voltage question on the A7V

  • Thread starter Thread starter ^JR^
  • Start date Start date
J

^JR^

Howdy,

The hardware monitor was halting the POST because the +3.3v has moved up to
+3.9v. I switched that to ignore in the power setup and it POSTs fine. Now
I've been running the Asus probe software constantly and I'm noticing spikes
in other voltages. Most noteable the -12v will spike to -8.xx for a second.
with no regualr interval. other voltages spike while running applications.
I'm sure this is fairly normal. I'm just concerned with the +3.3 holding so
far out of the threshold. Now, the system operation doesn't seem affected.
Should I be concerned?

I have the Allied 400w PS (dont have exact model number handy)
Asus A7V
Duron 1.2 Ghz

Thanks in advance!!!!!
^JR^
 
Howdy,

The hardware monitor was halting the POST because the +3.3v has moved up to
+3.9v. I switched that to ignore in the power setup and it POSTs fine. Now
I've been running the Asus probe software constantly and I'm noticing spikes
in other voltages. Most noteable the -12v will spike to -8.xx for a second.
with no regualr interval. other voltages spike while running applications.
I'm sure this is fairly normal. I'm just concerned with the +3.3 holding so
far out of the threshold. Now, the system operation doesn't seem affected.
Should I be concerned?

I have the Allied 400w PS (dont have exact model number handy)
Asus A7V
Duron 1.2 Ghz

Thanks in advance!!!!!
^JR^
Click to expand...

Maybe your ps dont have enough power on the 12v and the 3.3v (AMP)
the way psu are made, there are compose of little other psu. As i'm
reading stuff over the internet i learn that FXcard and cpu(xp) are
running on the 12 volts.

I know that Video card are basicly running on 3.3volt at 1x and .5volt
on 8x) concidering that lowering the voltage increase the (amp curent)
this means that the output amp on the 12volt (should be higher than
the current amp that you and i have.

For your problem i notice that i'm getting fluctuation in Different
voltage to, But it happend only when i change my FSB (cpu)
i'm actualy running my fsb at 140/140memory
i know that my voltage is still going down and returning to normale
but hey, it's stable...
 
"^JR^" said:
Howdy,

The hardware monitor was halting the POST because the +3.3v has moved up to
+3.9v. I switched that to ignore in the power setup and it POSTs fine. Now
I've been running the Asus probe software constantly and I'm noticing spikes
in other voltages. Most noteable the -12v will spike to -8.xx for a second.
with no regualr interval. other voltages spike while running applications.
I'm sure this is fairly normal. I'm just concerned with the +3.3 holding so
far out of the threshold. Now, the system operation doesn't seem affected.
Should I be concerned?

I have the Allied 400w PS (dont have exact model number handy)
Asus A7V
Duron 1.2 Ghz

Thanks in advance!!!!!
^JR^
Click to expand...

You should only use one monitoring program at a time. If you are running
MBM and Asus Probe at the same time, they can conflict with each other's
use of the SMBUS. The monitor chip sits on the SMBUS, which is a low speed
serial bus. If one program does an SMBUS read, and then a few hundred
microseconds later the other program attempts one, the readings can get
corrupted. The programs should all be using a hardware mutex, to avoid
this, but Asus Probe is known not to do this properly.

As for the performance of the power supply itself:

1) ATX power supplies have a minimum current draw requirement. Each
product is slightly different, so check the specs. For the least
used rails, like -5V or -12V, it is easy to find them out of
regulation if they are unloaded.

2) All ATX power supplies have a transient recovery time. This time
is significant, like 1 millisecond or so. I haven't read anything
definitive about this, like how much the voltages can vary when
there is a large change in current. ATX power supplies are not
tightly regulated like they would be on a linear regulated supply,
so if the monitor program catches the supply, just after a big
surge (like you just started Prime95 or CPUburn in another
window), you might see the voltages change.

3) The 3.3V supply is the most tightly regulated output. Examine the
20 pin ATX power connector going into the motherboard. One of the
pins has two wires going into the same pin. The thin brown one
is a feedback lead, and it is a mechanism for the power supply to
"observe" what is happening, as close to the motherboard as possible.
So, of all the supplies, the +3.3V should be the best. I find it
disconcerting that yours reads +3.9V.

You should find someone with a decent multimeter, to verify the real
voltages. The +3.3V shouldn't go more than about 5% higher than normal,
and that is +3.47. So, if it really is +3.9, that is out of spec.
While destruction might not be instantaneous, it is something you
should fix, especially if the power supply is telling you it is about
to fail. (If you cannot figure out where to measure the voltage,
strap a straight pin to one of the multimeter leads and push the pin
through the insulation of one of the +3.3V wires from the PS. That is
what mechanics do to car wiring harnesses when they are debugging
electrical faults. The pin should be small enough not to leave a
visible hole after it is removed. A ground connection can be
picked up from one of the motherboard connector shrouds. Be careful
not to short the +3.3V supply, because it could melt the needle on
you!)

Chances are that this is just a problem with the monitor chip or
the software used to read it. The multimeter can be used to
independently verify that the power supply is really OK.

I like to verify the power supply before I use it in a new
computer, by connecting some dummy loads made from some power
resistors I bought. I select the resistors to draw at least
the minimum specified load for each power rail. So, if the supply
has a minimum of 2A on the +12V, I connect a 6 ohm resistor with
at least a 24W power rating. Yes, they do get hot and I run
a fan next to them while testing. You need to connect PS_ON# to
an adjacent COM pin, to get an ATX supply to turn on.

HTH,
Paul
 
one more quick question... You say I can pick up a ground from one of the
motherboard connector shrouds. I'm not familiar with the term. Direct me
to a shroud, if you would please, sir :)

Thanks again
^JR^
 
"^JR^" said:
one more quick question... You say I can pick up a ground from one of the
motherboard connector shrouds. I'm not familiar with the term. Direct me
to a shroud, if you would please, sir :)

Thanks again
^JR^
Click to expand...

The shroud is the metal shell that surrounds the pins. You can either use
that, or the two nuts used to secure a cable to the back of the computer
would also be useful. The reason I like something like this, is because
then the meter lead cannot fall into other circuitry while you are
working - you can concentrate on probing with the other meter lead.

Paul
 
Hello, me again.

Well, I tried to get a reading on that +3.3v. I say tried, because all I
have is an analog meter and I'm not sure what range to set for a DC voltage
reading. On top of that I'm not even sure I'm reading the correct line of
numbers. Guess, I'm gonna have to hold out for a digital meter.

^JR^
 
"^JR^" said:
Hello, me again.

Well, I tried to get a reading on that +3.3v. I say tried, because all I
have is an analog meter and I'm not sure what range to set for a DC voltage
reading. On top of that I'm not even sure I'm reading the correct line of
numbers. Guess, I'm gonna have to hold out for a digital meter.

^JR^
Click to expand...

Use the "DC 10V" scale. The needle should come about a third of the
way up the scale. Don't forget to zero the meter - with no voltage
input, turn the screw that forms the "hub" for the meter needle
until the meter reads zero. You should do the zero with the meter
laying flat on a table and make your measurements the same way
as well. My old analog meter is only good to +/- 3%, so don't
expect spectacular results. The cheapest digital I've got ($25 on
sale) would do the same measurement +/- 1%, but is much worse on
some of the other scales.

HTH,
Paul
 
Ok then, my reading was correct. It is coming out to be +3.9v. Is this far
enough out of spec to warrant replacing the PS? I've only been running my
machine sparingly since this problem started to limit damage.

You've been a great help, Thanks again,
^JR^
 
Ok then, my reading was correct. It is coming out to be +3.9v. Is this far
enough out of spec to warrant replacing the PS? I've only been running my
machine sparingly since this problem started to limit damage.
Click to expand...
......
Well, did you measure the PSU voltages with a load, the MB connected?
If not, the reading could be wrong. Almost every PSU for PCs I have
seen, needs at least a minimum load to work correctly.
Armin
 
Yup, motherboard connected. I did not remove any cards. But I did not run
any applications during the test.

^JR^
 
Hello, again. I have something interesting to report. To be safe, I
swapped out power supplys with a skyhawk 300w I had lying around. The
hardware monitor reads the +3.3v out of spec to the same degree, +3.9x.
This strikes me as odd that 2 PS's would read the same like that. Guess
I'll be REchecking with my meter. By the way, I did eliminate some of the
other possibilities for the cause by trying a new power cord and plugging in
directly to the wall outlet rather than my surge protector. same results.
Stay tuned.....

^JR^
 
"^JR^" said:
Yup, motherboard connected. I did not remove any cards. But I did not run
any applications during the test.

^JR^
Click to expand...

If it really is 3.9 volts, it should be replaced. The reasoning being,
if it is 3.9 today, how high will it be tomorrow ? If it is out of spec,
it is a danger to the rest of your machine. Replacing the power supply is
cheaper than replacing all the hardware. (And, Armin is right, the
voltages should only be measured under load. I should have made that
more clear. But my recipe, to use a needle to pierce the wire, was
intended for you to make the measurement with the motherboard
connected.)

In terms of limits, I just checked a couple of chips to see what kinds
of limits they have. One 3.3V chip had an absolute max of 3.6V, while
another could handle 4.6V. So, the ability to withstand excess voltage
varies from chip design to chip design. The chip with the lowest value
will be the first one to be damaged.

Paul
 
"^JR^" said:
Hello, again. I have something interesting to report. To be safe, I
swapped out power supplys with a skyhawk 300w I had lying around. The
hardware monitor reads the +3.3v out of spec to the same degree, +3.9x.
This strikes me as odd that 2 PS's would read the same like that. Guess
I'll be REchecking with my meter. By the way, I did eliminate some of the
other possibilities for the cause by trying a new power cord and plugging in
directly to the wall outlet rather than my surge protector. same results.
Stay tuned.....

^JR^
Click to expand...

Interesting. Since both power supplies do the same thing, maybe you have
an inter-rail short ? That is a connection between say the +5V signal
and the +3.3V signal. The +5V signal tries to lift the +3.3V, and that
is how it is rising to +3.9 on both power supplies. This will make the
+5 supply lead much more heavily loaded, while the +3.3V supply gives
practically no current at all.

To debug this further, remove all the AGP and PCI cards. Handle them
by the edges and store them in the antistatic bag they came in.
Power up the system (yes, it will beep madly because the video card
is missing) and make your +3.3 measurement again. If the voltage has
returned to normal, one of the PCI/AGP cards has a rail-to-rail short.
Replace the cards one at a time, until you find that the PS is
overvolting again. The last card you put in is the faulty one.

If it still reads 3.9V with the motherboard alone, the rail to
rail short could be on or inside the motherboard. The motherboard
consists of four layers of copper sandwiched with non-conducting
fiberglass. You cannot examine the inner layers for faults. At
this point, I would pull the motherboard and do a thorough visual
inspection, such as looking for a brass standoff underneath the
motherboard that shouldn't be there etc. The standoffs are only
supposed to go, where they line up with holes in the motherboard.
Also, look for a conductor of some sort that has fallen across
the copper tracks on the motherboard.

An example of a manufacturing fault is pictured here. This is a
solder blob on a P4C800, and it is shorting two tracks together.
Look carefully at the lower right clump of solder and how it is
shorting to an adjacent track. That is the kind of thing you would
be looking for.

http://koti.mbnet.fi/~nightops/eki/DSC00249.JPG

If, after all of this work, you still cannot see the problem,
then RMA the motherboard. When you ship the motherboard, place
a note in the box explaining that you think there is a
rail-to-rail power short. In many cases, the board will be
discarded, as these faults cannot be fixed if they are inside
the PCB.

How we handle this in a lab situation, is we use an ohmmeter
to establish "norms" for a circuit board design. We measure the
resistance between the rails and ground, or from rail to rail.
It is amazing how closely a batch of boards agree on their
resistance between rails. After establishing what a "good" value
is for the design, when a freshly manufactured board comes in, a
quick check with the ohmmeter will tell whether there is something
grossly wrong with the board. In the case of a rail-to-rail
short, it is then detected before it can do any harm. Since you
only have one board to work with at home, you don't have that
luxury (of establishing a "norm").

HTH,
Paul
 
Well, I took my PS and put it in my daughters computer and it's hardware
monitor shows voltage within spec. I put her PS (which also showed within
spec on her monitor) and put it in my case and it reads out of spec as the
others have. I also removed all of my cards, as you suggested with the same
result using the MB alone.

While I haven't inspected the board itself yet, I'm now concluding my PS to
be good more than likely. My present generic case has a few years on it, so
I'm just gonna replace it. This case didn't even use brass stand offs
except at the top corner near the peripheral connections. All other
connection points are screwed into these little aluminum risers that clip on
to the MB plate and I never was happy with that so a new case for me, it is.
:)

By the way, I put my machine aside and am using my daughter's until I get a
new case.

And that about wraps it up. Thanks for you help and patience.

^JR^
 
"^JR^" said:
Well, I took my PS and put it in my daughters computer and it's hardware
monitor shows voltage within spec. I put her PS (which also showed within
spec on her monitor) and put it in my case and it reads out of spec as the
others have. I also removed all of my cards, as you suggested with the same
result using the MB alone.

While I haven't inspected the board itself yet, I'm now concluding my PS to
be good more than likely. My present generic case has a few years on it, so
I'm just gonna replace it. This case didn't even use brass stand offs
except at the top corner near the peripheral connections. All other
connection points are screwed into these little aluminum risers that clip on
to the MB plate and I never was happy with that so a new case for me, it is.
:)

By the way, I put my machine aside and am using my daughter's until I get a
new case.

And that about wraps it up. Thanks for you help and patience.
Click to expand...

Good debugging work! Keep us posted when the new case comes in.

Paul
 
well, I guess I was wrong about the case. I put everything into my
daughters case with the same result. So I guess (or hope, rather) that its
the monitoring chip. I wanna replace the board, but that would mean new
memory too and cash is too short for what I want to get.

Damn me

^JR^

Paul said:
"^JR^" said:
Well, I took my PS and put it in my daughters computer and it's hardware
monitor shows voltage within spec. I put her PS (which also showed within
spec on her monitor) and put it in my case and it reads out of spec as the
others have. I also removed all of my cards, as you suggested with the same
result using the MB alone.

While I haven't inspected the board itself yet, I'm now concluding my PS to
be good more than likely. My present generic case has a few years on it, so
I'm just gonna replace it. This case didn't even use brass stand offs
except at the top corner near the peripheral connections. All other
connection points are screwed into these little aluminum risers that clip on
to the MB plate and I never was happy with that so a new case for me, it is.
:)

By the way, I put my machine aside and am using my daughter's until I get a
new case.

And that about wraps it up. Thanks for you help and patience.
Click to expand...

Good debugging work! Keep us posted when the new case comes in.

Paul
^JR^
Paul said:
Hello, again. I have something interesting to report. To be safe, I
swapped out power supplys with a skyhawk 300w I had lying around. The
hardware monitor reads the +3.3v out of spec to the same degree, +3.9x.
This strikes me as odd that 2 PS's would read the same like that. Guess
I'll be REchecking with my meter. By the way, I did eliminate some
Click to expand...
of
the
other possibilities for the cause by trying a new power cord and plugging in
directly to the wall outlet rather than my surge protector. same results.
Stay tuned.....

^JR^

Interesting. Since both power supplies do the same thing, maybe you have
an inter-rail short ? That is a connection between say the +5V signal
and the +3.3V signal. The +5V signal tries to lift the +3.3V, and that
is how it is rising to +3.9 on both power supplies. This will make the
+5 supply lead much more heavily loaded, while the +3.3V supply gives
practically no current at all.

To debug this further, remove all the AGP and PCI cards. Handle them
by the edges and store them in the antistatic bag they came in.
Power up the system (yes, it will beep madly because the video card
is missing) and make your +3.3 measurement again. If the voltage has
returned to normal, one of the PCI/AGP cards has a rail-to-rail short.
Replace the cards one at a time, until you find that the PS is
overvolting again. The last card you put in is the faulty one.

If it still reads 3.9V with the motherboard alone, the rail to
rail short could be on or inside the motherboard. The motherboard
consists of four layers of copper sandwiched with non-conducting
fiberglass. You cannot examine the inner layers for faults. At
this point, I would pull the motherboard and do a thorough visual
inspection, such as looking for a brass standoff underneath the
motherboard that shouldn't be there etc. The standoffs are only
supposed to go, where they line up with holes in the motherboard.
Also, look for a conductor of some sort that has fallen across
the copper tracks on the motherboard.

An example of a manufacturing fault is pictured here. This is a
solder blob on a P4C800, and it is shorting two tracks together.
Look carefully at the lower right clump of solder and how it is
shorting to an adjacent track. That is the kind of thing you would
be looking for.

http://koti.mbnet.fi/~nightops/eki/DSC00249.JPG

If, after all of this work, you still cannot see the problem,
then RMA the motherboard. When you ship the motherboard, place
a note in the box explaining that you think there is a
rail-to-rail power short. In many cases, the board will be
discarded, as these faults cannot be fixed if they are inside
the PCB.

How we handle this in a lab situation, is we use an ohmmeter
to establish "norms" for a circuit board design. We measure the
resistance between the rails and ground, or from rail to rail.
It is amazing how closely a batch of boards agree on their
resistance between rails. After establishing what a "good" value
is for the design, when a freshly manufactured board comes in, a
quick check with the ohmmeter will tell whether there is something
grossly wrong with the board. In the case of a rail-to-rail
short, it is then detected before it can do any harm. Since you
only have one board to work with at home, you don't have that
luxury (of establishing a "norm").

HTH,
Paul


Click to expand...
Click to expand...
Click to expand...
 
"^JR^" said:
well, I guess I was wrong about the case. I put everything into my
daughters case with the same result. So I guess (or hope, rather) that its
the monitoring chip. I wanna replace the board, but that would mean new
memory too and cash is too short for what I want to get.

Damn me

^JR^
Click to expand...

For the monitor chip to be wrong, it could be the resistors used
for the voltage divider outside the chip, or it could be the chip
itself. As long as you've verified the supply voltages with your
meter, there won't be any harm running it with a defective
hardware monitor chip. It will just be annoying. If you have a
spare computer, is there any warranty time left on the motherboard ?

HTH,
Paul
Paul said:
"^JR^" said:
Well, I took my PS and put it in my daughters computer and it's hardware
monitor shows voltage within spec. I put her PS (which also showed within
spec on her monitor) and put it in my case and it reads out of spec as the
others have. I also removed all of my cards, as you suggested with the same
result using the MB alone.

While I haven't inspected the board itself yet, I'm now concluding my PS to
be good more than likely. My present generic case has a few years on it, so
I'm just gonna replace it. This case didn't even use brass stand offs
except at the top corner near the peripheral connections. All other
connection points are screwed into these little aluminum risers that clip on
to the MB plate and I never was happy with that so a new case for me, it is.
:)

By the way, I put my machine aside and am using my daughter's until I get a
new case.

And that about wraps it up. Thanks for you help and patience.
Click to expand...

Good debugging work! Keep us posted when the new case comes in.

Paul
Click to expand...
Click to expand...
 
Back
Top