Power Meter

The forum for non-component-related silent pc discussions.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

andyb
Patron of SPCR
Posts: 3307
Joined: Wed Dec 15, 2004 12:00 pm
Location: Essex, England

Post by andyb » Wed Nov 29, 2006 5:36 am

From an overall power usage/electricity cost point of view.

A 28W system running 24/7 for a whole year would cost £34.34 to run at the cost of £0.14p per KWh. Compared to my current system (68W) at a cost of £83.40p for a year. It would take 4 or less years to pay back the cost of the system so this is not financialy viable (quickly) unless you are looking at a new/another PC anyway. I have no idea what the cost of electricity is in other parts of the world, and to be honest I am not 100% sure that it costs £0.14p pew KWh in the UK at the moment, I can update my spreadsheet as necessary.


Andy

nutball
*Lifetime Patron*
Posts: 1304
Joined: Thu Apr 10, 2003 7:16 am
Location: en.gb.uk

Post by nutball » Wed Nov 29, 2006 7:12 am

andyb wrote: I have no idea what the cost of electricity is in other parts of the world, and to be honest I am not 100% sure that it costs £0.14p pew KWh in the UK at the moment, I can update my spreadsheet as necessary.
From my supplier it's about 20p/kWh up to a certain threshold, then 11p after that, so 14p won't be miles off.

Anyway those are very impressive power consumption figures, much better than I've been achieving with my Sempron/GF6100-based systems.

andyb
Patron of SPCR
Posts: 3307
Joined: Wed Dec 15, 2004 12:00 pm
Location: Essex, England

Post by andyb » Wed Nov 29, 2006 7:29 am

Something that has just struck me, when I was thinking about the PSU efficiencies. The Power Meter takes the readings of whatever the device is that is plugged into it (a power suppy in this instance), the power usage is "A/C power at the socket", and not the actual power that the PC is using DC.

Having looked at the SPCR review of this PSU, the efficiency at this power draw might only be 55-65%. Judging by the difference between the 40W, 65W and 90W DC output levels, and the rate that the efficiency drops off at I would be more likely to believe that the efficiency at 28W power draw to be nearer 55% efficiency.
http://www.silentpcreview.com/article596-page4.html

Feel free to have a stab in the dark as to how much the power draw of this system could be if a power brick and Pico-PSU were to be used, I was thinking of efficiencies in the range of 75-85%.

Tested here, with fantastic efficiency.
http://www.silentpcreview.com/article601-page3.html

There are 2 drawbacks to a PICO PSU+ Powerbrick solution, it adds £40+VAT to the cost of a system (plus a case fan), AND a system fan of some kind would be needed, wheras the fan in the S12-330 would be enough to cool the whole system and itself.


Andy

andyb
Patron of SPCR
Posts: 3307
Joined: Wed Dec 15, 2004 12:00 pm
Location: Essex, England

Post by andyb » Wed Nov 29, 2006 5:07 pm

No testing today :(

I had to actually do some work :evil:


Andy

andyb
Patron of SPCR
Posts: 3307
Joined: Wed Dec 15, 2004 12:00 pm
Location: Essex, England

Post by andyb » Fri Dec 01, 2006 7:50 am

I have done some testing today, and my plan to make this system a single fan PC looks workable.

I will post some information later this evening.


Andy

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Fri Dec 01, 2006 10:18 am

andyb wrote:I am very very impressed with my new toy, and im not talking about the power meter.....

this morning I recieved my GA-M61PM-S2 Motherboard.
http://www.gigabyte.com.tw/Products/Mot ... uctID=2373

And have put it together with a 3500+, a single 512MB stick or DDR2-667 a slimline SATA 7200.9 160GB and a SeaSonic S12 330W PSU, and an LG DVD-RW.

The lowest power usage i have seen it run at is just 35W, and the highest so far (installing drivers) has been 63W. On the desktop, with only Windows XP and the drivers installed it is varying between 35W and 37W, Coon n Quiet is enabled, DVD drive not spinning.

33-34W as above but without the optical drive attached.
Your report was pretty amazing, so I fired off an email to my contacts at Gigabyte, citing your power measurements. I received a reply last night. They were also surprised... and a bit dubious about the accuracy of the measurements:
gigabyte wrote:M61PM-S2 is a great integrated product with low power consumption, but I am not sure the accuracy of your reader's measurements.
The AC power measurement device you are using, I presume, is very similar to the Kill-a-Watt. I've been informed by a contact at Ecos Consulting (80 Plus) that at lower loads, the Kill-a-Watt gives less than perfectly accurate readings depending partly on PF correction. I don't have their data nor do i know exactly what conditions prevailed, but it would be great to have those results double-checked with some other measurement device if at all possible.

Ideally with something like this extech, you'll be assured of better than 1% accuracy.

Given that the PSU has APFC, you might also be able to use a low impedance resistor as a shunt in series with the AC input and measure the voltage drop across the resistor. Then you can use ohm's law to calculate the power. (P=V2*R) With a 0.1 ohm (25W) resistor, 37W power draw would show a ~1.92V drop across the resistor. This method relies on the PF being very high (which you have with the seasonic) and a highly accurate multimeter. It's also a bit dangerous...

Palindroman
Posts: 247
Joined: Fri Oct 20, 2006 1:00 am

Post by Palindroman » Fri Dec 01, 2006 11:23 am

Still, let's hope Andy's measurements are accurate. 8)

Okay, right at the start of my researching the lowest power possible system I checked a list (http://balusc.xs4all.nl/srv/har-cpu-amd-k8.php) to check out the various TDPs of different procs. This is what I found:

Turions 25-35 W
Sempron Manila 35 W
Athlon64 X2 3800+ 35 W

and

Athlon 64 3500+ 35W

This could very well be the one Andy used in his system, and thus partially explain the extreme low idle measurements. Andy, what's the part number of your CPU? Is it ADD3500IAA4CN or ADD3500CNBOX?

andyb
Patron of SPCR
Posts: 3307
Joined: Wed Dec 15, 2004 12:00 pm
Location: Essex, England

Post by andyb » Fri Dec 01, 2006 1:17 pm

Having read your post Mike, I just started thinking about how electronics could interfere with a reading, and I just realised that my first bout of testing had different results to todays testing.

This is how the power gets from the wall socket to the PSU, some might laugh others will be disturbed.

Wall Socket > Surge Strip > UPS > Power Strip > Power Strip > PSU.

Todays configuration only had a power strip, and the lowest wattage was 29W (briefly), and it was typically at 30W and thats after I had disconnected the CPU fan, so the wattage should if anything have gone down.

The maximum power draw I have persuaded the PC to take was today running "Prime 95" and it drew 62W.

I can lay my hands on another identical power meter, but that wont be until the 20th December.

Spending a small fortune on a really high quality power meter is not possible, and I dont know anyone in this part of the world who has anything that I can borrow.

How exactly would I go about the resistor test.??? I dont know how accurate my Voltmeter is either, I could however test the Mains with the Voltmeter and then compare it to the power meter.

What I could really do with is testing a few items that are known to draw a specific amount of power in the 20W-60W range and then identify how much the power meter is reporting. A desk lamp with a 60W bulb is an obvious choice, but how do I really know that the 60W bulb actually uses 60W :?

Any and all ideas are welcome to help me calibrate my meter, a graph can then be drawn to identify a more accurate measurement.

I have got stacks of old crap PSU's that have all been tested and work, I can identify what their power factor is (power meter function), would this help shed some light on the subject.???

I dont know the part number on the chip, I will look tomorrow, also those power usage figures are the MAX and not the MIN, and they will vary anyway from one sample to the next, and more specifically from one batch to the next.


Andy

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Fri Dec 01, 2006 1:52 pm

What I could really do with is testing a few items that are known to draw a specific amount of power in the 20W-60W range and then identify how much the power meter is reporting. A desk lamp with a 60W bulb is an obvious choice, but how do I really know that the 60W bulb actually uses 60W?
Well, I have an energy-saving lightbulb that is rated at 11W, and my power meter shows 12.25W, so it seems to be pretty accurate.

andyb
Patron of SPCR
Posts: 3307
Joined: Wed Dec 15, 2004 12:00 pm
Location: Essex, England

Post by andyb » Fri Dec 01, 2006 2:00 pm

I will try a few lightbulbs in a desklamp to see what the power meter says.

I am pretty sure I have 40W, 60W and 100W bulbs to try, but nothing below 40W.


Andy

nutball
*Lifetime Patron*
Posts: 1304
Joined: Thu Apr 10, 2003 7:16 am
Location: en.gb.uk

Post by nutball » Sat Dec 02, 2006 12:43 am

My power meter is a different model (brennenstuhl PM230). I think I got it off eBay for ~£10-15 IIRC.
The maximum power draw I have persuaded the PC to take was today running "Prime 95" and it drew 62W.
That's pretty close to the maximum I see from my box under load (75W if it has a hard-drive attached, ~65W without). My idle load is ~50W though. I wonder if the difference in the differences is telling us something -- ???

Felger Carbon
Posts: 2049
Joined: Thu Dec 15, 2005 11:06 am
Location: Klamath Falls, OR

Post by Felger Carbon » Sat Dec 02, 2006 1:00 am

MikeC wrote:The AC power measurement device you are using, I presume, is very similar to the Kill-a-Watt. I've been informed by a contact at Ecos Consulting (80 Plus) that at lower loads, the Kill-a-Watt gives less than perfectly accurate readings depending partly on PF correction. I don't have their data nor do i know exactly what conditions prevailed, but it would be great to have those results double-checked with some other measurement device if at all possible.
Mike: The way to check the "low level" (in)accuracy of the Kill-A-Watt is very simple: take a second power strip, or 2-1 adaptor, and plug it into the KAW. Now you can put two loads on the KAW. Plug an incandescent lamp into one of the outlets, preferably 80-120W. Wait for KAW to stabilize (for the lamp to warm up). Now plug the suspect computer load into the other outlet. Watts are additive. If the KAW reading increases 35W, then it was accurate in the first place. If not, you'll see the result since the reading is out of the mud. :D

Palindroman
Posts: 247
Joined: Fri Oct 20, 2006 1:00 am

Post by Palindroman » Sat Dec 02, 2006 2:25 am

Here's my system:

Arctic Silentium T4 350 Watt
Asus A8N-VM CSM
AMD Athlon 64 3200+ Venice 2.0 GHz, 2000 MHz, Boxed
Samsung HDD SpinPoint PS80SD 80 GB, 7200 Rpm, 8 MB, S-ATA II/300 Kingston ValueRam 1024 MB, PC3200, 400 MHz, 3
NEC DVD-RW ND-4550A
Neovo F-417 17" TFT

Code: Select all

	      standby   idle   load
Monitor      1.3   24.6   24.6
Modem        6.2    6.5    6.5
PC           3.0   54.0   86.2

Total       10.5   85.1  117.3

Load wasn't Prime95 or CPUBurn, just 100% CPU while converting avi to dvd.

andyb
Patron of SPCR
Posts: 3307
Joined: Wed Dec 15, 2004 12:00 pm
Location: Essex, England

Post by andyb » Sat Dec 02, 2006 2:28 am

I have a list of items that I can test, that have a relativly well know power consumption.

HDD's (various makes and models).

4 Spare Seasonic PSU's (all of which draw a small amount of power whilst on standyby, and as Felger said I can use them addititvely). But again, during my first testing I found that the PSU drew a tiny 2W, now its drawing 4W.!!!

Monitors various, mostly 17" CRT's.

I like the Idea that Felger came up with, add more identical items and identify the consumption of each additional item.

I am going to hook up 4 additional Seagate 7200.9 single platter drives (all PATA 1 is 80GB the other 3 are 160GB not thet it should make any difference).


I can test 4x S12-330W PSU's, I will test them all individually, and then add the results with each addition and see if it adds up.


Andy
Last edited by andyb on Sat Dec 02, 2006 3:27 am, edited 1 time in total.

andyb
Patron of SPCR
Posts: 3307
Joined: Wed Dec 15, 2004 12:00 pm
Location: Essex, England

Post by andyb » Sat Dec 02, 2006 3:19 am

OK this is what I plan to do.

I have 2 baselines to test additional drives with, both baselines have the PSU (S12-330) plugged into the Power Meter, which is plugged into an ordinary 4gang extension lead (not surge protected), which is plugged into the wall.

The incoming voltage goes up and down quite a lot, it varies from 235v to 242v.

Baseline 1 and 2 specs that wont change: S12-330 PSU, GA-M61PM-S2, AMD 3500+ (I will update with model number later), 1 stick of Corsair Value Select 667MHz CAS-5, 1x Seagate Momentus 5400.3 80GB PATA (single platter).

What will change is addindg additional HDDs (as mentioned above).

Baseline 1 is going to be tested in the BIOS, therefore no power management can interfere with any results, and the CPU etc etc are doing nothing at all, but they are powered.

Baseline 1: 53W.
Baseline 1: with 1x additional HDD 58W
Baseline 1: with 2x additional HDD 64W
Baseline 1: with 3x additional HDD 69W
Baseline 1: with 4x additional HDD 74W

Looking at the increased Wattage per additional drive. 5W, 4W, 5W, 5W.

The evidence is quite conclusive that these drive consume 4-5W at idle, when the elctronics are not doing anything, they are only powered, they are not connected to the motherboard.

Baseline 2 is going to be tested on the desktop, with no drivers installed at all, except the AMD CnQ driver, I have 2 windows open, CPU-Z to identify that the CPU is running at 1GHz, and Task manager to identify that the CPU is not being used.

Baseline 2: 33W.
Baseline 2: with 1x additional HDD 37-38W
Baseline 2: with 2x additional HDD 42W
Baseline 2: with 3x additional HDD 47W
Baseline 2: with 4x additional HDD 52W

Looking at the increased Wattage per additional drive. 4-5W, 4-5W, 5W, 5W.

Again this looks to be pretty conclusive, I would expect very small changes in the Wattage used by the additional drives between Basleine 1 & 2. This is because the PSU efficiency tails of quite sharply the lower it goes.

If I get time, I will also test a higher powered system (100W~ baseline) with those extra drives to identify whether the Power Meters accuracy drops a lot as the power draw drops.

I hope that these figures make sense to the SPCR readers.


Andy

mb2
Posts: 606
Joined: Sun Jun 06, 2004 2:42 pm
Location: UK

Post by mb2 » Fri Dec 22, 2006 8:24 am

andyb wrote:There are 2 drawbacks to a PICO PSU+ Powerbrick solution, it adds £40+VAT to the cost of a system (plus a case fan), AND a system fan of some kind would be needed
anyone else thinking that when the 65nm semprons come out this may actually offer us the holy grail of 0 fan computing? :twisted: and the pico + psu shouldn't be £40 more than a SS (which are quite expensive here anyway).. u can get the pico's really reasonably on ebay (due to $/£ E.R. and light shipping costs).. and low-ish power bricks aren't too bad..
and a new post was warranted methinks andy :P seems good..

edit: MikeC.. i think they are exactly the same internals as the kill-a-watt /SS power angel IIRC.

Devonavar
SPCR Reviewer
Posts: 1850
Joined: Sun Sep 21, 2003 11:23 am
Location: Vancouver, BC, Canada

Post by Devonavar » Sun Dec 24, 2006 12:41 pm

They can't be exactly the same. Kill-a-watt / Power Angels are 110V power only. I'm don't doubt that they come from the same place though...

andyb
Patron of SPCR
Posts: 3307
Joined: Wed Dec 15, 2004 12:00 pm
Location: Essex, England

Post by andyb » Sun Dec 24, 2006 6:10 pm

Pictures of the unit will be posted, and the manual, but the manual is identical to the one posted by me in one of my first posts within this thread (PDF format).

Exacly how accurate the meter is is not of the greatest imortance to me. When you look at the fact that the manufacturer claims a max of 2% and I would imagine a worst of 10x (20%) that amount we are still looking at a very low power PC, 40W. If the margin of error is 10%, you are still looking at a 36.5W PC :P

Either way, for my purposes that is brilliant.


Andy

andyb
Patron of SPCR
Posts: 3307
Joined: Wed Dec 15, 2004 12:00 pm
Location: Essex, England

Post by andyb » Sat Feb 17, 2007 5:08 am

At long last I have found something that is very low power to test.

A Seagate external HDD (3.5")

The external drive specifications for power usage are non-existant, but the HDD that is in the caddy is a known quantity.

http://www.seagate.com/docs/pdf/datashe ... 7200_9.pdf

The model of drive tested is an 80GB model (I would guess that its a 7200.7 and not a 7200.9 that is in the above link, however power usage is going to be near identical).

We also need to work into the equation the power wasted by the 230v --> 12v adapter, and the power used by the PATA --> USB adaptor.

The adaptor is made by "Li-Shin", a well known quantity in the AC/DC adapter market, with the model No. 0322B1224.

It has a worldwide rated input of 100-240v, and an output of 12v @2A, which equates to 24 Watts.

Upon turning the HDD on the Power Meter measured a peak of 28W, which of course is 4W over the maximum output of the AC/DC adapter, which is something that I will re-test on Monday.

The power immediately dropped down to 12W which is less than the rated seek of the drive, let alone the power used while "spinning up".

The rated power usage of the drive whilst Idle is 7.2W yet the measured power usage settled at 6-7W once the drive had finished spinning up and became idle.

Once I started using the drive it was hovering between 7-8W.

I believe that my power meter is under-reporting the power usage at low power levels, as it seems to be an unreasonably low level, even if Seagate's rated power consumption is a little high.


Andy

kater
Posts: 891
Joined: Thu Sep 07, 2006 11:20 pm
Location: Poland

Voltcraft power meter - a real man's toy :D

Post by kater » Thu Feb 22, 2007 1:23 pm

Hello :)

I recently bought a nifty power meter – Voltcraft Energy Check 3000. Made in Holland, cost me EUR 30. Conrad carries them in EU, but I bought mine in a tiny shop for local ultrageeks. Anyway, it comes with an LCD display and a variety of useful functions including the recording of all-time lowest and highest readings, two tariffs etc. Producer claims accuracy of 1% - pretty good compared to other power meters I found around here. Can withstand up to 3000W.

I have now spent some time playing with it and would like to present a few results. We'll start with my machine. Later on I'll add other computers I have access to.

My current machine:
CPU – A64 939 Manchester X2 3800+ @ default (2000) or @ 2400 (auto vcore in BIOS, 1,1 idle, 1,4 load); TDP 89W
Mobo – Asus A8NE, disabled audio & LAN
VGA – Nvidia 7900GTX @ default (650/800) or OC’ed to 680/850, no volt mods or anything, stock fan at 25% in idle, at 40% when in Oblivion and at 100% under max load.
RAM – 4 x 512 GeIL 3200 @ auto voltage
HDD – Seagate Barracuda 7200.10 250GB (it was a gift, not my choice, you know)
DVD – LG GSA-4165B (whatever)
Audio – vanilla SB Live!
Dial-up modem (I've just moved, stop laughing, you)

Few more words of explanation on how I tested things.
Idle – Windows, after booting up. RightMark CPU, Avast, Riva Tuner, Everest in the tray.
Work – MS Word, MP3 player, IE. Hitting the keys, moving the mouse.
Booting up – with ODD full.
Oblivion – playing Oblivion for at least 30 minutes, 1024 x 768, eye candy maxed up (with only some shadow options reduced) so as to achieve quite fluent gameplay (at least +20 FPS outdoors and +40 indoors, according to Fraps), Fraps to catch fps, no other programs running – only the tray.
Shutting down – well, you figure.
Max load stress – Orthos, 3DMark 06 benchmarking, defragmenting one partition, copying between partitions, burning a DVD, blasting MP3 player, Everest, lots of tray programs, MS Word, all fans at max RPM. (If you think I still can do sth. to make the machine suffer more please let me know!)

I used the record function to store the hi/lo readings and ran each test a few times. Max readings are the very highest numbers recorded by the device, even if they appear only momentarily.

PC with CPU & VGA @ default settings
IDLE 94 - 98
BOOTING UP 166
WORK 98 - 101
OBLIVION 212
MAX LOAD 235
SHUTTING DOWN 134

PC with OC’ed CPU & VGA
IDLE 96 - 99
BOOTING UP 179
WORK 99 - 102
OBLIVION 222
MAX LOAD 254
SHUTTING DOWN 140

Some additional explanations.
Oblivion usually required 15-20W less, especially when the game wasn’t pushed.
The same with the max load situation – it usually drew ca 15W less, and only bumped to +250 a few times.
After a few minutes of idling the consumption would even drop several times to 94W for stock machine and to 96W for OC’ed machine (HDD doesn’t shut down, mind you) – these are the lowest number recorded.
Work results are what can be seen on the LCD most of the time – I wanted to give you an idea of what the machine draws most of the day. Yeah, the power draw seldom shoots up to 105W or 108W when I work. But it mostly stays within the indicated range.

I’ve also played with the CPU OC’ed to 2600 with vcore bumped to 1,5V, but the results were so strange (sky high!) I’ll have to repeat the tests. I’ll also check some other PC’s I have access to and post more results later.

I hope you enjoyed reading this longish but otherwise neat post and will find it useful :D
Last edited by kater on Thu Feb 22, 2007 2:31 pm, edited 2 times in total.

Post Reply