tried several times to boot it, but nothing happened.

Every spec I can think of is the same on both sticks. Manufacturer,
memory type, speed, the width of the data path, voltage, even the damn
latency.

Now the bad news- it's possible the new memory module is
better than the old one, that only replacing the new module
won't solve the problem but replacing the old one and
keeping the new one might. However, i would suspect it is
still more likely the old module is ok, as OEMs will often
test two of the same exact module together because they
often offer memory upgrade configurations when you buy one
new, that simply add another of same module. In past years
this was to be avoided since it meant all memory slots are
filled, and abandonment of current memory for a further
upgrade, but today the situation has changed since the
typical new laptop uses dual channel memory but has only two
slots, so will ideally have two modules installed.

- Does anyone recognize the symptoms above? What's going on?
- Is this simply a compatibility problem, or is some setting in my BIOS
wrong?
- Can anyone think of a reason why the two memory sticks don't seem to
like each other, anymore?
- Can anything be done, or should I just throw the new stick at a
waterfowl? (a Finnish
proverb)

Well since it tests stable by itself, if you can't return it
to seller you can just re-auction it, but the most honest
way of doing so would be to mention that it's specified for
an XE3 but does not work properly on an N160 with original
memory still installed.

 

[paulmd]

I know there are a zillion other posts all around the web concerning
this, but I still felt that I had to post one myself. Apologies in
advance. =)

I have a Compaq Evo N160 with 256Mbs on RAM. I bought a 128Mb stick
from an internet auction. Though the stick I bought was from a HP
Omnibook XE3, the specs seemed to match mine.

I have seen one or two cases where both sticks of RAM are fine singly,
but do NOT like working together.

 

[Steven Spits]

Hi,

I have seen one or two cases where both sticks of RAM are fine singly,
but do NOT like working together.

Same here.

2 modules of 256 MB each. Testing them seperatly always worked like a charm
for days and even weeks but when both modules where used, the system would
freeze about once every few hours.

Steven

- - -

 

[sandman.150284]

The plot thickens, as they say...

I'm typing this with both sticks of memory in the system. Just for the
hell of it I plugged the new stick in today and presto, it works. I
haven't done anything to the system.

I ran Memtest86+ on the system for four and a half hours, it completed
eight passes and found no errors.

With the Memtest running time, the system has been up for some eight
hours now, without any lockups. The memory is shown on Windows.

As far as I can tell, both stick are CL3. So memory timing shouldn't be
a problem, am I right?

If it's of any help, here's a pic of the new stick:
http://img172.imageshack.us/img172/3163/muistikampa9sk.jpg

 

[paulmd]

The plot thickens, as they say...

I'm typing this with both sticks of memory in the system. Just for the
hell of it I plugged the new stick in today and presto, it works. I
haven't done anything to the system.

I ran Memtest86+ on the system for four and a half hours, it completed
eight passes and found no errors.

With the Memtest running time, the system has been up for some eight
hours now, without any lockups. The memory is shown on Windows.

As far as I can tell, both stick are CL3. So memory timing shouldn't be
a problem, am I right?

If it's of any help, here's a pic of the new stick:
http://img172.imageshack.us/img172/3163/muistikampa9sk.jpg

Dirty contacts on the 2nd RAM slot perhaps? Plopping it and out a few
times may have provided just enough cleaning. But since you got it
working, I'd hesitate to ask you to pull it out just to check....

 

[Steven Spits]

Hi,

I ran Memtest86+ on the system for four and a half hours, it completed
eight passes and found no errors.

In my case, Memtest86 could not find any errors but still Windows would
freeze up.

Steven

- - -

 

[larry moe 'n curly]

In my case, Memtest86 could not find any errors but still Windows would
freeze up.

Did you try www.goldmemory.cz ? It's failed some of my memory that
MemTest86 passed, and vice-versa.

Somebody recently said that HCImemory detected errors missed by
MemTest86, and last year a person reported that Microsoft's memory test
(http://oca.microsoft.com/en/windiag.asp) found errors missed by both
Gold and MemTest86. OTOH there seems to be no reason to bother with
DocMem (www.simmtester.com).

 

[kony]

Hi,


In my case, Memtest86 could not find any errors but still Windows would
freeze up.

If the board allows raising the memory bus speed some, you
might try doing that. I insist on a margin above the
regular running speed, that this higher (temporarily)
setting also test stable with memtest86+. You might need to
manually lock the memory timings though, in some cases
raising the memory bus speed will cause the bios to raise
(slow down a little) the memory timings automatically which
would foul up the test, make the memory seem more stable due
to this happening.

On a related note, manually setting those timings higher
yourself is a way to improve/regain stability in
almost-stable situations.

 

[sandman.150284]

Kony, no such luck. The BIOS won't let me tamper with the memory
timings and I am very uneasy about doing it manually with any sort of
jumper settings, even if that could be done. With my expertise in
hardware, that would be a surefire way to kiss goodbye to the whole
laptop. =)

I'll have to give the Windows Memory Diagnostics a run, once I get a
chance to burn it on a disk somewhere. Also, I haven't booted the
system since I got it up, it's only been on standby. So I will have to
see what happens when I do shut it down properly.

I'll keep you posted.

Sandman

 

[sandman.150284]

Well, the system has been running fine since last Thursday, so I think
the problem actually did solve itself. I guess dirty connectors were
the reason, but there's no way to know. I sure as hell ain't taking
anything out anymore, unless the system locks up again. =) For once I'm
going to follow my Dad's advice: "If it isn't broken, don't fix it."


Lesson to be learned? When installing used memory sticks on an old
system, clean the connectors.

Thank you, everyone, for your help.

 

[kony]

Well, the system has been running fine since last Thursday, so I think
the problem actually did solve itself. I guess dirty connectors were
the reason, but there's no way to know. I sure as hell ain't taking
anything out anymore, unless the system locks up again. =) For once I'm
going to follow my Dad's advice: "If it isn't broken, don't fix it."


Lesson to be learned? When installing used memory sticks on an old
system, clean the connectors.

Thank you, everyone, for your help.

One trick that can help is to take a non-lint type of paper
(Or at least low lint, not a paper towel), cut it to the
width of the memory module, spray contact cleaner on it
(enough but not dripping off of it), and quickly insert the
module before the cleaner evaporates off. If the paper
looks dirty afterwards, repeat this with a fresh piece until
it doesn't look dirty. Before doing this it might help to
gently dust out the system... particularly if the dust had
been accumulating for years it was due to have that done
anyway.

 

[Lee]

One trick that can help is to take a non-lint type of paper
(Or at least low lint, not a paper towel), cut it to the
width of the memory module, spray contact cleaner on it
(enough but not dripping off of it), and quickly insert the
module before the cleaner evaporates off. If the paper
looks dirty afterwards, repeat this with a fresh piece until
it doesn't look dirty. Before doing this it might help to
gently dust out the system... particularly if the dust had
been accumulating for years it was due to have that done
anyway.

The simple way of cleaning the contacts on ram, or the fingers of any
insertable card is to rub lightly with a PINK PEARL eraser and wipe off
with a lint free cloth.

Pink Pearl because it doesn't have any grit in it.
Regards
Lee