Power supply 5vsb dying ?

[pierre_catello]

Hello,

I have a problem with my one my home pc : all my power supplies, after
12-18 months of use, always end in failing to power my PC in standby
(PC won't wake up from suspend to ram, both with windows and linux).
Except for the standby problem, everything works fine, so I suspect the
5v line supplying the standby mode to be failing. It is my third PSU
which is doing that, and I'm wondering what is killing them like
that...

The system is a regular athlon XP 3000+, not overclocked, with 2x512
DDR cas2.5 Corsair ram modules, 6600GT graphic adapter (AGP).
The computer is not used for 3D games (so the GPU shouyld not be hungry
on the PSU), but as I do a lot of DV editing, encoding and archiving, I
often use several drives :
- 2 SATA 7200 drives, 1 ATA 7200 drive, 2 SCSI DVD burner and 1 ATA DVD
reader, 1 firewire external drive and 1 USB2 external drive (the two
latter one are not powered by the USB bus so they shouldn't cause
troubles).

My last PSU is a 420W ("LC Power" brand) average quality. The two
previous ones were Enermax in the 400W and 80 Euros range, and failed
the same way (no more standby, and occasional hard drive reset) after
about 18 months.

- Do you think my system overloading the PSU ?
- I often let the system in standby (suspend to ram), sometimes several
days. Is it this mode which may be killing the 5vsb ? is it bad for the
PSU ?

As a side point, I wondering if it is bad to continue to run a PC what
it seems to be a dying PSU (giving up the standby mode, as the PC works
fine otherwise), can the PSU kills the motherboard or other components
? In fact, I'm wondering if it would be safe to "re-cycle" this PSU for
a small server which is running 24h/day (not needing standby), or if
I'm running a risk of killing it.

Thanks for your help

 

[Conor]

Hello,

I have a problem with my one my home pc : all my power supplies, after
12-18 months of use, always end in failing to power my PC in standby
(PC won't wake up from suspend to ram, both with windows and linux).
Except for the standby problem, everything works fine, so I suspect the
5v line supplying the standby mode to be failing. It is my third PSU
which is doing that, and I'm wondering what is killing them like
that...

Probably nothing. It could be all down to bad drivers.

 

[pierre_catello]

Conor wrote :

Probably nothing. It could be all down to bad drivers.

Drivers have nothing to do with that :
- Replacing the the PSU solves the issue without changing any
software... And it is the third time!
- ACPI handling for both linux (hard compiled in the kernel) and
windows (Nvidia Nforce2 support software) were never reconfigured, so
the problem can in no way comes from improper driver support.

P.

 

[Rod Speed]

I have a problem with my one my home pc : all my power supplies, after
12-18 months of use, always end in failing to power my PC in standby
(PC won't wake up from suspend to ram, both with windows and linux).
Except for the standby problem, everything works fine, so I suspect
the 5v line supplying the standby mode to be failing. It is my third PSU
which is doing that, and I'm wondering what is killing them like that...

Yeah, rather odd.

The system is a regular athlon XP 3000+, not overclocked, with 2x512
DDR cas2.5 Corsair ram modules, 6600GT graphic adapter (AGP).
The computer is not used for 3D games (so the GPU shouyld not be
hungry on the PSU), but as I do a lot of DV editing, encoding and
archiving, I often use several drives :
- 2 SATA 7200 drives, 1 ATA 7200 drive, 2 SCSI DVD burner and 1 ATA
DVD reader, 1 firewire external drive and 1 USB2 external drive (the
two latter one are not powered by the USB bus so they shouldn't cause
troubles).

Shouldnt be a problem even if they were.

My last PSU is a 420W ("LC Power" brand) average quality.
The two previous ones were Enermax in the 400W and 80
Euros range, and failed the same way (no more standby,
and occasional hard drive reset) after about 18 months.

- Do you think my system overloading the PSU ?

Shouldnt produce that result even if it was.

- I often let the system in standby (suspend to ram), sometimes
several days. Is it this mode which may be killing the 5vsb ?

Thats certainly a significant load on the 5VSB.

is it bad for the PSU ?

Shouldnt be, it should be able to deliver that forever
as long as the 5VSB current is within the power supply
specs for the 5VSB, particularly with the Enermax.

As a side point, I wondering if it is bad to continue to
run a PC what it seems to be a dying PSU (giving up
the standby mode, as the PC works fine otherwise),
can the PSU kills the motherboard or other components ?

Yes, a badly designed PSU can kill what its powering as it dies.

In fact, I'm wondering if it would be safe to "re-cycle"
this PSU for a small server which is running 24h/day
(not needing standby), or if I'm running a risk of killing it.

If the 5VSB is dying because of bad caps in the power supply,
there is certainly some risk that there are more caps going bad
in the power supply and killing the system when more dies.

 

[pierre_catello]

Rod Speed wrote :

If the 5VSB is dying because of bad caps in the power supply,
there is certainly some risk that there are more caps going bad
in the power supply and killing the system when more dies.

Thanks for the warning...
Throwing away the PSU is likely the safe choice; My hesitation is
because I'm not sure whether the PSU is really dying, or whether is has
just became a bit weak on the 5vsb, and the motherboard is maybe a tad
too sensitive?

P.

 

[Rod Speed]

(e-mail address removed) wrote

Rod Speed wrote

Thanks for the warning...
Throwing away the PSU is likely the safe choice; My hesitation is
because I'm not sure whether the PSU is really dying, or whether is
has just became a bit weak on the 5vsb, and the motherboard is maybe
a tad too sensitive?

It isnt trivial to prove that.

What might be happening is that the 5VSB load is higher than
the 5VSB rail from the power supplys can supply and they are
shutting down the 5VSB rail but not the other rails, but that is
very unlikely given that it takes 12-18 months to happen.

I'd personally just measure the current the motherboard is taking
on the 5VSB rail when in suspend to ram mode, and compare that
with the 5VSB rail current rating with the Enermax particularly,
but it isnt that easy to measure rail currents without the right
equipment, its easiest with a clamp meter on the wire itself.

Its more likely that the continuing over current on the 5VSB rail
eventually kills the 5VSB rail, and its separate from the main
rails, but its hard to believe that that current which is only a
couple of amps at most is likely to kill anything. You'd expect
that the most it might do is cause the 5VSB rail to sag below
what it should be voltage wise. And still hard to see why you'd
get the 12-18 months effect. That 12-18 months effect sound much
more like bad caps, but its hard to believe you'd get that sort
of reproducible effect 3 times in a row, with three completely
different power supply designs.

 

[omagawd]

I often use several drives :
- 2 SATA 7200 drives, 1 ATA 7200 drive, 2 SCSI DVD burner and 1 ATA DVD
reader, 1 firewire external drive and 1 USB2 external drive (the two
latter one are not powered by the USB bus so they shouldn't cause
troubles).

My last PSU is a 420W

With that many drives I'd use a much bigger power supply. The surge
current at boot must be funny. Quit screwing around and get
something like this:

http://www.newegg.com/Product/Product.asp?Item=N82E16817703005

 

[lsmartino]

Rod Speed ha escrito:

(e-mail address removed) wrote


It isnt trivial to prove that.

What might be happening is that the 5VSB load is higher than
the 5VSB rail from the power supplys can supply and they are
shutting down the 5VSB rail but not the other rails, but that is
very unlikely given that it takes 12-18 months to happen.

I'd personally just measure the current the motherboard is taking
on the 5VSB rail when in suspend to ram mode, and compare that
with the 5VSB rail current rating with the Enermax particularly,
but it isnt that easy to measure rail currents without the right
equipment, its easiest with a clamp meter on the wire itself.

The diagnosis and troubleshooting method is right, but a clamp ammeter
will not be useful here because clamp ammeters only work with AC, and
the computer PSU produces DC. DC voltage flow produces no alternating
magnetic field around the conductor, which is precisely what is used by
a clamp ammeter to measure current.

To measure DC current you need to insert the ammeter in the circuit.

Its more likely that the continuing over current on the 5VSB rail
eventually kills the 5VSB rail, and its separate from the main
rails, but its hard to believe that that current which is only a
couple of amps at most is likely to kill anything. You'd expect
that the most it might do is cause the 5VSB rail to sag below
what it should be voltage wise. And still hard to see why you'd
get the 12-18 months effect. That 12-18 months effect sound much
more like bad caps, but its hard to believe you'd get that sort
of reproducible effect 3 times in a row, with three completely
different power supply designs.

Agreed.

 

[Rod Speed]

Rod Speed wrote

The diagnosis and troubleshooting method is right, but a clamp ammeter
will not be useful here because clamp ammeters only work with AC,
Wrong.

and the computer PSU produces DC. DC voltage flow produces
no alternating magnetic field around the conductor, which is
precisely what is used by a clamp ammeter to measure current.

That last is just plain wrong.

 

[pierre_catello]

Thanks to all of you for the inputs
I still don't figure out what is making my PSU troublesome after some
months of use (weird problem, only affecting standby mode), and I will
get a bigger PSU in the 600W range just to be safe. The new Corsair HX
(rebranded Seasonic?) offer 5 year of guarantee, so at least I will be
able to get a new one if it comes to fail like the others. I don't have
a clampmeter, but I was also thinking that they were only designed for
mesuring AC current (from induction effect)...

P.

 

[Rod Speed]

Thanks to all of you for the inputs
I still don't figure out what is making my PSU troublesome after some
months of use (weird problem, only affecting standby mode), and I will
get a bigger PSU in the 600W range just to be safe. The new Corsair HX
(rebranded Seasonic?) offer 5 year of guarantee, so at least I will be
able to get a new one if it comes to fail like the others. I don't
have a clampmeter, but I was also thinking that they were only
designed for mesuring AC current (from induction effect)...

Nar, thats just the most common style.

Thats what I meant about unusual test hardware, most dont have
a DC clamp meter available and its not trivial to measure it otherwise.

 

[w_tom]

Thanks to all of you for the inputs
I still don't figure out what is making my PSU troublesome after some
months of use (weird problem, only affecting standby mode), and I will
get a bigger PSU in the 600W range just to be safe. The new Corsair HX
(rebranded Seasonic?) offer 5 year of guarantee, so at least I will be
able to get a new one if it comes to fail like the others. I don't have
a clampmeter, but I was also thinking that they were only designed for
mesuring AC current (from induction effect)...

And that is the point, right? You keep replacing things and problem
remains. The 600 watt supply will not solve anything. You have
already done shotgunning to no purpose. No purpose? Did you first
identify the problem - or just start curing symptoms? You don't even
know if power supply or other parts of a power supply system are
defective. Now do what engineers do to solve problems the first time.
First get facts.

Start by measuring +5VSB (purple wire) voltage with system off and AC
receptacle connected. It must measure above 4.87, must remain stable
also when power switch is pressed, and when in standby mode. If so,
that power supply output is OK. Move on.

Next measure green wire with computer off. It must measure above 2
volts and immediately measure less than 0.8 volts when switch is
pressed. Do these numbers appear same both when system does and does
not start? BTW, what we are doing is also confirming current - and
without an expensive clamp meter.

Also important are gray wire voltage that must rise well above 2.4
volts within but seconds of when power switch is pressed - and remain
above 2.4 volts.

Now for voltages on any of red, orange, and yellow wires. Each must
rise when switch is pressed to above 3.23, 4.87, and 11.7. Furthermore
these voltages must remain above those numbers when everything else is
being used simultaneously (multitasking to all peripherals). If these
numbers are correct, then more on to other suspects - all currents are
more than sufficient. If currents are marginal, then the voltage will
also drop to or below those limits values.

Finally meter can also confirm integrity of CMOS battery - typically
a 3 volt coin cell mounted on motherboard. If voltage is below 2.8,
battery is gettting old. If voltage is below 2.4, then it might
contribute to problems.

IOW don't just 'try this and try that'. Either it is problematic, or
move on to other suspects. Number problems can identify which
component of a power supply 'system' has failed - and without all these
silly power supply purchases.

Why do you replace something? Because well proven diagnostic
procedures first identify a suspect. Don't shotgun. Get a 3.5 digit
multimeter which becomes one of your most useful tools for all
electronic and electrical solutions. If +5VSB is dying or cannot
provide sufficient current, a multimeter voltage reading will identify
that problem immediately - without buying more power supplies..

 

[Rod Speed]

And that is the point, right? You keep replacing things and problem
remains. The 600 watt supply will not solve anything. You have
already done shotgunning to no purpose. No purpose? Did you first
identify the problem - or just start curing symptoms? You don't even
know if power supply or other parts of a power supply system are
defective. Now do what engineers do to solve problems the first time.
First get facts.

Start by measuring +5VSB (purple wire) voltage with system off and AC
receptacle connected. It must measure above 4.87, must remain stable
also when power switch is pressed, and when in standby mode. If so,
that power supply output is OK. Move on.

Next measure green wire with computer off. It must measure above 2
volts and immediately measure less than 0.8 volts when switch is
pressed. Do these numbers appear same both when system does and does
not start? BTW, what we are doing is also confirming current - and
without an expensive clamp meter.

Also important are gray wire voltage that must rise well above 2.4
volts within but seconds of when power switch is pressed - and remain
above 2.4 volts.

Now for voltages on any of red, orange, and yellow wires. Each must
rise when switch is pressed to above 3.23, 4.87, and 11.7.

He keeps spewing this pig ignorant drivel, that isnt what the ATX specs say.

 

[pierre_catello]

Why do you replace something? Because well proven diagnostic
procedures first identify a suspect. Don't shotgun. Get a 3.5 digit
multimeter which becomes one of your most useful tools for all
electronic and electrical solutions. If +5VSB is dying or cannot
provide sufficient current, a multimeter voltage reading will identify
that problem immediately - without buying more power supplies..

What I'm trying to figure out is what is killing my PSUs... It seems
obvious that the PSUs which failed have a defective standby power.
I'm not sure to understand how your well defined procedure will help in
determining the cause of the problem?

P.

 

[Rod Speed]

What I'm trying to figure out is what is killing my PSUs... It seems
obvious that the PSUs which failed have a defective standby power.
I'm not sure to understand how your well defined procedure will help
in determining the cause of the problem?

It wont, its a canned response he just keep respouting
whenever anyone mentions a power supply.

And much of it is just plain wrong too.

 

[Rod Speed]

Rod Speed wrote :

Thanks for the warning...
Throwing away the PSU is likely the safe choice; My hesitation is
because I'm not sure whether the PSU is really dying, or whether is
has just became a bit weak on the 5vsb, and the motherboard is maybe
a tad too sensitive?

On further thought its more likely to be something as basic as
3 power supplys that havent been properly designed on the 5VSB
side, that gets too hot in standby mode when running at your high
5VSB loads due to suspending to ram, and that eventually kills
the 5VSB side of the power supply in the hottest weather.

If thats what is actually happening, and its not that easy to prove,
unless you actually open the failed supplys and check what has
actually failed in the 5VSB supply, it should be fine to use those
supplys in systems where you dont need a 5VSB rail.

Quite a few systems do need it for something as basic as starting tho.

 

[w_tom]

What I'm trying to figure out is what is killing my PSUs... It seems
obvious that the PSUs which failed have a defective standby power.
I'm not sure to understand how your well defined procedure will help in
determining the cause of the problem?

Currently you have posted nothing that others can use to answer your
question. Those numbers from a meter report what is and is not
defective. Numbers are essential to ask why. Without specific facts,
then every reply would be nothing more than wild speculation - no
different from shotgunning. Remember, making something appear to be
fixed -shotgunning - can result in future failures because shotgunning
only addresses symptoms and never asks why. Use the meter and report
all numbers. Without numbers, nobody knows anything. The procedure
does something that is fundamental - asks for 'whys' before fixing
anything.

Again, shotgunning implies one need not know why. Collecting facts
before replacing anything means learning 'why' so that problem is
solved the first time.

 

[Davej]

I have a problem with my one my home pc : all my power supplies, after
12-18 months of use, always end in failing to power my PC in standby
(PC won't wake up from suspend to ram, both with windows and linux).
Except for the standby problem, everything works fine, so I suspect the
5v line supplying the standby mode to be failing. It is my third PSU
which is doing that, and I'm wondering what is killing them like
that...
[...]
- Do you think my system overloading the PSU ?
- I often let the system in standby (suspend to ram), sometimes several
days. Is it this mode which may be killing the 5vsb ? is it bad for the
PSU ?

Maybe you have an intermittent short between +5VSB and ground? You
could add an inline fuse to protect it. You could add a cheap open
frame 5V supply to power it.

 

[pierre_catello]

On further thought its more likely to be something as basic as
3 power supplys that havent been properly designed on the 5VSB
side, that gets too hot in standby mode when running at your high
5VSB loads due to suspending to ram, and that eventually kills
the 5VSB side of the power supply in the hottest weather.

Thanks again Rod. Now that you mention heat, I realize that at least
two of my PSUs failed in the summer...

 

[w_tom]

. Now that you mention heat, I realize that at least
two of my PSUs failed in the summer...

5VSB supplies so little power that standby should not overheat
anything. Many hardware problems that always existed may not become
apparent until room temperature rises slightly. This is why
manufacturers of higher reliability hardware do burn-in testing.

Contrary to popular myth, burn-in testing means putting room
temperature up to maximum, executing everything worse case, lowering
room temperature to absolute minimum, then doing same tests. Then
temperature cycle equipment from below min to above max to min to max
.... then do more testing at max and min temperatures.

Well, you are not going to do all that. But running a computer for
some time also at room temperatures that are or exceeds 100 degree F
can find defects early. That is perfectly normal temperature to all
computers, does not create any harm, AND tends to find a defect while
hardware is still in warranty. IOW if that 5VSB was defective, then
5VSB problem might have been detected long ago (before failure became
problematic).

Why would power supplies fail in summer? Because supply was always
defective, not tested previously in a 100 degree F room, and symptoms
of that failure became apparent in summer. Summer did not create the
problem. Summer heat finally detected an existing condition.

If a power supply 5VSB was harmed by standby memory load, that power
supply was defective when purchased. Standby power is just too trivial
to create failure in any properly designed hardware.