Trying to salvage computer from minor electical problem

[Damn Dan]

I could use some help troubleshooting my computer with some strange
electrical issues.

I was having a power problem, so I did some testing and figured out the
"remote" wire wasn't being properly grounded by the motherboard. Thus,
the power supply was never getting the signal from the motherboard to
turn on. So I spliced the remote line and hardwired it to the
(grounded) computer chassis. The only inconvenience is I have to cut
off power using the toggle switch in the back when I shutdown, or the
motherboard stays awake and the processor fan keeps spinning. And to
boot up, I hit the toggle switch (which gives power to the motherboard)
and then the power button in the front, which activates the BIOS.

My problem is the BIOS doesn't always activate. I usually have to rock
the toggle switch a few times while hitting the power button until the
computer eventually boots. I don't know anything about what the BIOS
is expecting when it boots up, so I'm not sure where to begin to
troubleshoot this. The only clue I've found so far is that when the
BIOS doesn't activate, the "RAM Status" LED on the motherboard is
usually strobing. And when the BIOS finally does decide to activate,
that LED goes steady.

My guess is that since I'm manually powering the motherboard on and
off, it's messing up some logic that the BIOS expects during boot up
and shutdown.

My motherboard is a Gigabyte K8NS Pro.

Any thoughts?

-Dan

 

[Paul]

I could use some help troubleshooting my computer with some strange
electrical issues.

I was having a power problem, so I did some testing and figured out the
"remote" wire wasn't being properly grounded by the motherboard. Thus,
the power supply was never getting the signal from the motherboard to
turn on. So I spliced the remote line and hardwired it to the
(grounded) computer chassis. The only inconvenience is I have to cut
off power using the toggle switch in the back when I shutdown, or the
motherboard stays awake and the processor fan keeps spinning. And to
boot up, I hit the toggle switch (which gives power to the motherboard)
and then the power button in the front, which activates the BIOS.

My problem is the BIOS doesn't always activate. I usually have to rock
the toggle switch a few times while hitting the power button until the
computer eventually boots. I don't know anything about what the BIOS
is expecting when it boots up, so I'm not sure where to begin to
troubleshoot this. The only clue I've found so far is that when the
BIOS doesn't activate, the "RAM Status" LED on the motherboard is
usually strobing. And when the BIOS finally does decide to activate,
that LED goes steady.

My guess is that since I'm manually powering the motherboard on and
off, it's messing up some logic that the BIOS expects during boot up
and shutdown.

My motherboard is a Gigabyte K8NS Pro.

Any thoughts?

-Dan

"RAM_LED" is connected to standby power (so says the manual). I interpret
that to mean, it is a motherboard LED like many others use, and is connected
directly to +5VSB rail. It is basically telling you when +5VSB is present.
If +5VSB is present, it is used to refresh the RAM while in standby suspend
to RAM, which is why you should not remove sticks of RAM while that LED
is lit.

If you see that LED flashing on and off, it means the +5VSB is being overloaded.
That could mean that the power supply is defective.

To lend sanity to the motherboard, while using your "grounded PS_ON#" trick,
you should try the RESET button. That will feed a reset pulse to the mothetboard,
allowing an orderly start. But if the +5VSB blinks after you do that, then
that'll drive things nuts. You see, the +5VSB is also used by the power supply,
as part of its supervisor circuit. The state of the +5VSB, is effectively the
state of the poor power supply's brain, and if the +5VSB blinks, then the other
outputs from the supply should stop at the same time.

Your problem could be a bad power supply, or it could be something on the
motherboard or otherwise, that is overloading the power supply. Unplugging
stuff, a bit at a time, may uncover a defective part, if that is the
problem. Swapping the power supply, may help highlight whether that is the
problem.

Based on frequency of occurrence, a bad power supply is the most likely
fault.

The RAM status LED on your board should not "strobe". A motherboard LED like
that is supposed to stay on solid, all the time that the switch on the back
of the PSU is in the ON position.

Paul

 

[Rod Speed]

I could use some help troubleshooting my computer with some strange
electrical issues.

I was having a power problem, so I did some testing and figured out
the "remote" wire wasn't being properly grounded by the motherboard.
Thus, the power supply was never getting the signal from the
motherboard to turn on. So I spliced the remote line and hardwired
it to the (grounded) computer chassis. The only inconvenience is I
have to cut off power using the toggle switch in the back when I
shutdown, or the motherboard stays awake and the processor fan keeps
spinning. And to boot up, I hit the toggle switch (which gives power
to the motherboard) and then the power button in the front, which
activates the BIOS.

My problem is the BIOS doesn't always activate. I usually have to
rock the toggle switch a few times while hitting the power button
until the computer eventually boots. I don't know anything about
what the BIOS is expecting when it boots up, so I'm not sure where to
begin to troubleshoot this. The only clue I've found so far is that
when the BIOS doesn't activate, the "RAM Status" LED on the
motherboard is usually strobing. And when the BIOS finally does
decide to activate, that LED goes steady.

My guess is that since I'm manually powering the motherboard on and
off, it's messing up some logic that the BIOS expects during boot up
and shutdown.

My motherboard is a Gigabyte K8NS Pro.

Any thoughts?

Likely the same fault that stops it grounding the PS_ON#
line is what is producing that multiple attempts to reboot.

Likely bad caps on the motherboard, even if they arent visibly bad.

Could also be a lack of a proper reset since you arent using
the normal power button anymore. Try doing both, using your
new switch AND the power button to start the system.

 

[Damn Dan]

Likely the same fault that stops it grounding the PS_ON#
line is what is producing that multiple attempts to reboot.

Likely bad caps on the motherboard, even if they arent visibly bad.

Could also be a lack of a proper reset since you arent using
the normal power button anymore. Try doing both, using your
new switch AND the power button to start the system.

Thank you both for your input. Capacitors have been a suspicion of
mine all along and I think that's what it is in this case. I noticed
today that flicking the toggle switch in the back several times is not
necessary. Instead, I have to turn it on once and wait for the RAM LED
(it strobes for about 60 seconds and then goes steady). After that,
recycling power once successfully boots the computer. I don't even
have to press the power button in the front. Also, if I shutdown and
immediately power up again, the RAM LED stays steady and it'll boot
right back up. If I wait a while between shutting down and booting up,
I have to wait for the strobing to stop again. I can't think of
anything other than a faulty capacitor that could cause these symptoms.

So then, the big question: is it safe to continue using my "hack"
bootup method? I know my motherboard or power source is bad and will
inevitably need replacement eventually. However, is there any risk to
the other components in my computer?

 

[Rod Speed]

Thank you both for your input. Capacitors have been a suspicion of
mine all along and I think that's what it is in this case. I noticed
today that flicking the toggle switch in the back several times is not
necessary. Instead, I have to turn it on once and wait for the RAM
LED (it strobes for about 60 seconds and then goes steady). After
that, recycling power once successfully boots the computer. I don't
even have to press the power button in the front. Also, if I
shutdown and immediately power up again, the RAM LED stays steady and
it'll boot right back up. If I wait a while between shutting down
and booting up, I have to wait for the strobing to stop again. I
can't think of anything other than a faulty capacitor that could
cause these symptoms.

Yeah, me too.

So then, the big question: is it safe to continue using my "hack" bootup method?

Can you try another power supply easily ?

The problem is that if the power supply is the problem,
cheap ones can die rather spectacularly and kill what is
powered from them, so if its the power supply thats
taking a long time to bring up the 5VSB which is what
the LED is powered by, you are taking a real risk
continuing to use it with known bad caps in it.

I know my motherboard or power source is bad and will
inevitably need replacement eventually. However, is there
any risk to the other components in my computer?

If you get the same effect with a different power supply,
the motherboard caps are the problem and the risk of
damaging anything by continuing to use it like that is small.

Not zero, but small.

 

[jt3]

How old is your mbd? GigaByte has a 3-year warranty, dating from the date
of manufacture, and if you still fall into that slot, consider requesting an
RMA. They did it for mine, same mbd; came back with *all* the electrolytics
replaced. Works fine, since then.

Hth,
Joe

 

[Damn Dan]

Yeah, me too.


Can you try another power supply easily ?

The problem is that if the power supply is the problem,
cheap ones can die rather spectacularly and kill what is
powered from them, so if its the power supply thats
taking a long time to bring up the 5VSB which is what
the LED is powered by, you are taking a real risk
continuing to use it with known bad caps in it.


If you get the same effect with a different power supply,
the motherboard caps are the problem and the risk of
damaging anything by continuing to use it like that is small.

Not zero, but small.

Unfortunately, I don't have a spare power source to throw in.
Although, I did pull out the block and test all the voltages (by first
grounding the PS_ON wire). Granted, the power source was not under
load, but all the voltages did check out, including the 5V'er (what's
the "SB" stand for on 5VSB?).

 

[Damn Dan]

How old is your mbd? GigaByte has a 3-year warranty, dating from the date
of manufacture, and if you still fall into that slot, consider requesting an
RMA. They did it for mine, same mbd; came back with *all* the electrolytics
replaced. Works fine, since then.

Hth,
Joe

jt3,

Interesting you had a Gigabyte with capacitor failures. Did it have
similar symptoms as mine before you got it repaired? Thanks for
mentioning the warranty. I'll check the date on mine tonight and give
them a call. Although.... isn't Gigabyte in Taiwan or something? Did
you have to ship yours over seas or do they a facility closer?

 

[Rod Speed]

Unfortunately, I don't have a spare power source to throw in.
Although, I did pull out the block and test all the voltages (by first
grounding the PS_ON wire). Granted, the power source was not under
load, but all the voltages did check out, including the 5V'er (what's
the "SB" stand for on 5VSB?).

Stand by, what powers the motherboard for the power
switch and start on mouse and keyboard etc to work.

 

[jt3]

Some of the early symptoms were. Later, I started having problems with it
recognizing the memory in two of the slots (switching the modules showed
that).

The factory is in Taiwan, or perhaps somewhere else thereabouts. But the
service location was in Southern California, City of Industry, or such, as I
recall. They were actually pretty prompt about it. If I hadn't made
certain to record all the printed stuff on the mbd, I'd not have been
certain it was my mbd. That and being able to tell the caps had been
resoldered. Anyhow, XP gave no trouble on the return.

Good Luck,
Joe

 

[w_tom]

So far all your attempt do not and cannot identify a problem. They
are only $20 - were recently on sale for $10 at Sears. That 3.5 digit
multimeter reports in but two minutes the source of your problem. Even
if lights illuminate, that voltage can still be 100% defective. The
meter is as essential as a screw driver.

Start by measuring purple wire voltage (from power supply to
motherboard). Notice everything done without disconnecting anything.
With computer off but power cord connected to AC receptacle, that
voltage must exceed 4.87 VDC. Next measure green wire voltage. That
must be more than 2.0 volts and then drop to below 0.8 volts when power
switch is pressed.

Well if those are correct, then power supply controller told power
supply to turn on. Next measure gray wire voltages as switch is
pressed. That voltage must rise to well above 2.4 volts within a
second after power switch is pressed. This is power supply telling
motherboard that it is OK. Next monitor one of red, yellow, and orange
wires. Each will rise above 3.23, 4.87, or 11.7 VDC within seconds and
remain there. Those voltages should remain constant (above minimal
limits) even as all peripherals are multitasked - accessed
simultaneously. If any voltage does not try to rise, then that voltage
has too much load (too many peripherals, short circuit, bad capacitor,
etc). If any voltage drops too low as all peripherals are accessed,
then that one part of supply has a defect.

Not only should this information identify a suspect in but minutes.
Posting those numbers here can provide even more useful information.
Currently your symptoms are those of an intermittent that could have
been identified long ago.

Swapping power supply sometimes can even complicate problems. Once
those measurements identify the suspect, only then do you replace that
part. Then confirm that part solved the problems by taking voltage
measurements again. Solve a problem the first time; don't shotgun.

+5VSB is 5 volts standy. Power supply controller needs that voltage
to monitor front panel power switch.

 

[Rod Speed]

So far all your attempt do not and cannot identify a problem.
They are only $20 - were recently on sale for $10 at Sears.

Unfortunately no one can actually sell you an understanding of how they work.

That 3.5 digit multimeter reports in but two minutes the source of your problem.

Bet it doesnt.

Even if lights illuminate, that voltage can still be 100% defective.

Nope, 100% means its not there at all, stupid.

The meter is as essential as a screw driver.

Wrong, as always.

Start by measuring purple wire voltage (from power supply to motherboard).

He's already proved that that isnt being generated
by the motherboard by replacing it, stupid.

Notice everything done without disconnecting anything.
With computer off but power cord connected to AC
receptacle, that voltage must exceed 4.87 VDC.

Wrong, as always.

Next measure green wire voltage. That must be more than 2.0 volts
and then drop to below 0.8 volts when power switch is pressed.

He's already proved that that doesnt happen, stupid.

Well if those are correct, then power supply
controller told power supply to turn on.

He's already proved that that doesnt happen, stupid.

Next measure gray wire voltages as switch is pressed.
That voltage must rise to well above 2.4 volts within a
second after power switch is pressed.

He's already proved that that doesnt happen, stupid.

This is power supply telling motherboard that it is OK.

He's already proved that that doesnt happen, stupid.

Next monitor one of red, yellow, and orange wires. Each will rise
above 3.23, 4.87, or 11.7 VDC within seconds and remain there.

You've plucked those numbers out of your arse, as always.
Thats nothing like what the ATX spec says they must be.

Those voltages should remain constant (above minimal limits)
even as all peripherals are multitasked - accessed simultaneously.

Not a ****ing clue, as always.

If any voltage does not try to rise, then that voltage has too much
load (too many peripherals, short circuit, bad capacitor, etc).

Wrong, as always.

If any voltage drops too low as all peripherals are
accessed, then that one part of supply has a defect.

Wrong, as always.

Not only should this information identify a suspect in but minutes.

Pity he's already proven where the fault is, stupid.

Posting those numbers here can provide even more useful information.

But you are absolutely guaranteed to flaunt your complete pig
ignorance of what the ATX specs says about what they should be.

Currently your symptoms are those of an intermittent

Wrong, as always.

that could have been identified long ago.

Swapping power supply sometimes can even complicate problems.

Hardly ever.

Once those measurements identify the
suspect, only then do you replace that part.

Pity you cant even manage to grasp what the ATX spec says
about what voltages are acceptible, or anything else at all, either.

Then confirm that part solved the problems by taking voltage
measurements again. Solve a problem the first time; don't shotgun.

You wouldnt know what a shotgun was if one was fired at your lard arse.