Problem with PSUs

[Adrian Galperin]

Hi.

I bought one PC with the following components 6 months ago:

Motherboard: A7N8X-E Deluxe
Micro: AMD Barton 2800 MHZ
Memory: 2x512 MB DDR (Generic)
VGA Card: Chaintech GeForce MX 4000
Chipset NVIDIA GeForce4 MX 4000
Hard Disk: Seagate 80 GB (ST380013AS)
CD-RW LG
Modem Encore 56K

I installed it in one office. I connected the power supply to one
power stabilizer. Actually, I replaced an old computer that always
worked fine, with this new PC (with ASUS mainboard).
Next to this PC, there is another one connected directly (without
stabilizer) to the wall power. In another room there is another
computer working fine.
The problem is that since I started to work with this PC, the PSU
(power supply unit) of it got broken 3 times. Every time it broke, we
replaced it with a new one, usually different brand and models. All of
them were 350W standard PSUs which I think it is enough for giving
power to the few components my PC has.
Due to there are other computers working without any problem since
long time ago in the place where I installed the computer, I assume
that the problem is not related to the external electrical
infrastructure. Besides, I assume that it is neither related to the
PSU because 3 different PSUs broke. So, it seems to be some component
inside the PC is causing these failures.
Generally the problem happened this way: I was working with the PC and
suddenly the monitor lose the signal and the PSU stopped working. The
only thing still on was the ON/OFF led of the front of the cabinet. I
pushed the button until the led turned off. Then when I tried to turn
on the PC again, it did not work. Then I changed the PSU and generally
between 20-40 days later the same situation happened again.
What do you think about this? Could the problem be related to the
motherboard?

I will appreciate your comments and opinions.
Thanks.

Adrian

 

[Ghostrider]

Hi.

I bought one PC with the following components 6 months ago:

Motherboard: A7N8X-E Deluxe
Micro: AMD Barton 2800 MHZ
Memory: 2x512 MB DDR (Generic)
VGA Card: Chaintech GeForce MX 4000
Chipset NVIDIA GeForce4 MX 4000
Hard Disk: Seagate 80 GB (ST380013AS)
CD-RW LG
Modem Encore 56K

I installed it in one office. I connected the power supply to one
power stabilizer. Actually, I replaced an old computer that always
worked fine, with this new PC (with ASUS mainboard).
Next to this PC, there is another one connected directly (without
stabilizer) to the wall power. In another room there is another
computer working fine.
The problem is that since I started to work with this PC, the PSU
(power supply unit) of it got broken 3 times. Every time it broke, we
replaced it with a new one, usually different brand and models. All of
them were 350W standard PSUs which I think it is enough for giving
power to the few components my PC has.
Due to there are other computers working without any problem since
long time ago in the place where I installed the computer, I assume
that the problem is not related to the external electrical
infrastructure. Besides, I assume that it is neither related to the
PSU because 3 different PSUs broke. So, it seems to be some component
inside the PC is causing these failures.
Generally the problem happened this way: I was working with the PC and
suddenly the monitor lose the signal and the PSU stopped working. The
only thing still on was the ON/OFF led of the front of the cabinet. I
pushed the button until the led turned off. Then when I tried to turn
on the PC again, it did not work. Then I changed the PSU and generally
between 20-40 days later the same situation happened again.
What do you think about this? Could the problem be related to the
motherboard?

I will appreciate your comments and opinions.
Thanks.

Adrian

The only common points are: (a) Working inside the computer
and (b) The connection between the PSU to the power. If the
computer case is open and the computer fails, then one could
assume an errant burst of static or accidental shorting of a
circuit through carelessness. OTOH, surge protectors are not
very effective --- UPS's are better --- and with other systems
on the same power circuit, a burst from their PSU's could be
an issue, but not likely. Solution: Replace the PSU with at
least a 450-Watt Unit, and put all 3 computers on their own
UPS's, and stop fiddling with the insides of this particular
computer.

 

[w_tom]

One (of many reasons) for your symptoms would be a lockout
in the power supply controller. The power supply system is
more than just a power supply. This is only another reason
why a multimeter is used to find a defect before just
*assuming* power supply failure.

Numbers taken during some of these procedures make it
possible for the more technically informed to provide a useful
answer. As has always been well understood (and you have
demonstrated by example) is that swapping power supplies can
cure a symptom - not the problem. Procedures and how that
power supply system works are lists in two discussion:
"Computer doesnt start at all" in alt.comp.hardware on 10
Jan 2004 at
http://tinyurl.com/2t69q
"I think my power supply is dead" in alt.comp.hardware on 5
Feb 2004 at
http://www.tinyurl.com/2musa

Pictures to better understand where to take those
measurements:
http://techrepublic.com.com/5102-10586-5566528.html
www.ochardware.com/articles/psuvolt/psuvolt.html

Without numbers and other essential data, then replies can
only be speculation. Numbers empower those who can really
answer your question. Based upon what has been posted, one
can only wildly speculate.

What is this power stabilizer?

 

[Paul]

Hi.

I bought one PC with the following components 6 months ago:

Motherboard: A7N8X-E Deluxe
Micro: AMD Barton 2800 MHZ
Memory: 2x512 MB DDR (Generic)
VGA Card: Chaintech GeForce MX 4000
Chipset NVIDIA GeForce4 MX 4000
Hard Disk: Seagate 80 GB (ST380013AS)
CD-RW LG
Modem Encore 56K

I installed it in one office. I connected the power supply to one
power stabilizer. Actually, I replaced an old computer that always
worked fine, with this new PC (with ASUS mainboard).
Next to this PC, there is another one connected directly (without
stabilizer) to the wall power. In another room there is another
computer working fine.
The problem is that since I started to work with this PC, the PSU
(power supply unit) of it got broken 3 times. Every time it broke, we
replaced it with a new one, usually different brand and models. All of
them were 350W standard PSUs which I think it is enough for giving
power to the few components my PC has.
Due to there are other computers working without any problem since
long time ago in the place where I installed the computer, I assume
that the problem is not related to the external electrical
infrastructure. Besides, I assume that it is neither related to the
PSU because 3 different PSUs broke. So, it seems to be some component
inside the PC is causing these failures.
Generally the problem happened this way: I was working with the PC and
suddenly the monitor lose the signal and the PSU stopped working. The
only thing still on was the ON/OFF led of the front of the cabinet. I
pushed the button until the led turned off. Then when I tried to turn
on the PC again, it did not work. Then I changed the PSU and generally
between 20-40 days later the same situation happened again.
What do you think about this? Could the problem be related to the
motherboard?

I will appreciate your comments and opinions.
Thanks.

Adrian

If this was my computer, I would check the label on the side
of the PSU. With a 3200+ and dual channel memory, I've measured
16.6amps being drawn from +5V. Allow another 5.5amps for a high
end video card (one of the video cards with a separate power
cable). A disk drive uses 1A from +5V. That means 5V@25A is a
reasonable starting rating for the power supply. Not all 350W
standard power supplies will have that high a rating.

The thing is, different Asus motherboards use different rails
as their main source of power. The Pentium P4 boards place a
premium on the +12V current, while the AthlonXP Asus boards
like to use lots of +5V.

If you are a PC installer/builder/professional, I recommend owning
a clamp-on DC ammeter. There are some examples in this datasheet:

http://www.repaircalibration.com/lib/cal-tek/380947.pdf

To measure +5V, for example, you take the four +5V wires and
place all four within the jaws of the ammeter. The ammeter
will sum the current flowing through the four wires, and tell
you how much current is flowing to the motherboard. You can
do a similar measurement, using the +5V wire leading to a
disk drive or CD drive.

By using an application like Prime95 or CPUBurn, and a clamp-on
ammeter, you can measure system current consumption. Compare
the consumption to the rating on the side of the PSU. That will
tell you how much bigger a supply is needed. The label on the
side of the supply, gives a breakdown of what each PSU rail
is capable of. It will also say, what combined limits there are
for the +3.3 and +5V power.

It is also possible that something is shorting on the bottom
of the motherboard. Try to ensure the holes in the motherboard
are centered over the brass standoffs, to reduce the risk that
an off-center standoff is causing something to short.

The four +5V wires and pins on the ATX 20 pin connector, are
rated for 6 amps each. That is a total of 24 amps. If the
motherboard is drawing more than 24 amps (say due to a malfunction),
then the ATX 20 pin connector will get hot enough to melt the
plastic around the connector pins. If there is a defect bad
enough to do that, you might see visible damage to the ATX
power connector.

HTH,
Paul

 

[Adrian Galperin]

Dear folks,

Thanks for your replies.
I have checked with ASUS Support and they told me to download one ASUS
utility (ASUS PC Probe) for measuring some voltages. I found by using
this software the following values:

+12V 12.544
+5V 4.73
+3.3V 3.328
Vcore 1.664

As a result, ASUS support replied me:
"your +5V value is too low and the +12V value is too high; that means
your Power supply unit is not so good.
if possible ,i think you need test with another PSU, you can see the
every output value in the PSU label,
you need use a PSU with :
+5V higher than 28A,
+12V higher than 16A"

I checked my current PSU and it is:
Output 350W
+3.3V 22A
+5V 25A
+12V 12A

What do you think? Do you think the problem I am having (I already
damaged 3 PSUs) is because I need a better PSU which can achieve the
values recommended by ASUS Support.

I would like to know your comments about this.

Thanks.
Regards,

Adrian

 

[w_tom]

An onboard voltage monitor is only a monitor. It is not
sufficiently calibrated to measure voltage - only monitor for
voltage changes. But calibrated after confirmed with a
multimeter. What Asus told you (and more) was provided in
those previously cited posts.

Limits for those numbers were provided in the previous
discussion entitled "I think my power supply is dead". The +5
volts (if that really is the number) is low. Meets lower
limit in those specs. But the measured voltage must be above
4.87 for other reasons. The +12 is at and just under the
upper limit - acceptable. But again, until the multimeter
(with sufficient accuracy) reports those voltages, only then
can we be sure what your voltages are.

The difference what Asus demands and what the power supply
is rated at is quite trivial. That trivial difference alone
would not cause such voltage variances. But is that power
supply really sufficient? Does it really do what the label
says?

The procedure is simple. Measure those voltages on power up
as described in previous posts. Then repeat those
measurements when the system is accessing most peripherals
simultaneously. For example, soundcard is playing a sound
while small command line programs are repeatedly reading
directories of the C:, A:, and CD-Rom; and while doing
something via the network. This creates a maximum load.
Accessing peripherals simultaneously should not cause a major
voltage drop. Post numbers so that others with experience can
help you understand what those numbers mean.

How much does the voltage drop (either using multimeter or
motherboard monitor)? These numbers determine whether the
supply is too small (and therefore does not really output what
they claim) or whether problem is elsewhere.

BTW, computer cannot damage a minimally sufficient power
supply. Even one too small must never be damaged. But
swapping supplies may only cure symptoms - as you have
demonstrated previously. Were those previous supplies really
damaged? Or were they just protecting themselves? Better to
take numbers using the multimeter on both the suspect supply,
and then later after installing a new supply. Voltages at
startup and with supply under maximum load.

Power supplies that are bought on price (are missing
essential functions) can be damaged by a computer. If
computer damaged a supply, then other essential functions also
were probably forgotten inside that supply. Missing
functions that could then damage motherboard, disk drives,
etc. A minimally acceptable supply first provides a long
list of numerical specs. You may not understand what those
specs say. But if specs are provided, then 1% who really know
anything about computers can warn you about "all those missing
essential functions". If supply does not provide those
numbers, then essential functions are most likely forgotten -
not inside that supply.

A procedure to measure and understand power supplies was
posted previously. Anything Asus was saying (and other
important facts that Asus never mentioned) were in those
previous posts. Some clone power supplies don't really output
what is on the label. How do you know? Did it come with a
long list of numbers for other essential functions? If not,
suspect the worst.

 

[Paul]

Dear folks,

Thanks for your replies.
I have checked with ASUS Support and they told me to download one ASUS
utility (ASUS PC Probe) for measuring some voltages. I found by using
this software the following values:

+12V 12.544
+5V 4.73
+3.3V 3.328
Vcore 1.664

As a result, ASUS support replied me:
"your +5V value is too low and the +12V value is too high; that means
your Power supply unit is not so good.
if possible ,i think you need test with another PSU, you can see the
every output value in the PSU label,
you need use a PSU with :
+5V higher than 28A,
+12V higher than 16A"

I checked my current PSU and it is:
Output 350W
+3.3V 22A
+5V 25A
+12V 12A

What do you think? Do you think the problem I am having (I already
damaged 3 PSUs) is because I need a better PSU which can achieve the
values recommended by ASUS Support.

I would like to know your comments about this.

Thanks.
Regards,

Adrian

Well, I don't know if dwelling on the details of PSU design is
going to help you resolve this problem or not. Occasionally, I
run into reports here, where the symptoms suggest there is an
internal fault in a motherboard, but the only way to determine
if that is the case, is either by observing visual damage to
some component of the system, or to measure the current flow.

For example, I have seen a Vcore failure, where the area around
the CPU socket was charred from the heat. I have also read a report
of a power supply casing getting too hot to touch. Those are signs
that something bad is happening, and would be cases where the
current being consumed is way above either the ratings of the
power connector or of the power supply itself.

First, a little note about power supplies. I have a couple of
different kinds here. The simplest kind has one feedback wire on
a 3.3V signal. If you look at the ATX 20 pin wire harness, there is
a 3.3V pin, which has one thick wire, and one slightly thinner wire.
The thinner wire is a remote sense, and the power supply observes
the voltage measured at the end of the thinner wire. You will
notice that the value of the 3.3V output is very close to the
correct value, and the sense wire helps make that possible.

The older power supply designs had one primary circuit. One of the
points to note here, is that some of the output voltages are
determined by the transformer turns ratio, and are not that
precisely regulated.

http://www.pavouk.comp.cz/hw/en_atxps.html

Power supplies like the Antec Truepower have separated the outputs
somewhat, and there is a sense wire for +3.3, +5, and +12V.
This means the power supply outputs are separate stabilized.
So, not all power supplies have the same internal architecture.

Depending on the design, some power supplies have a limit on
the combined total power coming from +3.3V and +5V. This may
present a practical limit which is lower than the one listed
as the maximum current on a winding.

When I look at your voltages:

+12V 12.544
+5V 4.73
+3.3V 3.328

what they tell me is -

1) Your supply uses a single thin wire on a 3.3V power pin.
The power supply is tightly regulating the 3.3V voltage.
The +5V and +12V voltages are related to the 3.3V voltage
by means of the fixed turns ratio of the output transformer.

2) The +12V is higher than normal.
3) The +5V is lower than normal

4) Based on (2) and (3), what this tells me, is the +5V winding
is quite heavily loaded, which is to be expected for your
type of motherboard (Asus athlonxp motherboards power the
processor from +5V). Comparatively speaking, the +12V is
not being loaded nearly as much, so its voltage runs higher
than normal. I am more concerned with the 4.73 number, than
with the +12V going a little high.

The +5V reading of 4.73 is at the lower limit of the usual
5% tolerance listed on the label on the supply. That means
the number of amps flowing must be getting in the neighbourhood
of the capacity of the power supply.

So, what to do about it:

1) Buy a "boat anchor". For example, this power supply claims to
be able to provide +5V @ 61 amps, which is not physically
possible. It costs $64 US, and is not the most expensive
power supply you could buy. (It isn't really a 61 amp
supply, because the combined output limit on +3.3 and +5V is
235 watts, and even if the current flowing on the +3.3V output
is 0 amps, the +5V can make 47 amps.) This power supply will
teach your motherboard who is the boss

http://www2.newegg.com/Product/Product.asp?Item=17-101-406&depa=0
VOLTAGE +5V +12V +3.3V -5V -12V +5VSB
PSP4ATX50F 61A 28A 35A 0.5A 0.5A 2A

2) Buy a separately regulated supply. Any of these Antec
Truepower supplies will do the job. The True380 may be
enough current for +5V, to avoid problems. A True550 would
cost too much to be practical. There are some other brands
that also separately regulate the outputs, like OCZ 420ADJ
(http://www2.newegg.com/Product/Product.asp?Item=N82E16817104150)
where adjustment knobs are provided for each power rail. There
might even be an Enermax that does that.

VOLTAGE +5V +12V +3.3V -5V -12V +5VSB
TRUE330 30A 17A 28A 0.5A 1.0A 2.0A
TRUE380 35A 18A 28A 0.5A 1.0A 2.0A <--- $65 US
TRUE430 36A 20A 28A 0.5A 1.0A 2.0A
TRUE480 38A 22A 30A 1.5A 1.0A 2.0A
TRUE550 40A 24A 32A 0.5A 1.0A 2.0A <--- $95 US

420ADJ 30A 30A 28A 0.5A 0.5A 2.0A <--- $89 US

3) Buy an efficient supply. When your motherboard draws more power
than it should, a supply like this will not get nearly as
hot internally. Only problem with some of these, is they
are way too expensive. So, I think the "boat anchor" brute
force approach may be more economic than the elegance of
a high efficiency supply.

http://www.tomshardware.com/howto/20050228/power_supply-37.html

4) Get another motherboard ? Like maybe Abit NF7-S (I've heard
comments that the S2/S2G is not quite as nice, so read up
on which one is the best one to get). The NF7-S family run
the processor from +12V, at a lower current. The NF7-S has
likely been out of production for some time, leaving only
the S2 and S2G in the retail channel.

http://www.abit-usa.com/products/mb/

HTH,
Paul

 

[Adrian Galperin]

Hi.

The person who said that the onboard monitor is not accurate was
totally right.
I checked the voltages by opening the cabinet and using a digital
multimeter and found that the +5V value is about 4.95V with the PC
turned on but without working with it. This value dropped to 4.87V by
working doing several things at the same time (I started the virus
scan for my hard disk, opened a file from my CD-RW drive and asked my
scanner to scan one page).
As for the +12V value, I found that is about 12.66V in steady state
and increased to 12.92V by doing the above 3 things at the same time.
I have to say that everytime my PSU got damaged I was not making too
many tasks with the computer. Even once I was doing nothing. Just the
PC was turned on. Nothing else.
Somebody asked me if I see some damaged components or cables by
looking at it. I do not see anything wrong with the motherboard and
cables. When my PSUs failed in the past, nothing burnt, just the PSU
stopped working in one second.

What do you think? Should I measure anything else?

Adrian

 

[w_tom]

If damage can be seen, then you would have problems plus.
Basic electronic damage is not observable. Only catastrophic
(possibly life threatening and totally unacceptable) damage is
observable.

That 12 volts is too high. This can only be a power supply
problem. 12.9 will not cause hardware damage, but is
unacceptable from any minimally acceptable supply. A number
that suggests other possible problems inside that supply.

I suspect this is one of those supplies dumped into North
America because so many computer assemblers don't even have
basic electrical knowledge and yet claim they are experts. The
supply only exists because naive North American computer
assemblers make it so profitable to dump such supplies. Again,
"power supplies that are bought on price (are missing
essential functions) can be damaged by a computer."

A minimally acceptable supply can even have all output wires
shorted together and still must not be damaged. IOW the
problem is not so much the power supply as much as a problem
located elsewhere. Same problem that probably caused previous
power supply failures.

One simple way to know far more than most computer
assemblers. If the power supply does not provided with a long
list of numeric specs, then suspect the worst. Examples of
things a minimally acceptable supply will specifically claim,
in writing and with numbers:
Specification compliance: ATX 2.03 & ATX12V v1.1
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%

That means a minimally acceptable supply starts at $60
retail. Obviously, that does not mean all $60+ supplies are
acceptable. Price can only indicate a defective ('dumped')
power supply. Price cannot prove supply is minimally
acceptable.

In the meantime, your supply has indications of being a
'dumped' supply. Next time, first demand detailed and written
specs - or walk away from that vendor. Written specs being
the first of many requirements. You will discover most don't
want you to know these numbers. They are not marketing to
informed consumers. They are marketing to the ill informed
computer assembler. To dump inferior power supplies (that are
also missing other essential functions), manufacturer must not
claim anything; provide no written specifications.

Again, another fact about minimally acceptable supplies. A
load cannot harm such supplies. Either the minimally
acceptable supply works, or it shuts down without damage.
This being a fact that every minimally acceptable computer
assembler AND your power supply vendor are suppose to know -
since if was defacto standard even 30 years ago.

The meter makes problems immediately apparent. Furthermore,
the next power supply should be confirmed by meter when
installed so that this problem will be solved before it can
happen again. A meter being that essential to knowledge and
computer repair. The next power supply should have all three
key voltages within upper 3/4 of those limits even when
computer operates with all peripherals accessed
simultaneously. Supply voltages must remain acceptable even
under 100% load. A test that can only be performed with
supply fully connected to system and by using that meter.

The meter can also calibrate motherboard voltage monitor.
Set that monitor alarm at modified settings 'calibrated' by
the meter.

One other possibility about those previous power supplies.
They may not have failed. They simply shut down to protect
themselves. We would not know without further testing.

 

[Paul]

Hi.

The person who said that the onboard monitor is not accurate was
totally right.
I checked the voltages by opening the cabinet and using a digital
multimeter and found that the +5V value is about 4.95V with the PC
turned on but without working with it. This value dropped to 4.87V by
working doing several things at the same time (I started the virus
scan for my hard disk, opened a file from my CD-RW drive and asked my
scanner to scan one page).
As for the +12V value, I found that is about 12.66V in steady state
and increased to 12.92V by doing the above 3 things at the same time.
I have to say that everytime my PSU got damaged I was not making too
many tasks with the computer. Even once I was doing nothing. Just the
PC was turned on. Nothing else.
Somebody asked me if I see some damaged components or cables by
looking at it. I do not see anything wrong with the motherboard and
cables. When my PSUs failed in the past, nothing burnt, just the PSU
stopped working in one second.

What do you think? Should I measure anything else?

Adrian

If you want to see the +5V voltage take a big drop, get
a copy of Prime95 from mersenne.org. Run the "torture
test" option, then use the meter. Prime95 loads the
processor to 100%, and should give you a good indication
of how low the +5V will drop under worst case conditions.
If the +5V drops below 4.75V, you need more "amps".

I really think a more capable power supply is the answer.

Or, change to a motherboard that powers the processor from
the +12V rail, instead of +5V.

HTH,
Paul