ORIGINAL MESSAGE------------------
> I found that when I cool down the
> area around cpu (around socket 7 without cool the cpu), the problem
> appear. On other computers, I tested with same cooling and no problem at
> all.
>
> How could this happen? Can anyone give me an explanation of this? I've
> seen a lot of heat problems, but never seen this kind of problem before.

FIRST REPLY-----------------------
> Hi J Qian,
> I am confident that you have a cold solder joint around one or more of the
> 321pins
> that connect the CPU to the motherboard.....a careful inspection with a
> magnifying
> device may reveal the culprit....as the device heats up, the connection is
> made and it work ok.... when it cools down the metals contract and a very
> tiny crack opens up
> that prevents the system from booting.
> Hope this helps,,,and good luck in finding the one that is causing this to
> happen!!!
> Joseph H. Bishop
> PuterDoc
> www.puterdoc.com
> ICQ #270514

SECOND REPLY ----------------
> In my Limited experience with electricity it sounds like a broken trace on
> the MoBo. When the MoBo is cold the trace doesn't complete the circuit.
> When it warms up the MoBo the MoBo then reconnects the broken trace and
> WaLa a circuit is born.
>
> BCoe CM Productions
------------------------------
I saw this once, but I have seen many, many cracked, "cold" solder joints in
industrial electronics.  They were usually under massive components like
transformers, and appeared to have been caused by metal fatigue due to
vibration (although flexing a board repeatedly - as in pressing a large
chip into a socket several times) could also cause it.  Sometimes the
lead was corroded and the solder failed to wet it, but usually I found a
microscopic crack where the solder is thinner, in a full circle around the
lead, about 2/3 the diameter of the hole.

To see the crack, look on the solder side of the board, using 10x to 30x
magnification, with a strong light from the side, close to the board.

If you can, set up so that you can safely access both sides of the board
while it is operating - without it falling or causing a short.  Use the
cold spray to narrow down the area where the problem is.  Be patient,
you will have to let the board warm up between tries, but it beats
inspecting hundreds of pins.  Using a small wooden stick (such as the
handle of a long cotton swab), gently poke all the components in the area.
Gently try to push them in all directions, but not hard enough to
permanently bend the leads of taller components such as capacitors.

With luck, you should find the culprit.  Now, repeat the prodding from
the solder side, trying to wiggle each lead where it comes through the
solder.  Be prepared to widen the search - sometimes a slight flexing of
the board causes "false positives" when testing like this.  When you find
it, the gentlest of touches will usually cause the open circuit to occur,
and you will be able to see the crack under magnification.

To repair, resolder, making sure that it is well-fluxed, and melted all the
way through to the component side.  I assume you are familiar with static
and stray current precautions.

Hope this helps.

Boyd Ramsay

[log in to unmask]

            Do you want to signoff PCBUILD or just change to
                    Digest mode - visit our web site:
                    http://nospin.com/pc/pcbuild.html