PSU problems, is it faulty?

[sienk700]

Last week I found that my computer would no longer respond to the
power switch on the case. I tested the switch itself, it was fine. I
managed to boot the machine by grounding the green wire from the PSU,
but I still had to find out what was wrong. I tested the PSU in
another machine and it booted fine, so here I assumed the motherboard
was the problem.

After RMAing the motherboard and getting a replacement, the problem
was still there. No response from the power switch, no beeps or
anything.
I checked the voltage on the green wire from the PSU, it was around
3.5V. When I pressed the power switch it dropped to 0.1V, but the
machine still wasnt doing anything.
I then checked the purple wire which read over 5V when unconnected,
and around 3.7V when connected to the motherboard. (Which I read
somewhere is acceptable)

So I still had a machine that wouldnt boot and no idea what was going
on, then it just came on, a minute or two after I had pressed the
power switch it just blurted to life. After trying this a few times
the system seems to respond to the switch fine now, but when it does
the fans (the PSU and CPU fan) spin quite slowly for a few seconds
before picking up.

Anyway, my problem is I am worried whether I have a faulty PSU, as my
problem seems to have magically disappeared but I don't want to
continue running the machine as normal with something faulty and risk
damaging things.

The PSU is an Antec TruePower II 480w.

Any views? Or any further testing you can recommend I do?

Thanks

 

[Paul]

Last week I found that my computer would no longer respond to the
power switch on the case. I tested the switch itself, it was fine. I
managed to boot the machine by grounding the green wire from the PSU,
but I still had to find out what was wrong. I tested the PSU in
another machine and it booted fine, so here I assumed the motherboard
was the problem.

After RMAing the motherboard and getting a replacement, the problem
was still there. No response from the power switch, no beeps or
anything.
I checked the voltage on the green wire from the PSU, it was around
3.5V. When I pressed the power switch it dropped to 0.1V, but the
machine still wasnt doing anything.
I then checked the purple wire which read over 5V when unconnected,
and around 3.7V when connected to the motherboard. (Which I read
somewhere is acceptable)

So I still had a machine that wouldnt boot and no idea what was going
on, then it just came on, a minute or two after I had pressed the
power switch it just blurted to life. After trying this a few times
the system seems to respond to the switch fine now, but when it does
the fans (the PSU and CPU fan) spin quite slowly for a few seconds
before picking up.

Anyway, my problem is I am worried whether I have a faulty PSU, as my
problem seems to have magically disappeared but I don't want to
continue running the machine as normal with something faulty and risk
damaging things.

The PSU is an Antec TruePower II 480w.

Any views? Or any further testing you can recommend I do?

Thanks

The "Purple" wire is +5VSB. In whose book would 3.7V be acceptable
on a 5V output ? Definitely not acceptable. Typical output is 5V @ 2A
with maybe a tolerance of +/- 5 percent. The 3.7V is way out of spec.
Something is loading your "Purple" signal, such as an add-in card in
the motherboard.

Start stripping down hardware, until the signal goes back to 5.0V
or thereabouts. You don't even need to fully power on the machine
while you are doing this strip down. For each test, turn on the PSU
at the back, and see if the measured voltage on the purple signal is
5V or not. Then switch off again and remove the next piece of hardware.
With the main ATX power connector plugged to the motherboard, you can
access the purple potential via the back of the connector. You should
be able to jam your multimeter into the back of the connector, and
touch some metal on the purple pin.

A minimum test ?

Connect just the bare motherboard to the PSU. Nothing connected to the PANEL
header. Switch on the PSU at the back. Does the voltage still rise to only
3.7V ? If so, there are two possibilities. Either the PSU has a defective
(weak) +5VSB. Or the motherboard is drawing more than 2A through the
purple wire (internal motherboard short, external motherboard short like
a standoff, or a defective IC).

Having a clamp-on DC ammeter handy, would make measuring the current in
the purple wire easy. But the meter to do it is relatively expensive.
If you want to make a DC current measurement with a cheap multimeter,
you'd purchase an ATX power extender cable, cut the purple wire in two
on the extender, and carefully connect your multimeter, on the amps scale,
into the purple wire path. Generally speaking, you don't want the ammeter
connections to fall off while you are working, so by "carefully", what I
mean is, make sure the meter connections are secure. And that the meter probes
won't fall and contact anything else. If you have a well stocked electronics
store near you, you may be able to find banana plugs with fittings on
them suitable for helping you splice into the purple path. Banana plugs
are what will fit into the face of your multimeter.

If the motherboard is only drawing 1 ampere, and the supply is rated
at 2 amps output, then you know the power supply is weak and should be
replaced. And your motherboard manual may contain recommendations for
what kind of load the motherboard will put on the +5VSB line, for
things like STR (suspend to RAM) and the like. The loading could even
be coming from a bunch of USB devices you've got running off the +5VSB.
Some motherboards have a header for selecting +5V or +5VSB, to run
USB devices. The USB devices shouldn't be drawing current when not
being used, but things like USB battery chargers or USB vacuum cleaners
will do what they want.

Paul

 

[sienk700]

The "Purple" wire is +5VSB. In whose book would 3.7V be acceptable

on a 5V output ? Definitely not acceptable. Typical output is 5V @ 2A
with maybe a tolerance of +/- 5 percent. The 3.7V is way out of spec.
Something is loading your "Purple" signal, such as an add-in card in
the motherboard.

Start stripping down hardware, until the signal goes back to 5.0V
or thereabouts. You don't even need to fully power on the machine
while you are doing this strip down. For each test, turn on the PSU
at the back, and see if the measured voltage on the purple signal is
5V or not. Then switch off again and remove the next piece of hardware.
With the main ATX power connector plugged to the motherboard, you can
access the purple potential via the back of the connector. You should
be able to jam your multimeter into the back of the connector, and
touch some metal on the purple pin.

A minimum test ?

Connect just the bare motherboard to the PSU. Nothing connected to the PANEL
header. Switch on the PSU at the back. Does the voltage still rise to only
3.7V ? If so, there are two possibilities. Either the PSU has a defective
(weak) +5VSB. Or the motherboard is drawing more than 2A through the
purple wire (internal motherboard short, external motherboard short like
a standoff, or a defective IC).

Having a clamp-on DC ammeter handy, would make measuring the current in
the purple wire easy. But the meter to do it is relatively expensive.
If you want to make a DC current measurement with a cheap multimeter,
you'd purchase an ATX power extender cable, cut the purple wire in two
on the extender, and carefully connect your multimeter, on the amps scale,
into the purple wire path. Generally speaking, you don't want the ammeter
connections to fall off while you are working, so by "carefully", what I
mean is, make sure the meter connections are secure. And that the meter probes
won't fall and contact anything else. If you have a well stocked electronics
store near you, you may be able to find banana plugs with fittings on
them suitable for helping you splice into the purple path. Banana plugs
are what will fit into the face of your multimeter.

If the motherboard is only drawing 1 ampere, and the supply is rated
at 2 amps output, then you know the power supply is weak and should be
replaced. And your motherboard manual may contain recommendations for
what kind of load the motherboard will put on the +5VSB line, for
things like STR (suspend to RAM) and the like. The loading could even
be coming from a bunch of USB devices you've got running off the +5VSB.
Some motherboards have a header for selecting +5V or +5VSB, to run
USB devices. The USB devices shouldn't be drawing current when not
being used, but things like USB battery chargers or USB vacuum cleaners
will do what they want.

Paul

Thanks for the quick and detailed response.

With only the CPU and memory in the board I'm still getting around
3.7v on the "+5VSB" line. I unplugged everything and the reading is
still the same.

Unfortunately I don't have a way of safely checking the current (aside
from hand wiring each pin) but I'm quite sure the problem doesn't lie
with the motherboard since I've had these power issues on 2 different
boards now. I guess my best bet would be RMAing the PSU.

Thanks again Paul.

 

[Rod Speed]

Last week I found that my computer would no longer respond to
the power switch on the case. I tested the switch itself, it was fine.
I managed to boot the machine by grounding the green wire from
the PSU, but I still had to find out what was wrong. I tested the
PSU in another machine and it booted fine, so here I assumed
the motherboard was the problem.

After RMAing the motherboard and getting a replacement, the
problem was still there. No response from the power switch, no
beeps or anything. I checked the voltage on the green wire from
the PSU, it was around 3.5V. When I pressed the power switch
it dropped to 0.1V, but the machine still wasnt doing anything.

I then checked the purple wire which read over 5V when unconnected,
and around 3.7V when connected to the motherboard. (Which I read
somewhere is acceptable)

No its not, the power supply cant supply enough 5VSB current. Thats the problem.

So I still had a machine that wouldnt boot and no idea what
was going on, then it just came on, a minute or two after I
had pressed the power switch it just blurted to life. After
trying this a few times the system seems to respond to the
switch fine now, but when it does the fans (the PSU and
CPU fan) spin quite slowly for a few seconds before picking up.

Looks like the power supply is failing.

Anyway, my problem is I am worried whether I have a faulty PSU,

Yes you do.

as my problem seems to have magically disappeared

Because its an intermittent fault.

but I don't want to continue running the machine as
normal with something faulty and risk damaging things.

Yeah, the 12V rail sagging initially isnt great for the hard drive(s)

The PSU is an Antec TruePower II 480w.

Any views? Or any further testing you can recommend I do?

You'll likely find that the 12V rails (yellow) sag when the fans
spin slowly and that isnt going to be great for the hard drive(s)

 

[Rod Speed]

The "Purple" wire is +5VSB. In whose book would 3.7V be acceptable
on a 5V output ? Definitely not acceptable. Typical output is 5V @ 2A
with maybe a tolerance of +/- 5 percent. The 3.7V is way out of spec.
Something is loading your "Purple" signal, such as an add-in card in
the motherboard.

Start stripping down hardware, until the signal goes back to 5.0V
or thereabouts. You don't even need to fully power on the machine
while you are doing this strip down. For each test, turn on the PSU
at the back, and see if the measured voltage on the purple signal is
5V or not. Then switch off again and remove the next piece of
hardware. With the main ATX power connector plugged to the motherboard, you can
access the purple potential via the back of the connector. You should
be able to jam your multimeter into the back of the connector, and
touch some metal on the purple pin.

A minimum test ?

Connect just the bare motherboard to the PSU. Nothing connected to
the PANEL header. Switch on the PSU at the back. Does the voltage
still rise to only 3.7V ? If so, there are two possibilities. Either the PSU has a defective
(weak) +5VSB. Or the motherboard is drawing more than 2A through the purple wire (internal
motherboard short,

Unlikely given that he's swapped the motherboard.

external motherboard short like a standoff,

And the best test for that is running it loose on the desktop.

or a defective IC).

Unlikely given that he's swapped the motherboard.

Having a clamp-on DC ammeter handy, would make measuring the current
in the purple wire easy. But the meter to do it is relatively expensive.
If you want to make a DC current measurement with a cheap multimeter,
you'd purchase an ATX power extender cable, cut the purple wire in two
on the extender, and carefully connect your multimeter, on the amps
scale, into the purple wire path. Generally speaking, you don't want
the ammeter connections to fall off while you are working, so by
"carefully", what I mean is, make sure the meter connections are secure. And that the meter probes
won't fall and contact anything else. If you have a well stocked electronics store near you, you
may be able to find banana plugs with fittings on them suitable for helping you splice into the
purple path. Banana plugs are what will fit into the face of your multimeter.

Much more viable to swap the power supply given that the motherboard
has already been swapped and the 12V rail appears to be sagging.

If the motherboard is only drawing 1 ampere, and the supply is rated
at 2 amps output, then you know the power supply is weak and should be
replaced. And your motherboard manual may contain recommendations for
what kind of load the motherboard will put on the +5VSB line, for things like STR (suspend to RAM)
and the like. The loading could even be coming from a bunch of USB devices you've got running off
the +5VSB.

Some motherboards have a header for selecting +5V or +5VSB, to run
USB devices. The USB devices shouldn't be drawing current when not
being used, but things like USB battery chargers or USB vacuum cleaners will do what they want.

Much more likely to be a flakey power supply.

 

[Paul]

No its not, the power supply cant supply enough 5VSB current. Thats the problem.


Looks like the power supply is failing.


Yes you do.


Because its an intermittent fault.


Yeah, the 12V rail sagging initially isnt great for the hard drive(s)



You'll likely find that the 12V rails (yellow) sag when the fans
spin slowly and that isnt going to be great for the hard drive(s)

There is another interpretation. The 5VSB is also a measure
of the state of the supervisor portion of the power supply.
The low 12V is because the supply is only "half switched on".
It eventually transitions out of that zone, as the control
in the supply, eventually switches the supply fully on.

Paul

 

[Rod Speed]

Rod Speed wrote

There is another interpretation. The 5VSB is also a measure
of the state of the supervisor portion of the power supply.

No it isnt. The ATX standard doesnt say that.

The low 12V is because the supply is only "half switched on".

Not even possible. The standard says that it
should shut down cleanly if it cant start properly.

It eventually transitions out of that zone, as the control
in the supply, eventually switches the supply fully on.

Its much more likely that its just a flakey power supply
that cant deliver the full 5VSB current and cant start
properly either and that the fault is intermittent.

Its certainly possible that a short to case outside the power supply
is the problem with both rail voltages, but again, the standard requires
that the supply shut down cleanly in that situation, not stagger along
with sagged rails and so its much more likely just a flakey power supply.

 

[Ken Maltby]

No it isnt. The ATX standard doesnt say that.


Not even possible. The standard says that it
should shut down cleanly if it cant start properly.


Its much more likely that its just a flakey power supply
that cant deliver the full 5VSB current and cant start
properly either and that the fault is intermittent.

Its certainly possible that a short to case outside the power supply
is the problem with both rail voltages, but again, the standard requires
that the supply shut down cleanly in that situation, not stagger along
with sagged rails and so its much more likely just a flakey power supply.

OK you two, in both situations the solution is to replace the PSU,
which the OP has figured out by now.

Peace, Kumbaya, Can't we all just get along?

Luck;
Ken

 

[Rod Speed]

OK you two, in both situations the solution is to replace the PSU, which the OP has figured out
by now.

Peace, Kumbaya, Can't we all just get along?

What matters much more is discussing what can be diagnosed
with unusual symptoms like that to make fault finding quicker in
future, particulary not bothering with the unlikely short to case
possibility until a PSU swap makes no difference.