MeNotHome said:
I have had my A7N8X-E Dlx board for about 1.5 yrs now and have changed
the power supply 4 times.
The last 3 times have changed the power supply the symptom was the
computer would just shut off on its' own and would not power back up
until I unplugged the unit from the UPS and waited about 30 seconds.
Plug it back in and it would start up.
The newest power supply issue is a startup problem. I boot the
computer and I could hear the hard drive starting and stoping and
starting again. I thought it was a bad hard drive so I replaced it.
Still does it.
So during bootup the other day I went into bios instead of letting it
come up and went to the hardware screen that shows the voltages.
The voltages seemed fine at first but after about 20-30 seconds I
noticed the +5 dropping.
As it warms up it starts at about 4.6-4.7v and then as the temp starts
rising the +5 starts falling down to about 4.35 and then I hear the
hard drive stop and start. After the temp reaches about 110deg F the
+5v starts climbing back to normal around 4.5-4.7v.
If I let the system boot after it gets back up to above 4.5v it works
perfectly. Otherwise I have to keep hitting the reset.
The current powersupply is a 430w antec. The system has two hard
drives, 1 cd burner, 1 floppy drive and that's about it.
Any recommendations on a better powersupply. I have built dozens and
dozens of PCs and haven't seen this problem before. I'm begining to
think there is something wrong with the motherboard to be causing the
powersupply numerous times.
Thanks for any advice
I would start by pulling the motherboard out of the case.
Assemble your basic system outside the computer case, using a
thick phone book as a support for the motherboard.
You could have something shorting to the bottom of the motherboard,
and assembling the system outside the case might show you some
different symptoms.
Having a separate voltmeter, like a cheap one from Radio Shack, would
allow you to monitor the voltage independently from the hardware
monitor chip. Like, when the processor is not there to read the
values. You can plug the meter into a disk drive connector, and
pick up +5V and GND from two of the pins on there. Be careful not
to short the leads on the meter (using two separate disk drive
connectors, for the two meter leads, might be a bit safer).
To start the motherboard, outside the case, you can touch a
screwdriver tip, to the two pins on the PANEL header where
normally the power switch is connected.
One test you could do, is pull the processor out of its socket, and
start the motherboard without a processor. Using your Radio Shack
voltmeter, check the voltage and see if it is wandering around at
its previous low level (4.5V) or not. If the voltage is staying up,
then chances are the processor is involved some how.
+5V is used to power the processor. On my A7N8X-E, I've tried various
test conditions on my board, and in the worst case (multiple copies
of Prime95, running in Linux), the board draws 16.6 amps from +5V.
The 3.3V consumption is pretty low, and varies from 5.2 to 6.4 amps.
There is only enough +12V being used to run the fans. If you have
a video card with a separate power connector (like my 9800pro), then
a card like that increase the +5V consumption by at most another
5.5 amps on +5V. A disk drive uses 1A of +5V, and allocate about
the same for a CD/DVD drive when it is playing a CD.
An Antec Truepower 430W has 36 amps on +5V to offer. Even under
gaming conditions, you are drawing maybe 23 amps or a bit more.
Actually, consumption on the processor drops a bit when gaming,
as the processor has to wait for the video card, so 22-23 is
a good enough estimate for a worst case.
The part of the current that flows through the ATX 20 pin connector
in this case, is the 16.6A number. There are four +5V pins on the
connector, and each one is rated for 6 amps. That means, if the
current goes over 24A, the pins on the connector will start to get
warm or hot. In extreme cases, the plastic will start to melt, and
the finish on the pins might show signs of heat damage.
If there is a short on the motherboard, the short could contribute
to the power supply loading. Similarly, if there is a fault inside
the processor, that causes more Vcore to be consumed, that will also
increase the current flowing through the Vcore converter, and
from the power supply. If the processor is doing this, you would
notice an abnormally high temperature for the processor.
The processor would likely start crashing and/or triggering Asus
COP to shutdown on thermal overload.
If the short is between conductors on the motherboard, there could
be a warm spot on the motherboard. When a short circuit is good
enough to squash the output of the power supply, the reduced output
voltage of the power supply, means the place that is shorting, can
no longer get hot enough to be noticed. So, whether a short is
observable or not, really depends on how good a short it is -
either a partial short (hot) or a complete short (cold).
With a complete short, the power supply should shut off. There are
a number of super-cheap power supplies, that have no protection at
all, and those ones would likely burn under these conditions.
Based on your observations on the voltage, I would expect the power
supply to be getting pretty warm. Perhaps the PSU fan is speeding up
a bit as well ?
To make my current measurements, I use a clamp-on DC ammeter. That
kind of meter is based on a Hall probe, a semiconductor that can
measure magnetic fields. To use it, all I have to do is place the
clamp around the four +5V wires coming from the power supply, and
the meter tells me how many amps are flowing. That kind of meter is
too expensive for ordinary debugging (maybe $400 CDN but price may
have dropped). If you could borrow a meter like that, it is
definitely a good tool for determining how much current is flowing.
In your case, the readings of the voltage will be your indirect
indication of how much current is flowing. Say, for example, you
borrow another processor from someone. It doesn't even have to be
a high performance processor, like a 3200+. Just as long as you
can get the BIOS to post, so you can read the voltage, and see if
the voltage does its usual 4.5V thing, or the voltage perks up to
5.0V. If the voltage perks up, then you would have a reasonable
idea that there is a fault inside your processor. If the voltage
stays low, the fault is inside the motherboard.
Similarly, if you operate the board, using the original processor,
sitting on the phone book, and the voltage perks up, you know a
standoff was shorting a +5v conductor.
If you wish to do none of the above, RMA the motherboard under
warranty, indicating that it is damaging power supplies. Asus
can then look for a short, when they rework the board.
HTH,
Paul
