Hardware lockup - need suggestions for diagnosing

[Ben Scott]

On 12/9/06, Larry Cook <lcook at sybase.com> wrote:
> I'm trying to diagnose my friends Dell Latitude C600 P3 @ 800MHz.

  Commentary: Computers (even laptops) have become so cheap (for those
of us lucky enough to live in the US) that any trouble that resists
diagnosis usually means the unit should be discarded.  But that's no
fun...

>  It just locks up w/o rhyme or reason.

  You've made a complete back-up of everything on the system, right?

  If not, the first thing you should do is remove the hard drive,
attach it to a working system as a secondary disk, and make a back-up.

> ... it happens with two version of Windows and four different Linux Live CDs
> makes me think it's a h/w problem ...

  Agreed.

> It also seems to happen more quickly when cold.  (Isn't it
> usually the other way around?)

  Indeed.

  What if you turn it off for a while, remove the battery, warm the
machine externally, and then try booting?  That might clue you in to
whether it really is thermal, or some other runtime-related thing (a
bad capacitor that does not accumulate energy at the rate it should,
maybe (I'm pulling that out of the air, but I've heard of weirder
things)).

> It hardly gets very far in the boot process when cold.

  The only things I can think of would be:

- Hard disk drive has a mechanical part that is sticking when cold,
but starts to move more freely when warmed.  Can you try a different
hard disk drive, just to see what the machine does?  One with Linux
installed would be ideal, since Linux doesn't freak if the host
hardware changes radically.

- Marginally bad solder joint that makes a better connection when heated.

> Any good diagnostic programs I can run?

  The Dell diagnostics are actually fairly comprehensive.  But it
rarely hurts to get a second opinion.

  Memtest86, Memtest86+, and "Windows Memory Diagnostic" are all free.

[1] http://www.memtest86.com/
[2] http://www.memtest.org/
[3] http://oca.microsoft.com/en/windiag.asp

  In the past, "badblocks -s -v -w" has smoked out defective hard
drives for me, but it's a destructive test (will DESTROY ALL DATA).
Also, more recently I've had drives pass that but still have
operational problems.  As drives get more intelligent, they get harder
to diagnose.  You can try determining the OEM and model of the drive,
and seeing if the OEM offers any manufacturer-specific diagnostics.

-- Ben