PSU watts reading

  • Thread starter Thread starter Frank
  • Start date Start date
F

Frank

Hi all,
Anyone knows if it's possible to monitor or take a reading of how much watts
the psu is giving out. I have a 400W psu and I would like to know if I'm
running it at full capacity or if it's just giving out 50% of its capacity.
Reason is the fan pulls out air which is 2x hotter that my case fan which I
would suspect the psu is in demande. But how much? Volts I can figure out
with a meter, but watts? How do I take a reading. anybody know?
Thanks
 
When supply starts to exceed wattage, then voltage starts to
drop. If volts drop to lowest 1/4 of limits or lower, then
power supply is probably maxed out. Setup XP so that it
accesses multiple peripherals simultaneously. Then observe
voltage. No significant voltage drop (do this for each of
three critical voltages on red, orange, and yellow wires)
means power supply has sufficient power.

But then most every computer will work just fine on a 250
watt supply. Your worry: many clone supplies do not even
provide power that they rated at. A problem created by too
many domestic computer experts who don't even know basic
electrical principles nor essential functions that a power
supply should provide. So many clone system supplies don't
even contain essential functions that were defacto standard 30
years ago. Measure the voltages with supply under full load
to answer you concern.
 
w_tom said:
When supply starts to exceed wattage, then voltage starts to
drop. If volts drop to lowest 1/4 of limits or lower, then
power supply is probably maxed out. Setup XP so that it
accesses multiple peripherals simultaneously.
Click to expand...

Why go to all the trouble of installing XP just to do that?
Then observe
voltage. No significant voltage drop (do this for each of
three critical voltages on red, orange, and yellow wires)
means power supply has sufficient power.

But then most every computer will work just fine on a 250
watt supply.
Click to expand...

<cough>bullshit<cough>

My CPU draws pretty much half that, my GPU pretty much the other half and
the mobo the rest. What are my drives supposed to run on?
Your worry: many clone supplies do not even
provide power that they rated at. A problem created by too
many domestic computer experts who don't even know basic
electrical principles nor essential functions that a power
supply should provide.
Click to expand...

Pure arrogance, a typical (of late) w_tom post. A man who assumes that if
you don't owm a DVM or multimeter you shouldn't own a PC.
 
Frank said:
Anyone knows if it's possible to monitor or take a reading of how much watts
the psu is giving out. I have a 400W psu and I would like to know if I'm
running it at full capacity or if it's just giving out 50% of its capacity.
Reason is the fan pulls out air which is 2x hotter that my case fan which I
would suspect the psu is in demande. But how much? Volts I can figure out
with a meter, but watts? How do I take a reading. anybody know?
Click to expand...

Easiest way is to measure the input power and divide that by 0.7
 
ric said:
Easiest way is to measure the input power and divide that by 0.7
Click to expand...

This won't take into account losses in the PSU it self, but would be a good
quide.
He then has the problem of measuring the input power, which I doubt will be
and easier. It will certainly be far more dangerous.
 
Frank said:
Hi all,
Anyone knows if it's possible to monitor or take a reading of how much watts
the psu is giving out. I have a 400W psu and I would like to know if I'm
running it at full capacity or if it's just giving out 50% of its capacity.
Reason is the fan pulls out air which is 2x hotter that my case fan which I
would suspect the psu is in demande. But how much? Volts I can figure out
with a meter, but watts? How do I take a reading. anybody know?
Thanks
Click to expand...

Well power=currentXvoltage

W(power)=I(current)V(voltage) or = V(squared)/R or = I(squared)R from (V= I
times R.)

So really you need an ampmeter.

Pass the current through the ampmeter(or ammeter) and multiply by
the voltage which you should measure at the same time. You could
*assume* the voltage is as given (say 5volts).

You can take reading at seperate times if you want though, as
the reading is unlikely to change and the meter itself, by design,
will have a negligible effect.

Of course there may be several voltage level (5V, 12V etc..)
in which case you need to add them (powers) all togeather.

Presumably your meter does not have a facility to read amps?
In which case you should have stole an ammeter (called AVO's in my day)
from the school physics lessons which you apparently didn't attend!!

Anyway if you have a mulit-meter like this one:-
http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&category=4678&item=3814673619&rd=1

It should so the trick.
I assume your meter will also measure amps as the one I found is
only £4.99.
I am seriously tempted to get one myself at that price!!!!!
Especially as someone here has been nagging me to get a meter.

I want to make a similar measurement on my PC so I am glad you asked!!

I think I will check a few local stores such as Maplin (UK) to see if they
have something at a similar price before I ebay though.

It looks as 'neat little baby' at £7 inc p&p not much more than a packet
of ciggarettes!

It seems good value too and it is within 1% accurate so it will be good
enough
for any job I want to do with it!!

Good luck.

half_pint.
 
half_pint said:
which I



This won't take into account losses in the PSU it self, but would be a good
quide.
He then has the problem of measuring the input power, which I doubt will be
and easier. It will certainly be far more dangerous.
Click to expand...
The value of 0.7 is the efficiency of the PSU, which is correct for
most. To measure the input power, just use a line cord with a 1 ohm
resistor in series and measure the voltage drop across it. As for
danger, judging from your posts, anything you do is dangerous!

Virg Wall
 
half_pint said:
which

Well power=currentXvoltage

W(power)=I(current)V(voltage) or = V(squared)/R or = I(squared)R from (V= I
times R.)

So really you need an ampmeter.

Pass the current through the ampmeter(or ammeter) and multiply by
the voltage which you should measure at the same time. You could
*assume* the voltage is as given (say 5volts).

You can take reading at seperate times if you want though, as
the reading is unlikely to change and the meter itself, by design,
will have a negligible effect.

Of course there may be several voltage level (5V, 12V etc..)
in which case you need to add them (powers) all togeather.

Presumably your meter does not have a facility to read amps?
In which case you should have stole an ammeter (called AVO's in my day)
from the school physics lessons which you apparently didn't attend!!

Anyway if you have a mulit-meter like this one:-
http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&category=4678&item=3814673619&rd=1

It should so the trick.
I assume your meter will also measure amps as the one I found is
only £4.99.
I am seriously tempted to get one myself at that price!!!!!
Especially as someone here has been nagging me to get a meter.

I want to make a similar measurement on my PC so I am glad you asked!!

I think I will check a few local stores such as Maplin (UK) to see if they
have something at a similar price before I ebay though.

It looks as 'neat little baby' at £7 inc p&p not much more than a packet
of ciggarettes!

It seems good value too and it is within 1% accurate so it will be good
enough
for any job I want to do with it!!

Good luck.

half_pint.
Click to expand...

Having said that I don't think that meters DC current range will be enough.
At 400W and 5V the current would be a huge 80 Amps!!!
So you will probably need a different type of meter, something like this
one.
http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&category=1504&item=4166950115&rd=1
or
http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&category=4678&item=3815122807&rd=1

There are probably also some 'tricks' you could do using various resistors,
but I
am too tired to go into it now.

Some links :-

http://www.otherpower.com/otherpower_experiments_ammeter.html


AND!!!

http://www.otherpower.com/cgi-bin/webbbs/webbbs_config.pl?noframes;read=5232

That 'TomW' name in the above link seems disconcertingly familiar!!!

I have the feeling I have have bitten off quiet a large piece in this
theads - gulp :O)
 
VWWall said:
The value of 0.7 is the efficiency of the PSU, which is correct for
most. To measure the input power, just use a line cord with a 1 ohm
resistor in series and measure the voltage drop across it. As for
danger, judging from your posts, anything you do is dangerous!

Virg Wall
Click to expand...

Yes well that maybe so!!! (see my follow up posts) it did seem to easy.

I am guessing a 'line cord' is a the 1 ohm resistor in series?
I though of something like that (on the DC side though? yes?).

I didn't realise the currents were so large.

I am a little surprised large current DC meters are hard to come by,
I reckon I could make one *easy*. (Provided I survived the experience).

I kind of realised I had 'put my foot in it' but I will come back later when
I am
less tired.

If his PSU is faulty can you rely on a 0.7 value?
 
half_pint said:
This won't take into account losses in the PSU it self, but would be a good
quide.
Click to expand...

Sure it will. (I meant multiply, BTW. You can divide by 1.414.)

Losses inside the PSU *are* the reason for the multiplication by 0.7,
as in 70% efficiency. Input power times efficiency = output power.
He then has the problem of measuring the input power, which I doubt will be
and easier. It will certainly be far more dangerous.
Click to expand...

It won't be easier to measure the power on ONE input line versus the dozen
or so lines coming OUT of the PSU? A power meter is easily rentable, or
simply measure the current in one AC line and do the math.
 
ric said:
Sure it will. (I meant multiply, BTW. You can divide by 1.414.)

Losses inside the PSU *are* the reason for the multiplication by 0.7,
as in 70% efficiency. Input power times efficiency = output power.


It won't be easier to measure the power on ONE input line versus the dozen
or so lines coming OUT of the PSU? A power meter is easily rentable, or
simply measure the current in one AC line and do the math.
Click to expand...


Well I don't know how that 0.7 figure is arrived at.
Anyway if the PSU is faulty I doubt it would be accurate.

With a proper ammeter which can take large currents it would not take
long to take the reading, and your results would not be open to question.
They would basically be 100% reliable.
 
half_pint said:
Well I don't know how that 0.7 figure is arrived at.
Click to expand...

You don't? Most PSU manufacturers specify their PSU efficiency. Most
are around 70%.
Anyway if the PSU is faulty I doubt it would be accurate.
Click to expand...

The OP made it quite clear that his PSU was not faulty. He just wondered
how much power it was putting out.
With a proper ammeter which can take large currents it would not take
long to take the reading, and your results would not be open to question.
Click to expand...

So, you are gonna measure all three 3.3v lines, all four +5v lines,
all of the +/- 12v, -5v, +5vsb lines, etc.? And monitoring ONE input
line wouldn't be easier?
 
I was planning on measuring the total watts used (3.3V,5V,12V) or maybe each
rail individually. However, I've studied slightly in electronics and one
thing I hate is blowing something up. No matter how careful one can be there
is always a risk and I can't afford it right now. So I trying to find the
most failsafe method out there. I went over to my local radioshack but the
guy at that store doesn't know too much. I heard of a clamp on amp meter but
it works for AC. Don't know if it could work with DC. I doubt. As for
cutting wires or pluging my multimeter in there is a no no.to risky and not
convienient to take measures while windows is booting. Amps being mesured in
series, I would have to hold everything by hand and shutdown before removing
the meter. Too complicated and risky. Anyways, thanks for all your help
guys, and if more ideas do come up, I'm open to suggestions.
thanks all
 
ric said:
You don't?
Click to expand...

Well I suppose I do actually, they would measure the input
current and voltage product appropiately, and also the sums of
the products of the output current and voltages in the method I
described, then divide the latter by the former. Doing anything
else would be mere guesswork.
Most PSU manufacturers specify their PSU efficiency. Most
are around 70%.
Click to expand...

And they have to do the above to determine that figure.
The figure they give may well be an average figure anyway,
just like the number of matches in a box.
The OP made it quite clear that his PSU was not faulty.
Click to expand...

No he doesn't, you are making an assumption. It is clear he cannot
determine the output power so he cannot determine it is working
efficiently to the manufacturers rating. What he says in part of his
post is a little unclear to me "I would suspect the psu is in demande"
He just wondered
how much power it was putting out.
question.

So, you are gonna measure all three 3.3v lines, all four +5v lines,
all of the +/- 12v, -5v, +5vsb lines, etc.? And monitoring ONE input
line wouldn't be easier?
Click to expand...

Well it might take a while but its the only way to get an accurate
reading. he indicates he may have a power supply problem?

I was under the impression that that there would be a maximum.
of 3 power levels, however I seem to remember a whole bunch of
wires come out of the PSU box and if that is the case it probably is not
going to be practical to do it unless you are very patient and diligent.

Perhaps if I had designed the PSU the job would be easier :O)
And perhaps a little more dangerous?

Also some people (me included) are not to happy playing
around with 240V mains.
 
I was planning on measuring the total watts used (3.3V,5V,12V) or maybe each
rail individually. However, I've studied slightly in electronics and one
thing I hate is blowing something up. No matter how careful one can be there
is always a risk and I can't afford it right now. So I trying to find the
most failsafe method out there. I went over to my local radioshack but the
guy at that store doesn't know too much. I heard of a clamp on amp meter but
it works for AC. Don't know if it could work with DC. I doubt. As for
cutting wires or pluging my multimeter in there is a no no.to risky and not
convienient to take measures while windows is booting. Amps being mesured in
series, I would have to hold everything by hand and shutdown before removing
the meter. Too complicated and risky. Anyways, thanks for all your help
guys, and if more ideas do come up, I'm open to suggestions.
thanks all
Click to expand...

Just a few random comments, not necessarily applicable to this point in
the thread but nonetheless together in a single post;

- A reading cannot be made and assumed as a constant. For example, hard
drives spinning up will typically consume upwards of 2.25A but later under
1A. Sitting at the desktop the video card isn't under much stress but
play a 3D game and potentially large increase in 3V. CPU will also change
current consumption by a large amount on 12V or 5V rail, particularly
when: A) OS uses ACPI w/HLT Idling B) Motherboard bios has this
feature enabled (when applicable, in general it's much less likley to be
enabled on an Athlon XP than a P4 chipset-m'board).

- If you can achieve what you consider a valid amperage reading, then what
do you compare it to? One could trust the specs on manufacturer's data
sheet or printed on the label but additional specs must be considered,
such as the MTBF rate and temp used to obtain that rating, and the
reputation of the manufacturer to provide accurate ratings or at least
disclose when their rating system deviates from most common practices.
Also with figures like MTBF it can even be necessary to "attempt" to
discriminate why a unit has a particular rating. For example, the rate
may be based on the MTBF rating of the fan alone, it being considered the
shorter-lived part.

Without measurement we can't know for certain exactly how much power your
system is using, but if you provided a list of the major components and
the make/model/capacity of power supply we can most likely tell you
whether the unit is nearer one or the other two alternatives you
mentioned, half-load or fully loaded. It's pretty hard to build a
"typical" PC that uses 400W though, generally it'd need highly o'c (or
multiples of) CPU, Video, and quite a few HDDs and/or peltiers to come
close to 400W consumption.
 
VWWall said:
The value of 0.7 is the efficiency of the PSU, which is correct for
most. To measure the input power, just use a line cord with a 1 ohm
resistor in series and measure the voltage drop across it. As for
danger, judging from your posts, anything you do is dangerous!
Click to expand...

I did think the 0.7 was some adjustment for root-mean-squared,
obviously not!
 
Frank said:
I was planning on measuring the total watts used (3.3V,5V,12V) or maybe each
rail individually. However, I've studied slightly in electronics and one
thing I hate is blowing something up. No matter how careful one can be there
is always a risk and I can't afford it right now. So I trying to find the
most failsafe method out there. I went over to my local radioshack but the
guy at that store doesn't know too much. I heard of a clamp on amp meter but
it works for AC. Don't know if it could work with DC. I doubt. As for
cutting wires or pluging my multimeter in there is a no no.to risky and not
convienient to take measures while windows is booting. Amps being mesured in
series, I would have to hold everything by hand and shutdown before removing
the meter.
Click to expand...

Yes I think you woud definately have to do that and it would take a while.
But see below.


You would not necessarilly have to cut wires, I was thinking along the lines
of opening up a 'snap'? connector. Then you would need some extra single
wires
to go across the gap, preferably with the right connectors on the end to
make the
job easier. You could feed each of the wires in turn through the meter (you
will
of course need an extra wire to do this.)

The job would not be *that* difficult *if* you had the right connectors.

You would also have to reboot each time you meaasured a different current
*unless* you made every connection a 'double' (two in parrellel) in which
case
you could put the meter in whilst the current went through the other wire
and
then remove the other wire when the meter is connected.

Still sound a bit daunting though doesn't it?


Diagram:-
------------------------------------------------------]] RED (say 12V)
--------single connector>>]]-------------------------]]
------------------------------------------------------]] YELLOW(say 5V)
--------single connector>>]]-------------------------]]
------------------------------------------------------]] BLUE(say 3.5V)
--------single connector>>]]-------------------------]]
Click to expand...
Click to expand...

so you could connect the meter as below
-----------------------remove this line(aa) after you connect the meter-----------]]
-------------------->>]]----------ammeter-------->>]]---------------------
Click to expand...
Click to expand...
-----]]

You then have to reconnect line (aa) before you remove the meter.


If you do plenty of preperation in making all the wires and connectors, ie
wires of a nice
lenght and correctly colour coded, and proper connectors which make a nice
fit leaving
no exposed wire/metal you *probably* can't go too far wrong.

However having said all that I think there are a whole bundle of connectors
coming out of a PSU, like these:-
http://www.buypcdirect.com/images/tutorials/powersupply/powersupply2.jpg
http://www.jdr.com/images/products/P/ps-300atxp4.jpg
http://www.marcee.org/images/Articles/PCPowerSupply3.jpg

So it really does look like a mammoth task!!
I was thinking more of those 4 pin connectors which connect drives,
looks like there are 3 or 4 of those and another big connector
(for the mobo?).
So I think measuring the input power (as suggested) is your only option
in reality. Unless you are very brave (or foolish!!).

However I don't really see why there should be so many wires
coming out of the PSU in the first place.
Too complicated and risky. Anyways, thanks for all your help
guys, and if more ideas do come up, I'm open to suggestions.
thanks all
Click to expand...

Yes I think those meters will only work for AC after having a think about it
A current produces an electrical field, which for AC, is a moving
field, which will induce a small current in circuits mearby (the meters
clamp).
However as the DC field is not moving, no current will be induced, so it
will not measure a DC field.

( I think the AC field is actually an electromagnetic field strictly
speaking
nut it doesn't really matter).

You could make a device to measure the DC field, and hence the
current though it, I think, as a DC field will deflect a compass
needle

I found this
http://www.allaboutcircuits.com/vol_1/chpt_8/4.html
which explains ammeter design, and how you use
shunt resistors so you can measure large currents.

Basically most of the current goes through a low value shunt
resistor and just a small bit goes through the actual ammeter.
A little bit of maths gives your the total current.
I was thinking of that in an earlier post when I was a little 'tired'

Anyway I have (re)learnt a bit myself in my investigations so it has
not been a waste of time!.

Good luck
 
half_pint said:
.... snip ...

I did think the 0.7 was some adjustment for root-mean-squared,
obviously not!
Click to expand...

The power factor for a switching PS is probably most peculiar.
This essentially means that current vs voltage does not follow
ohms law, or that impedance in not constant over time.
 
CBFalconer said:
... snip ...



The power factor for a switching PS is probably most peculiar.
Click to expand...

The use of a switching power supply can indeed mess up the input current
waveform. Ohm's law still applies to the voltage across the sampling
resistor. For better accuracy, it might be well to use a true RMS
voltmeter.
This essentially means that current vs voltage does not follow
ohms law, or that impedance in not constant over time.
Click to expand...
It's not Ohm's law that's in question here!
The problem is that W = EI, reguires that E and I be vectors. If E and
I an E in phase, (1.0 PF), than ordinary measurements will work.

I've found, (for the few cases I've actually measured), that the method
is within the expected measurement accuracy. What methods do you advise
using? Power is a tricky thing to measure, even with the best of
equipment. One of the reasons some people thought cold fusion worked!
;-)

Virg Wall
 
Back
Top