(weird) Power Supply failure, reason?

[Average User]

My power supply died. (400watt I believe) but the symptoms were really unusual,
the 5v and 12v lines continued to function, when it died, the picture remained
on the monitor and the sound card kept working. (sounded like a CD skipping)
This tells me the cards still had power...

Fans kept working too.

Turned it off, then back on. Drives powered up, no screen, no BIOS beep, no
nothing.

I tried w/out the video card. Same thing. (also, I couldn't operate the caps
lock/num lock, it appears as if the CPU and NOT the power supply is DOA.

I measured it with a voltmeter, the 5v and 12v lines had power.

For all intents, it really looked as though the CPU itself was dead.

I also disected the original PSU, no bulged capacitors or any sign of board
damage.

After a bit of research, I replaced the power supply with a 350 watt unit.
(it's all the shop had, a Seaside 350 watt supply)

Everything powered up, system worked. 16 hours later it died, same exact
symptoms. I was there when it happened and managed to open the case within
about 3 minutes, none of the heat sinks were unusually "hot".

There weren't any power supply brown-outs or surges at the time. (the lights
didn't flicker or anything)

I managed to get another power supply and it *could* be, I suppose that I just
happened to get a bad replacement PSU. But I want to make sure..

What would cause a power supply to fail like that? 12v, 5v still worked. Fans,
drives, video card still worked. It was exactly as if the CPU just stopped.

I won't be able to get another PSU and I'd really rather not fry another..

What could cause a power supply to fail in such an unusual way? (where the 5v
and 12v power lines are still functional, the power light still lights up
and the green LED on the board is still on)

The motherboard is an:

Motherboard: ASUS M2V
CPU: AMD dual core 5200

There weren't any unusual devices plugged in. Not even USB devices.

Is there anything I should consider before installing PSU #3 ? (there are some
extreme financial constraints, I can't afford to be buying more power supplies)

Any advice appreciated! I'm really baffled by what caused the PSU to fail!

Thanks!

Jamie

 

[Average User]

But what voltages were read? Did you measure 3.3V, 5VSB,
and if it were a dual rail PSU, the 2nd 12V rail going to
the CPU as well as the 12V to HDDs/etc?

5v and 12v from the drive cables, I couldn't read the others because it needed
to be plugged in to power up.

Are you sure it was "Seaside" and not Seasonic? Seasonic is
a fairly good brand, while Seaside is an unknown (generic)

Your right! seasonic.

If this 350W PSU was a "Seaside", I would insist on my money
back from the shop, or to have the price paid credited to be
applied towards a better PSU they order for you (so long as
they don't try to price-gouge you too much on the one they
have to order, you can see fair market prices at someplace
like Newegg.com, and if they ordered it from Newegg it will
usually arrive within 2-3 business days so you wouldn't have
too much downtime.

That was the only power supply they offered, so I'm kind of stuck
with it. :-(

Still, wouldn't it just shut itself off if the load was too high?

By same exact symptoms do you mean the looping music and
screen freeze too, or just that system wouldn't come on
after (resetting or shutting down?) whatever happened but
the fans and drives worked but it failed to initialize and
POST - thus no screen output or beeps?

Yes, same exact symptoms except, no looping sound. (I didn't have
one playing when it went out)

It would appear as though everything but the CPU was powered on.

If the replacement is another generic, I wouldn't even try
it, instead getting a known good name brand. You didn't
mention all the (major power consuming) parts in your system
so it is hard to say if an honestly rated 350W PSU will
suffice or not.

It's a pretty vanilla machine:

AMD AM2 5200 (65 watt)
M2V motherboard
Integrated sound card.
Video Card (nvidia GeForce)
SATA Hard drive
Atansic network card.
DVD drive
CDRW drive

You write that they "still worked" but we can't be entirely
sure of that. They could get power, show some kind of
activity due to having power, but it might not be a stable
operation because of dirty power, and similarly, a
poor/failing/marginally-capable-for-system-load PSU has a
harder time initially powering on a system than keeping it
running.

Hmm.. if I removed the cables to the CDROM's, would that likely
give it the extra margin? I can probably do without one of the cdroms.

This could be an insight!

It could be some other part, not a PSU fault, but it is hard
to speculate with several variables. You might try
unplugging all other things non-essential towards getting
the system to initialize and POST, leaving only the
following:

CPU, CPU heatsink/fan, 1 memory module, the least power
hungry video card you have.

Hmm.. what is the probability of a bad motherboard, processor or
other thing shorting out the PSU? Have you ever known something to
do that?

At the least I feel the shop owes you your money back. Even
if the system had a short circuit of some part, it should
not kill the PSU. Since we don't k

I do appreciate your time in answering. I understand there isn't a guarantee

Just looking for areas and things I can do on this next power supply to
hopefully keep it from dying.

We can't say for certain from what you wrote that the 5V and
12V rails are functioning correctly even if they may have
power. PSU may turn on but have instable power, or take too
long to regulate properly so system can't initialize and
POST. Reducing parts in the system would tend to help PSU
regulate sooner by reducing current consumption but the
largest consumers, motherboard processor and video card have
to be present either way so pulling other parts is done in a
hope to show some sign of life but can't be a conclusive
proof of anything if it still don't show POST so you have
any feedback on screen.

I did pull everything, even the video card.. the NUM-LOCK and CAPS lock
keys wouldn't turn the lights on/off, this indicated to me the CPU
wasn't on.

You shouldn't need to. We can't know yet if original PSU is
dead or not, but you should receive a refund for the second
one. Unfortunately, a quality PSU may cost more than a
generic (assuming still that the generic was Seaside instead
of Seasonic), but you may find that buying online allows you
to pay nearly the same for a good brand as a generic costs
from a shop due to the shop middleman overhead.

I kind of wish I had. But.. this is the only PSU I've got to work
with.

Maybe some other part is failing instead of PSU, leaving all
parts non-essential towards posting and finally running
windows would be a way to rule out some parts if this is
possible.

I'm reasonably sure it'll power up if I put this PSU in it. I just
don't know how long it'll last.

If all else fails you are left substituting other
known-good or new parts for each item in the system, or
trying the suspect parts in another compatible known working
system. Playing odds, of the main components that are still
needed as enumerated above, CPU memory motherboard video,
the motherboard would tend to be a more likely fault,
perhaps video card if the fan on it died or was very clogged
with dust till it overheated. CPU and memory are usually
fairly reliable though memory might be the next most likely
then CPU last.

Another vote for motherboard as "most likely" in terms of
bad components.

I should have said, there is virtually no dust at all, the motherboard
is surprisingly clean. The video card doesn't have a fan.

What are the ratings on the 1st replacement PSU, the
Seaside? What are the ratings on the second PSU. From
these and the brand name and model we might be able to
speculate about appropriateness for powering the system
based on the parts list I asked about above.

The original could have been 550W or 400W. (the original PSU seemed to indicate
550 on the label, but I can't be certain, the invoice for this machine said
400W)

There are other less likely things that could cause such a
problem like an intermittent short. Something like having a
loose screw stuck behind the motherboard, having a
motherboard standoff installed where there shouldn't be one
which doesn't correspond to any screw hole in the
otherboard. Even bugs that crawl into a slot and die can
cause shorts, we can only proceed with one thing at a time
trying to reduce the variables through all the things
mentioned above.

Hahah!

I actually checked for insects, that was one of the things I had
suspected.

There weren't any loose screws and I did take the motherboard out, inspected it
for scratches and anything that might cause a short... nothing :-(

Also check the battery voltage, it wouldn't be impossible
for you to have two separate problems, one being instability
that just causes windows to crash, and the second being a
dead battery that keeps the system from POSTing.

I think the battery is good, as it was able to keep its settings

I *did* have a seemingly unrelated problem with a hard drive SATA
cable. But I replaced the cable within 2 hours of powering up on
PSU2 (the most recent one that went out)

This wasn't a power cable, it was just the data cable.

Measure
battery voltage, individual board tolerance to low voltage
might vary but I would replace it if lower than 3.0V... at
least in a case where I was trying to rule out things since
a battery is cheap, about $1 online maybe cheapest on ebay
shipped in a paper envelope.

I'll check the battery next.

And.. remove the USB plug on the front panel, remove the CDROM from
the board and power it up, does this seem like a reasonable test?

When this does power up, do you recommend I DISABLE everything related
to sleep mode in BIOS?

I did look and all the jumpers seem to be set to regular 5V instead
of the sleep mode.

By the way, there's a bright blue LED on the front. I had removed it
when I first got the machine (it's too bright) is there any harm in keeping
it removed? (I don't think there is, and it'd likely help more than hurt..
but I felt I should ask)

Thank you VERY much for responding!

Jamie

 

[Paul]

5v and 12v from the drive cables, I couldn't read the others because it needed
to be plugged in to power up.

You can probe the backside of the main power connector, to
monitor the other voltages. There is the nylon shell, with the
wires coming out, but the ends of the metal pins are accessible
to your multimeter probe. (Clip your black lead to an I/O screw,
so you're only poking at the main connector with the red test
lead.)

The pattern of voltages, can tell you where the loading is.
For example, if I saw 3.25V, 5.05V, 12.25V as the three main
rail voltages, I might suspect that the 3.3V rail has the
most loading, relatively speaking. The relationship of the
voltages is mainly determined by turns ratio (when all the
voltages share a common transformer), and if the voltages
seem to be relatively accurate with no load present, then
the loaded values can be used to hint at where the load might
be. This loading idea gains the most credence, when the
voltages are quite far off. So if you saw 3.1V on the 3.3V rail,
you'd suspect something was partially shorting or overloading
on the motherboard.

That technique is a poor substitute for having a DC clamp-on
ammeter, but not everyone can afford one of those.

If the load presented by the motherboard gets high enough, the
pins on the main power connector can get burned. 6 amps per pin
is a rough estimate of the safe current carrying capacity. Above
that level, you might see eventual oxidation and failure to make
good contact. Oxidation is a runaway condition, so once a pin
is ruined, it only gets worse (resistance gets higher, temp
gets higher, pin oxidizes some more, soon there is no shiny
metal left). Cleaning up a pin, with emery paper or equivalent,
would only be a temporary solution. (I've tried it, on some lab
equipment I needed to stay running, and did experiments over
a two year period. I eventually soldered the power supply
right to the equipment, so the connector could no longer fail
Problem solved.)

Paul

 

[Average User]

Maybe you are that one person out of thousands who after
having their PSU die, unluckily ended up with a replacement
that had a defect, but I would not buy another PSU without a
warranty if your (Seasonic) actually doesn't have one, and
would be further examining the parts and trying to find
faults in motherboard, video card, and less likely memory or
CPU. Surely the shop can do something to accomdate your
needs, it shouldn't be much time for them to just plug your

I should clarify, they DID. I have now the second Seasonic PSU
and I want to make sure it'll work. (they wouldn't refund the $$,
but they did exchange it)

I can't exchage it AGAIN.. this is what I mean by "last chance".
(I've not yet powered it.. trying to collect data first)


Anyway, I plugged it in and fired it up. This time, it didn't do what
it did last time.. :-/

It did NOT boot, no POST, no BEEP, no nothing.

I pulled the RAM and attempted again.

No POST, no error about no ram or anything on the monitor, BUT it
did beep.

Is the CPU responsible for checking the presence of RAM or is that
some other circuit?

(pulling the ram resulting in a beep provided some insights?)

Thanks!

(I've NEVER had a problem like this before, this is just plain WEIRD!)

Jamie

 

[Paul]

I should clarify, they DID. I have now the second Seasonic PSU
and I want to make sure it'll work. (they wouldn't refund the $$,
but they did exchange it)

I can't exchage it AGAIN.. this is what I mean by "last chance".
(I've not yet powered it.. trying to collect data first)


Anyway, I plugged it in and fired it up. This time, it didn't do what
it did last time.. :-/

It did NOT boot, no POST, no BEEP, no nothing.

I pulled the RAM and attempted again.

No POST, no error about no ram or anything on the monitor, BUT it
did beep.

Is the CPU responsible for checking the presence of RAM or is that
some other circuit?

(pulling the ram resulting in a beep provided some insights?)

Thanks!

(I've NEVER had a problem like this before, this is just plain WEIRD!)

Jamie

The CPU checks for RAM. Part of the BIOS code runs without the benefit
of RAM, the CPU reads the SPD chip on the DIMM, and then attempts to
set up the RAM controller in the Northbridge.

If it beeps a "missing RAM" code, when the RAM is removed, then you
know the CPU has executed some code. If the computer no longer
beeps when the RAM is reinstalled, then the RAM is dragging down the
Northbridge, and preventing the CPU from getting to the BIOS chip,
to read instructions.

If those are the symptoms, do you have another stick of RAM to test ?
If the second stick does the same thing, it could be motherboard
(or a short to the bottom of the motherboard, like an extra standoff
where there should not be one). If it is a standoff, pressing on the
motherboard, might make the problem intermittent.

Paul