To Paul's reply:
Thank you for your feedback on my two issues.
The info on the Voice post is quite clear.But as I had suspected, I
have to plug a extra set of small speakers, just to hear the voice
post. Anyway, thank you for clearing this issue.
Regarding the P4T-E problem:
I don't think it's the PSU, as my new system with the P4C800E DLX is
running on the same PSU/case. With my new upgrade, I even added more
power requiring components - ie. Two SATA drives (besides the original
IDE drive), Asus nVidia FX5900 Ultra with it's extra power connector
(GeForce 2 Ultra previuosly), Prescott CPU.I have a Enermax Whisper
450W.
I took the Mobo to a friend's PC shop, and on a diferent PSU there,
the board was declared "DEAD" by them, as it didn't seem to operate
either.
When I mentioned, that I tinkered with the board, mainly I was
re-seating the CPU in the MB socket, start the system, re-seat then
start again until the system eventualy booted.In my opinion, it has to
do something with the CPU socket on the MB.CPU is fine and working, as
I checked it on diferent MB.Is there a way to further check, where the
broken link is on the MB?
Thank you for any further feedback.
Well, to be honest with you, it is really miraculous that so many
boards work so well. The big chips on a motherboard are surface mount
devices, and the chip type is called BGA (ball grid array). BGA
is a nice technology, in that, when they make them in the factory,
a BGA will "snap" into alignment, as the balls of solder melt to
the board. This "auto-alignment" of the chip to the board, is what
makes the yield in the factory so high.
Now, when the board is installed in a product, there is a second
issue. That is thermal stresses and mechanical stresses. Thermal
stress is where the CPU heats up and cools down, as you vary the
loading on the CPU. The constant change in temperature of the
material around the balls of solder, causes the solder to fatigue
and crack. A cracked ball can give you an "intermittent" connection,
where a slight flexing of the board may cause the connection to be
made again. But, from a mechanical stress perspective, flexing
the board is also bad for the solder joints, if a lot of stress
is applied. Inserting DIMMs, for example, is a source of stress
to the board, and a good reason to keep all the screws on the
motherboard tight (without overdoing it of course).
At the factory, one method used to detect bad soldering, is an
X-ray machine. After the board is passed through the wave solder
machine, an X-ray machine with a camera connected to it, takes
pictures of each BGA chip (like the Northbridge and the Southbridge).
The angle the camera makes to the board can be varied, to look on
either side of the tin/lead ball of solder. By using optical
comparison, this machine can even be automated, so the machine
can identify and reject badly soldered boards. Once the soldering
process is refined, the need to X-ray every motherboard can be
reduced to statistical sampling, where every one of N boards is
checked. Another test, using a continuity tester, is an additional
test, that is cheaper than X-raying, and is likely to be used on
every motherboard. This is done with a "bed of nails" tester,
equipped with a vacuum head, to hold the motherboard against the
spring loaded nails.
The larger the number of pins on a BGA, the higher the probability
that one of the balls of solder will make a bad connection. In a
report I read some time ago, a 750 ball BGA is considered to have
a lifetime of around 10 years or so. It is still possible for
an earlier failure to occur, but based on the small number of
failures you hear about in this newsgroup, the problem isn't
widespread.
Another place for problems, can be the ZIF socket for the processor.
If the contacts are bent or distorted, either due to a bad ZIF
socket from the ZIF manufacturer, or due to handling of the ZIF
by the end user, then that can be a source of intermittent problems
as well. As your board appears to be dead, if you don't think it
will ever be alive again, you could look at the plastic "lid" that
fits over the base of the socket. It may be possible to pry the
lid off, if you are careful. You could then, at least, look to see
if any of the socket pins are disturbed. I don't think it is
really practical to fix a distorted pin, because the act of bending
the metal would ruin the shape of the contact surface, and only
make matters worse.
You might also examine the pins on the CPU, to see if any are bent.
Maybe that is the problem ?
HTH,
Paul