Something is killing my power supplies

[Graham D]

I wonder if anyone can give me some tips to help trace a problem I am
having.

I recently upgraded my self-built system by changing the motherboard,
memory and video card. I had previously upgraded the hard drive and
various other bits and pieces.

At first everything worked fine, but after a couple of weeks the
entire system died. Checks revealed that the PSU had died. I carefully
checked everything in the system for short circuits, bad connections,
etc, etc, but all was fine. The computer shop I used suggested that
the new components had possibly placed too much demand on an old and
low-powered PSU.

I then bought a good quality 400W PSU, checked everything out, and it
all worked without any problems. I reassembled the system, and it
still worked without any problems.

After about two months, I left the computer running overnight going
through various utilities. By morning... dead again.

I am baffled by this. If it was a short circuit, the PSU would have
died immediately (wouldn't it???).

I don't think the problem lies with the HDD, CD-ROM, modem, etc, as
these all worked fine with the old motherboard for a long time.

So the culprit must be (I assume):

- the motherboard
- the PSU
- the new video card
- the memory

Does anyone have any ideas about what sort of problem would cause a
power supply to blow after two months of trouble free operation? I
don't want this problem to recur, and I don't particularly want to
have to replace all the components, either.

Any suggestions either as to the cause of the problem, or as to how I
might try and track it down would be appreciated.

Many thanks.

Graham D

 

[John McGaw]

I wonder if anyone can give me some tips to help trace a problem I am
having.

I recently upgraded my self-built system by changing the motherboard,
memory and video card. I had previously upgraded the hard drive and
various other bits and pieces.

At first everything worked fine, but after a couple of weeks the
entire system died. Checks revealed that the PSU had died. I carefully
checked everything in the system for short circuits, bad connections,
etc, etc, but all was fine. The computer shop I used suggested that
the new components had possibly placed too much demand on an old and
low-powered PSU.

I then bought a good quality 400W PSU, checked everything out, and it
all worked without any problems. I reassembled the system, and it
still worked without any problems.

After about two months, I left the computer running overnight going
through various utilities. By morning... dead again.

I am baffled by this. If it was a short circuit, the PSU would have
died immediately (wouldn't it???).

I don't think the problem lies with the HDD, CD-ROM, modem, etc, as
these all worked fine with the old motherboard for a long time.

So the culprit must be (I assume):

- the motherboard
- the PSU
- the new video card
- the memory

Does anyone have any ideas about what sort of problem would cause a
power supply to blow after two months of trouble free operation? I
don't want this problem to recur, and I don't particularly want to
have to replace all the components, either.

Any suggestions either as to the cause of the problem, or as to how I
might try and track it down would be appreciated.

Many thanks.

Graham D

What is the "good quality PSU" that you replaced the original with? For that
matter, what was the original? My first instinct is that the replacement
simply died a natural death in that it ran for "about two months" but I
guess there could be other reasons. For example you say that you "left the
computer running overnight" explaining when the problem occurred but my
question would be "is this the only time that it has run overnight?". I
suppose that there could be some hideous things going on with your mains
power that is striking only when the system is switched on in the early
morning hours although I'd expect other electrical equipment to show
problems if this were the case.

Typically if some component in the system was malfunctioning badly enough to
destroy a PSU there would be telltale signs like exploding obviously flamed
out components and nasty smells left behind so I seriously doubt that some
component like your memory is at fault. A possibility is that the mounting
points on your MB could be mislocated and be shorting out under odd
circumstances but this seems less likely because such a short is usually
pretty obvious since it can commonly be provoked by simply pressing on and
flexing the MB while the system is powered up.

In your situation I'd simply buy a new power supply from a top-line
manufacturer and go on as if nothing had happened. There are some relevant
PSU reviews at http://www6.tomshardware.com/howto/20040122/index.html .
Their tests reveal that there is a vast span of quality and rating honesty
among the main manufacturers.

 

[Dave C.]

OK, one old power supply died (common), and then one newish one died. (not
so common, but not unheard of)

I think you're just having bad luck. I don't think there is anything about
your hardware that is killing power supplies. Try a Seasonic or Fortron
brand power supply next. If THAT dies in a few months, I might change my
mind. But right now, I think two power supplies dead in a few months is
just bad luck. -Dave

 

[AlanW]

Have you considered voltage spikes from your house supply? Try one of
those spike protector gizmos.
Cheers
Al

 

[w_tom]

If a computer parts supplier tells you:

... new components had possibly placed too much demand on an old
and low-powered PSU.

Then go find a computer parts supplier that first learned
basic electrical knowledge. He lied. He should know its a
lie since even Intel specs make PSU requirements quite clear.
A power supply must either supply sufficient power or it must
shut down without damage. Power supply outputs can even be
shorted together and still no damage occurs to that supply.
Defacto standard 30 years ago. But then some computer parts
suppliers buy supplies dumped into America without essential
internal functions. Suppliers sell them only one
specification - price. Could not be bothered to verify or
even learn numerical and technical specifications.

Power supply supply fails. Supplier blames customer, too
much load, new motherboard, mythical spikes from AC mains, ...
anything except the technically ignorant parts supplier who
should have never been selling that supply.

The culprit is most likely the power supply vendor. He even
says:

new components had possibly placed too much demand on
an old and low-powered PSU.

Appreciate how technically naive (and now richer) he really
is.

What happens to many power supplies that cannot even bother
to meet standards of 30 years ago? Power supplies self
destructed before output maximum rated load could be
measured. That means supply failed multiple requirements - it
was that bad:
http://www6.tomshardware.com/howto/02q4/021021/index.html

Time and time again, our lab measurements were unable to
verify the output figures represented on the model
identification sticker.

Lets review some industry standards for power supplies.
Power supply outputs can even be shorted together and the
power supply will not be damaged. Power supplies already
contain any protection that can be effective at the power
supply - making that plug-in protector redundant as well as
ineffective. Even if the power supply suffers a massive
failure, still, it must not damage any components in
computer. No computer components can damage a power supply
sold by responsible vendors. If the power supply is a
problem, then a 3.5 digit multimeter will quickly identify
that failure - much faster and more accurately than even
swapping power supplies.

 

[DaveW]

Many so-called "400 watt" power supplies that are not high quality brands
put out no where near 400 watts into real world loads. You didn't describe
what components you have hooked up to the power supply in your computer,
but its possible that you are still straining the power supply and thus
burning it out. And as a recommendation I would highly suggest an Antec
power supply.

 

[R. Asby Dragon]

Have you considered voltage spikes from your house supply? Try one of
those spike protector gizmos.
Cheers
Al

Screw that idea; most of them are only good for "one hit" .

Go with a good UPS (Uninteruptable Power Supply)

I don't put a machine online without it in this house . I work for a
power utiliity--- not the one who supplies my 'juice' at home
unfortunately ....

I've got 4 of them; adding a couple more. It's cheap for me; I'm a
tech and know where to buy good batteries cheap.

 

[Guest]

I recently upgraded my self-built system by changing the motherboard,
memory and video card. I had previously upgraded the hard drive and
various other bits and pieces.

At first everything worked fine, but after a couple of weeks the
entire system died. Checks revealed that the PSU had died.
The computer shop I used suggested that the new components had
possibly placed too much demand on an old and low-powered PSU.

Maybe, but the power supply should have simply shut down, without
damage to itself or the equipment. All ATX supplies are built with
overpower and overcurrent protection, but it may not have been built
correctly. Also a high percentage of supplies, probably the majority,
have no overtemperature protection.

I then bought a good quality 400W PSU,
After about two months, I left the computer running overnight going
through various utilities. By morning... dead again.

What brand, and why did you think it was good? If the same shop sold
it to you, they owe you a new supply.

http://takaman.co.jp can help estimate the power load, but that
spreadsheet gives high estimates. Since you didn't list any specifics
about your equipment, there's no way to know whether it drew 50W or
250W, but don't assume that a supply rated for 300W will be adequate
for a 250W load because ratings are often inaccurate, and the current
draw from one of the voltage lines may have been more than what the
supply could have provided.

 

[Patrick]

one of



Screw that idea; most of them are only good for "one hit" .

Go with a good UPS (Uninteruptable Power Supply)

I don't put a machine online without it in this house . I work for a
power utiliity--- not the one who supplies my 'juice' at home
unfortunately ....

I've got 4 of them; adding a couple more. It's cheap for me; I'm a
tech and know where to buy good batteries cheap.

In Florida, I run 9 UPS boxen, 7 for computer systems, and one each for
the two dvd's, 2 TV's, 2 video game consoles...

Our power spikes in the AM and PM, from the switch-over of the mains to
get more feeds for the load, (plus, no upgrades of the 'plant' since
1974), just repairs. Our power runs in a somewhat serial direction,
North to South, with little of the feed structure of the rest of America.

I got damn tired of blown PSU's!

 

[Andy]

Our power spikes in the AM and PM, from the switch-over of the mains to
get more feeds for the load, (plus, no upgrades of the 'plant' since
1974), just repairs. Our power runs in a somewhat serial direction,
North to South, with little of the feed structure of the rest of America.

I got damn tired of blown PSU's!

I thought only California was that bad in the USA ? Years in the UK since I
even experienced a blackout and I have never had a blown PSU in 20 years
(plenty of failed fans as normal)

Is there anywhere is the States with good reliable Power ?

Andy

 

[Moderately Confused]

I thought only California was that bad in the USA ? Years in the UK since I
even experienced a blackout and I have never had a blown PSU in 20 years
(plenty of failed fans as normal)

Is there anywhere is the States with good reliable Power ?

Andy

Here in Pennsylvania on the east cost it's pretty good. In general, the
whole is east cost is pretty good electricity wise, as long as you overlook
the whole New York thing that happened a little bit ago...

MC

 

[Graham D]

Thank you for the prompt and helpful responses to my query. Back to
the supplier it is...

Graham D

 

[w_tom]

Any power supply that is minimally acceptable will provide a
long list of numerical specifications. Power supplies that
sell on cost - that forget to include essential functions -
also will not provide specs. Why get caught with a written
lie? A short example of specs that any decent power supply
manufacturer will provide (and there should be this many specs
plus more):
Specification compliance: ATX 2.03 & ATX12V v1.1
Acoustics noise 25.8dBA typical at 70w, 30cm
Short circuit protection on all outputs
Over voltage protection
Over power protection
100% hi-pot test
100% burn in, high temperature cycled on/off
PFC harmonics compliance: EN61000-3-2 + A1 + A2
EMI/RFI compliance: CE, CISPR22 & FCC part 15 class B
Safety compliance: VDE, TUV, D, N, S, Fi, UL, C-UL & CB
Hold up time, full load: 16ms. typical
Efficiency; 100-120VAC and full range: >65%
Dielectric withstand, input to frame/ground: 1800VAC, 1sec.
Dielectric withstand, input to output: 1800VAC, 1sec.
Ripple/noise: 1%
MTBF, full load @ 25°C amb.: >100k hrs

Demonstrated here is why power supplies fail. They don't
fail due to too much load. Load only kills supplies that
don't even claim to provide these specification - a failure
directly traceable to the human. BTW, many if not most clone
computer power supplies (including one seen in CompUSA) don't
even claim to meet these specs. With so many 'expert'
computer assemblers in America that don't even have basic
electrical knowledge (who make decisions only on price), then
N America has become a fertile market for dumping inferior
supplies. That is more often why power supplies fail. The
'bean counter' mentality who calls himself computer literate.

A power supply damaged by load is a failure directly
traceable to that technically naive 'bean counter' human who
bought the supply. Minimally acceptable power supplies are
not damaged by load.

 

[w_tom]

Any protectors that is good for one hit did not even protect
during that hit. Grossly undersized protectors sell to so
many technically ignorant by undersizing the protector. Power
supplies (must) have internal protection. Protection so good
that a small spike cannot damage a computer supply BUT will
damage the grossly undersized protector. Then the naive say
the protector sacrificed itself to protect from a destructive
surge. Myths promoted.

The effective protector costs about $1 per protected
appliance. That plug-in protector or the UPS (which has the
same protector circuit as a power strip) cost what - $15, $50,
or $70 per protected appliance. That is tens of times more
money for something that does not even claim to protect
protection from the type of surge that typically damages.
Read their numerical spec. They don't even discuss the
different type of surges nor really make any numerical claim
of protection. They only say enough for the technically naive
to *assume* protection from all types of surges exists.

Need effective protection? The primary protector is
provided by the utility. But you may need to inspect it:
http://www.tvtower.com/fpl.html
And rules for earthing:
http://www.tvtower.com/grounding_and_bonding.html

The secondary protector is installed by the homeowner
because we still don't build new homes as if the transistor
exists. Its called a 'whole house' protector and is also
defined by its connection to a single point earth ground:
"RJ-11 line protection?" on 31 Dec 2003 in pdx.computing at
http://tinyurl.com/2hl53 and
"Opinions on Surge Protectors?" on 7 Jul 2003 in the
newsgroup alt.certification.a-plus at
http://tinyurl.com/l3m9

Any protector that is minimally effective remains fully
functional after the direct lightning strike. But most
important is what a protector does. Those plug-in protectors
and UPSes do not even provide a numerical specification that
says they provide effective protection. Look at their specs.
But one need not even read to see why they are not effective.
Where is the dedicated 'less than 10 foot' connection to
central earth ground? No earth ground means no effective
protection. Earthing is why 'whole house' protectors have
been so effective since before WWII. The protector is only as
effective as its earth ground.