Real power draw of a PC, results from my own PC

gareth170 · 5 Jun 2009 at 15:42

Sasahara said:
The Hi-power could well be under powering (current, not voltage) components, and therefore using less Power

i have a hi-power 900w psu which i have had for over a year without any issues, about 4 months ago i had the psu tested because it was a year old, the currant, voltages, rails are all performing at the right spec, the psu doesn't under power components..

best psu i've bought

james.miller · 5 Jun 2009 at 15:51

if the psu was undervolting, but not enough to shut the pc down or affect the pcs operation, it would force the psu to supply more current, surely? a component will still need to same wattage, from one extreme of its operating voltage to the other - less voltage = more current. the end result is still the same at the plug socket, give or take depending on the psu's efficiancy at any given curent demand, and im sure it would be the same visa versa

Sasahara · 5 Jun 2009 at 16:05

Fortunately it doesn't work like that, or we'd be having house fires every time there was a dip in voltage supply.

If it's under voltage or under not supply the required current you just get lower performance, at this low level it would be hard to check.
As we are only talking about a very small amount here, it's well within the ATX spec.

But you can't use you wall power readings to compare efficiency.

And I'm not even saying the Hi-Power is lower, but it Could be.

BTW what sort of PSU test?

james.miller · 5 Jun 2009 at 16:10

Sasahara said:
Fortunately it doesn't work like that, or we'd be having house fires every time there was a dip in voltage supply.

no because the psu would shut down - we're not talking about the 12v rail dipping to 5v and still trying to operate - the tolerances are far smaller than that. a dip in performance you can measure very easily and if that did actually happen if the 12v rail drops a volt, we would have seen some kind of measurement of it by now.

here's a good example of testing at the wall on a corsiar hx620 http://www.hardocp.com/article.html?art=MTMwMSw0LCxoZW50aHVzaWFzdA==

120v input - 613w output / 760w input ( 6.33 amps @ 120v)
100v input - 613w output / 798w input (7.98 amps @ 100v)

efficiancy dropped and the input increased by 38w/1.65a but it still provided a clean output and didnt complain. less input voltage = more current required. now why wouldnt that work at the component level drawing power from the supply?

user1453 · 5 Jun 2009 at 16:13

actually if the mains voltage dips the power supply will draw slightly more current but the voltages that the motherboard sees would remain the same

most of the modern computer psus can work with 110v,, american voltage, this is because the voltage conversion is done by chrging up a set of capacitors, look up

it would take a huge dip in power to get anywhere near the 13A limit for a mains plug, ie down to the point at which a computer not running would be the least of your problems

because the pc powers supplies are switch mode the output voltages can remain within tolerence with the invput voltage being a bit different to the uk mains specification

btw; I'm a second year electrical engineering student

operation of pc power supplies

james.miller · 5 Jun 2009 at 16:21

Sasahara · 5 Jun 2009 at 16:21

This isn't about the mains supply to the PSU.

The PC components are not Switched-mode powers units and will not draw more current if they receive lower voltage.

gareth170 · 5 Jun 2009 at 16:23

Sasahara said:
BTW what sort of PSU test?

my friend tested it which is a professional electrician/engineering and he as all the professional equipment

james.miller · 5 Jun 2009 at 16:26

Sasahara said:
This isn't about the mains supply to the PSU.

The PC components are not Switched-mode powers units and will not draw more current if they receive lower voltage.

no but the PWM powersupplies on a motherboard that drive the cpu ram, and also the gpu's, are. drives arent, to my knowledge.

Sasahara · 5 Jun 2009 at 16:26

james.miller said:
no because the psu would shut down - we're not talking about the 12v rail dipping to 5v and still trying to operate - the tolerances are far smaller than that. a dip in performance you can measure very easily and if that did actually happen if the 12v rail drops a volt, we would have seen some kind of measurement of it by now.
?

Talk about wrong end of the stick. The house fires reference was in genereal to the idea that components will draw more current if they receive less voltage when they are not Switched-mode powers units like all the PC components and say for instance an electric oven. If ovens did draw more current everytime there was a brown out you have a hell of a lot of blow fuses.

james.miller · 5 Jun 2009 at 16:29

see above

user1453 · 5 Jun 2009 at 16:32

a decent psu will shut itself down if it canot supply voltages within the atx specification so the question of what happens if pc componets do not recieve the correct voltage is irelivent

and active componts such as switch mode regulators can draw more current if the voltage drops however siple things like resistors and lightbulbs cannot

Sasahara · 5 Jun 2009 at 16:33

gareth170 said:
my friend tested it which is a professional electrician/engineering and he as all the professional equipment

Not many electricians would have the equipment it test specificly for we have been discussing. Or even a reason to.

Sasahara · 5 Jun 2009 at 16:38

user1453 said:
a decent psu will shut itself down if it canot supply voltages within the atx specification so the question of what happens if pc componets do not recieve the correct voltage is irelivent

and active componts such as switch mode regulators can draw more current if the voltage drops however siple things like resistors and lightbulbs cannot

The ATX spec has a reasonable allowance and we are only talking about a few watts. Seem some people are adding thing that havent been written.

Nothing to do with extremes of power. Just you can't go by wall readings to compare PSU efficiency, simple no?

user1453 · 5 Jun 2009 at 16:44

I would suggest that you could measure the power draw at the wall plug for a psu, load your computer, then change psus and test again, the one that has less power draw at the wall will be the most effiecient at the load for that particular pc

Stonedofmoo · 22 Jul 2009 at 20:38

Well I've just changed from a Q6600 to a Q9550.

Both are clocked at stock speeds but undervolted.
Q6600 @ 2.4Ghz, 1.025v used 240w full load and 138w at idle
Q9550 @ 2.83Ghz, 0.96v uses 180w full load and 130w at idle

Both are excellent chips. I was thinking of using my Q6600 in a mini-itx HTPC, but none of the boards available allow you to undervolt which is irritating. The low voltage the Q6600 allows is it's best feature.

RJC · 22 Jul 2009 at 21:12

Intel I7 @ 4ghz with ATI 4870x2.

Running with Prime 95 & 3DMark 06 pulls 525w ~
Idle 240w ~

james.miller · 22 Jul 2009 at 21:27

still less than my tv LOL.

That's a huge draw though. blimey.

stonedofmoo thats a decent drop in usage from the die shirk. i wouldnt personally use the q6600 in a htpc environment though. not that ytou'd have any trouble cooling it or anything like that, but something like an e5200 would be better suited and would have a far lower power draw

domthecondom · 23 Jul 2009 at 12:38

How much are these meters?

Stonedofmoo · 23 Jul 2009 at 13:40

About £12 from various online retailers.