View Single Post
  #10   Report Post  
Posted to sci.electronics.repair
D Yuniskis D Yuniskis is offline
external usenet poster
 
Posts: 319
Default HP Pavillion disk read failure

root wrote:
I picked up and HP Pavillion which is just over
one year old. The guy who sold it said it just
needed an new hard disk. Not true. I put a
new drive in it and installed my OS. After an
hour or so the system couldn't read the disk.
I shut it down and what I think is the Southbridge
chip was very hot. It had a small passive heat
sink which I removed, cleaned the surfaces, and
put on fresh conductive paste. I also affixed
a small fan to the heat sink. After this I was
able to run the machine for three or four hours
before the same disk read failures started.

I replaced the power supply with one known
to be good and the problems came back after
three hours or so of operation.

Just now I removed the CPU heat sink/fan and
renewed the heat sink compound. The CPU was
at room temperature, so I don't think this
is the problem.

None of the other components on the motherboard
is even warm to the touch. Even the disk drive
was at room temperature.

I can't think of anything other than heat that
would explain the problem, but then I don't
see how the machine could run for several hours
and without overheating.

Any ideas?


Any time you see wonky behavior from a PC, check
for bad caps. Around the processor is a prime
area of concern (ripple currents can be tens of
amps, easily, so ESR becomes a big issue).

You can also check *inside* the power supply for
similar problems.

I started a list of (cap and PC) manufacturers that
I have encountered this problem and gave up as it
became too long to track effectively (some I
just recognize instinctively, now).

Note what the machine is doing at the time (idle vs.
working hard, etc.).

Cold spray (whatever form of politically/legally correct
"freon" is en vogue) can help if you suspect a solid
state device.

The components with heat sinks are probably *expected*
to run warm -- don't be surprised if they *do*! :

Does a quick cycling of power (completely removed before
being restored) give you another "3 hours" of use? Or,
does the problem come back "quickly" (i.e., if it
is thermal, it should come back quickly; if it is a
software problem, it could take 3 more hours for a
memory leak, etc. to manifest)