Ken said:
Maybe by your CPU are to hot? ("Akasa cheapo HS fan")
Are the cooling device properly mounted?
Maybe your PSU need to have at least 20-25A output
capacity on the +12V output.
A high +12V output won't help, because without the ATX 12V
2x2 plug, there is only one wire carrying current from the
PSU to the motherboard. A pin on the ATX 20pin connector is
rated for 6 amps, so if you draw more current than that,
the connector pin will burn. The Vcore is either powered by
that one wire, or else +5V is used for the Vcore. I have no
way to be certain which is the case. In any case, having
20 or 25 amps of output on +12V doesn't help, unless you can
use the 2x2 ATX12V connector.
As to figuring out what is happening, it will be difficult
without instruments. I have a clamp-on DC ammeter (could
be a Hall probe inside) and I can measure the current in
any wire, without making contact with the wire. With such
an instrument, in 2 minutes you can check all the wires and
see which one carries too much current. While it isn't likely,
contact a local computer repair shop and see if they are
equipped with a clamp-on DC ammeter, or they have a power supply
that has meters on all the outputs.
Without instruments, the process of finding the problem will
be slow, and potentially will endanger another PSU.
First, remove the motherboard from the case, and assemble the
system outside the case. (Before adding or removing components,
always unplug the line cord, to guarantee there is no power
on the motherboard.) Your problem could actually be a partial
short from a motherboard standoff to a power supply on the
bottom of the board, and that is the main reason for removing
the board.
With the motherboard sitting on a piece of cardboard or other
insulator, power just the motherboard from the PSU. Switch it
on and wait to see if the air coming from the PSU gets real hot.
This might take five or ten minutes to find out. (To start the
motherboard, you don't need any wires on the PANEL header.
Simply touch an (ESD drained) screwdriver tip to the two pins
on the PANEL header that you normally connect the case power
switch to.)
Add another component, like the processor plus heatsink/fan.
When you power it up this time, it should beep, indicating
it cannot find any memory or video card, etc. Run it for a
while and see if the air coming from the PSU is hot. If the
PSU has a variable fan, you may notice the fan speed up in
a matter of minutes, again indicating the PSU is heating up.
Continue adding components one at a time, until the PSU shows
it is being stressed. If the symptoms are not reproducable
(i.e. PSU won't blow up or air won't get hot or PSU fan
never speeds up), then it could be a standoff was touching
something it wasn't supposed to.
If at some point, adding a component causes an observable
change in the PSU, then that component could be bad.
There are a couple of ways a PSU can be tortured. Both the PSU
and the Vcore circuit are capable of supply lots of current,
so that a partial short can be drawing current, and yet the
motherboard continues to run. So, a partial short can overheat
the PSU, and the computer could still be working, right up
to the last minute the PSU is alive. This would be a rail to
ground partial short.
A second failure mode, is when one supply touches another
supply. This is a rail to rail short. This can screw up the
monitored voltage values, for example, so examining the BIOS
hardware monitoring page might show an anomaly. In fact,
before you take anything apart, perhaps you could connect a
third (good) PSU to the board, long enough to enter the BIOS,
and observe measured values for the PSU voltages. You might be
able to see one of the rails being lower or higher than it
should be, and that might give you an idea as to what has
failed.
HTH,
Paul
