Sign in with
Sign up | Sign in
Your question

Fun with a Kill-A-Watt: updated

Last response: in CPUs
Share
December 27, 2006 1:58:02 AM

I received a Kill-A-Watt meter for Christmas this year, so I decided to do some testing:

K6-2 test system: As described below

Relisys 15" CRT (built Nov. 97): 75 watt active, 4 watt standby
Computer-idle on desktop: 42 watt
Computer-100% CPU load: 67 watt
CD-ROM drive spinup: Add 6 watts to previous totals.
Hard Drive in standy: Subtract 3 watts (per hd) from previous totals.
Floppy Drive: forgot to test. :oops: 


Pentium M test System: Dell Inspiron 1200: 1.4Ghz Pentium M, Banias core, underclocked/undervolted @ 600MHz/0.700volt. 256MB DDR pc2700 ram, 30GB 4200rpm 2.5" hdd, 14" screen, CD-RW/DVD-ROM combo drive, factory 18V 62 watt max AC adapter.

laptop at lowest screen brightness, idle: 11 watt
full brightness, idle: 17 watt
CPU @ 100% load, lowest screen brightness: 15 watt
laptop in standby mode: 1 watt
combo drive spinning, laptop with lowest screen brightness: 15 watt
laptop idle, lowest screen brightness, with hdd in standby: 9 watt
Wi-fi add in card: 1 watt


I'll test my Pentium D system when I get home.

More about : fun kill watt updated

December 27, 2006 5:48:09 AM

Quote:
I received a Kill-A-Watt meter for Christmas this year, so I decided to do some testing:

K6-2 test system: As described below

Relisys 15" CRT (built Nov. 97): 75 watt active, 4 watt standby
Computer-idle on desktop: 42 watt
Computer-100% CPU load: 67 watt
CD-ROM drive spinup: Add 6 watts to previous totals.
Hard Drive in standy: Subtract 3 watts (per hd) from previous totals.
Floppy Drive: forgot to test. :oops: 


Pentium M test System: Dell Inspiron 1200: 1.4Ghz Pentium M, Banias core, underclocked/undervolted @ 600MHz/0.700volt. 256MB DDR pc2700 ram, 30GB 4200rpm 2.5" hdd, 14" screen, CD-RW/DVD-ROM combo drive, factory 18V 62 watt max AC adapter.

laptop at lowest screen brightness, idle: 11 watt
full brightness, idle: 17 watt
CPU @ 100% load, lowest screen brightness: 15 watt
laptop in standby mode: 1 watt
combo drive spinning, laptop with lowest screen brightness: 15 watt
laptop idle, lowest screen brightness, with hdd in standby: 9 watt
Wi-fi add in card: 1 watt


I'll test my Pentium D system when I get home.


Nice experiment... looking forward to the Pentium D results --- I hope it does not go off scale :) Richter scale? :wink:...or Geiger counter?
December 27, 2006 2:15:05 PM

This is by far one of the most fascinating posts I have ever read here. Please do as much as you can. For fun can you also do some of the following:

Television
Toaster Oven
Lamp with 60 watt lightbulb (ya, I know it should read out 60, but just for kicks)
Phone?
Computer with USB memory drive compared to a computer without one
Cable Modem
Wireless Router

And if you have a few extra G's could you test a Quad FX system?

Thanks!
a c 480 à CPUs
December 27, 2006 2:18:05 PM

Quote:
I received a Kill-A-Watt meter for Christmas this year


Now that's the kinda gift I would like to recieve.
December 27, 2006 2:29:59 PM

Yes, I bought the UPM Energy meter (similar to Kill A Watt, just a different brand) at Canadian Tire for $28 including tax, about a year ago. I have measured virtually everything in the house that is on a 15 amp service.

There are two differences between the UPM and the KAW: the UPM has non-volatile memory (if you install 2 batteries), and the KAW can read out power factor (efficiency from the wall to the consumption of whatever's being measured). The UPM does not have the power factor feature.

I have all my testing notes somewhere, but it would take too long to include results in this post. Currently, I have the UPM as the first device from the wall, then my UPS, then the tower, monitor, and two external hard drives.

These are really fun gadgets!

-Bob
December 27, 2006 2:32:37 PM

Quote:
I have all my testing notes somewhere, but it would take too long to include results in this post. Currently, I have the UPM as the first device from the wall, then my UPS, then the tower, monitor, and two external hard drives.


What's your reading and system specs?
December 27, 2006 2:46:01 PM

Quote:
I have all my testing notes somewhere, but it would take too long to include results in this post. Currently, I have the UPM as the first device from the wall, then my UPS, then the tower, monitor, and two external hard drives.


What's your reading and system specs?

For devices connected to the UPM:

The UPS is a Tripp-Lite SU1000XL, 1000VA, 800 watts, dual conversion type where power is always being drawn from the UPS battery, and it is always being charged: consumes 35 watts

The Chaintech VFN4 AMD 3500+ consumes 12 watts when the comp is shut down via XP

The LCD (Samsung 930 MP 19 inch) consumes 2 watts when shut down by XP

So, when my systen is shut down, it consumes 49 watts.

When it is running and XP is just idling, the UPS, tower and monitor consume 160 watts on average: monitor 35, UPS 35, meaning that the tower is consuming 90 watts.

Peak consumption (according to the UPM peak consumption variable) over the last 781 hours (since I last reset it to zero) is 222 watts.

I don't normally have the external 3.5 inch USB hard drives on, but they consume 7 watts each.

My 900 watt microwave consumes nearly 1400 watts :!:

-Bob
a b à CPUs
December 27, 2006 3:35:54 PM

I've had a Lexa PM300-consumptionmeter since last summer.
My whole pc system is currently connected to it, including AMD64 3000+, nVidia 7800GS, 1,5Gigs ddr400, 4 IDE harddrives and IDE dvd-burner in Antec Sonata-case with 380W Truepower2,0 PSU, 17inch Viewsonic TFT and Creative 2.1 speakers.
The peak consumption has been 229 watts and currently its using 170watts just writing this. Speakers and monitor unplugged the consumption drops about 30 watts.

Im not sure how reliable the measurements are. I've noticed there is some debate going on in some finnish forums about this particular meter and its ability to correctly measure rapid current spikes.

Tech.coordinator asked about the 60w lightbulb, and guess what, it read 62 watts, close enough :) 
December 27, 2006 4:18:55 PM

A lot of electronics will still consume electricity when 'off.' Pretty sinister if you ask me. Best thing to do? Unplug everything when not in use. :wink:
December 27, 2006 5:22:12 PM

Quote:


Im not sure how reliable the measurements are. I've noticed there is some debate going on in some finnish forums about this particular meter and its ability to correctly measure rapid current spikes.

Tech.coordinator asked about the 60w lightbulb, and guess what, it read 62 watts, close enough :) 


----------------------

The UPM claims 0.2% accuracy with wattage above about 30 watts, and 2% below that. This is plenty good enough for non-commercial applications. As far as I can tell, it refreshes readings every half second.

Yes, I would expect these household meters to have some difficulty reading current spikes, because they weren't designed for that. They are designed to give averages so that average current consumption can be measured over a period of time, so that the cost of electricity can be calculated. These are very different objectives from a commercial-grade tester, where the objectives are the quality and specifications of the circuit.

My household voltage reading is dead on at 119.9 + /- .8 because I had my household voltage checked by my utility company a couple years ago.

Speaking of light bulbs, I have converted to 15 watt flourescent compacts wherever I was using an incandescent. These 15 watt bulbs consume 18 watts.

-Bob
December 27, 2006 5:40:37 PM

Quote:
A lot of electronics will still consume electricity when 'off.' Pretty sinister if you ask me. Best thing to do? Unplug everything when not in use. :wink:


As a general rule, I agree. However, in my case the hog when my system is off is the UPS (35 watts). But it is impractical for me to turn it off. If I unplug it, it screams bloody murder! The only way to shut it up is to unplug everything from it so the load on it is zero, then pull the plug.

I can live with the 35 watt consumption in return for 35 minutes run time in case of a power failure.

-Bob
a b à CPUs
December 27, 2006 6:00:45 PM

The Lexa is reading the line voltage as 224V and when measuring with multimeter it reads 220V, so it is slightly off anyways. Funny thing that, I thought we had 230V household voltage nowadays here in Finland. Or maybe the old multimeter is finally tossing in the towel... hmm i need to check that one...

Lots of electronics have all kinds of inverters and choppers that draw current in non-sinusoidal form, and I've been told that these cheap meters cant really read it accurately, even in average. I was referring to this earlier with the current spikes.

edit: yep it should be 230V, it used to be 220 earlier like 10 years ago, guess we are a bit slow changing things around here :p 
December 27, 2006 6:53:54 PM

Quote:
Yes, I bought the UPM Energy meter (similar to Kill A Watt, just a different brand) at Canadian Tire for $28 including tax, about a year ago. I have measured virtually everything in the house that is on a 15 amp service.

There are two differences between the UPM and the KAW: the UPM has non-volatile memory (if you install 2 batteries), and the KAW can read out power factor (efficiency from the wall to the consumption of whatever's being measured). The UPM does not have the power factor feature.

I have all my testing notes somewhere, but it would take too long to include results in this post. Currently, I have the UPM as the first device from the wall, then my UPS, then the tower, monitor, and two external hard drives.

These are really fun gadgets!

-Bob
I want one. I want to test with my 3.0C, then switch it out for my 2.4A@3726 and watch with horror...what happens. :? I know that SANDRA 2004 SP2b estimates my 3.0C @ ~89w and my 2.4A @ ~125-130W. :x
December 27, 2006 6:55:36 PM

The following system:

Thermaltake PurePower 680W PSU
e6700 @ 3.3GHz, 1.425v vCore
8800GTX @ 645/2050
3*7200RPM IDE HDDs
1*7200RPM SATA HDD
SATA DVD±RW drive
2*80mm fan (cooling hdds)
1*40mm fan (cooling chipset)
1*110mm fan (Zalman CPU cooler)
1*40mm radial fan (cooling mosfets)
Asus P5W DH Deluxe
Soundblaster Audigy 1
HP LP2065 20.1" 4:3 TFT
Creative Gigaworks 5.1 speakers.

All plugged in, speakers + monitor on standby: 24W
Speakers on, monitor on standby: 49W
Idle in Windows, no CPU power management: 321W
Full Prime95 load, both cores, small FFTs: 366W
Full TAT load, both cores: 382W
3Dmark06 Return to Proxycon: 404W

Unsurprisingly, the GPU really ups the power usage.

Being in the UK, Ring main is rated @ 32A total, although I have a 13A fuse on the power strip.

Nominal voltage is reading at 245V, that means that I can hit up to 3185W before I blow the fuse in the plug :D 
December 27, 2006 6:59:53 PM

Quote:
The Lexa is reading the line voltage as 224V and when measuring with multimeter it reads 220V, so it is slightly off anyways. Funny thing that, I thought we had 230V household voltage nowadays here in Finland. Or maybe the old multimeter is finally tossing in the towel... hmm i need to check that one...

Lots of electronics have all kinds of inverters and choppers that draw current in non-sinusoidal form, and I've been told that these cheap meters cant really read it accurately, even in average. I was referring to this earlier with the current spikes.

edit: yep it should be 230V, it used to be 220 earlier like 10 years ago, guess we are a bit slow changing things around here :p 


No, thats correct.

Europe used to vary between 220V and 240V in the various countries. The UK for example was 240V±6% (225.6V-254.4V).

Then the EU rationalised all of Europe to 230V±10% (207V-253V).

As the existing ranges were all inside this new range, no supplies were ever actually changed. All that happened was that the manufacturers of appliances were forced to make devices to a 230V±10% tolerance, so that they could be sold across Europe.
December 27, 2006 7:29:54 PM

1Tanker: If you hurry up you can get to your local CT before closing :lol: 

The exact model is UPM Energy Meter EM 100. I found it in the electrical supplies (extension cords, etc.) section.

No geek should be without one :wink:

-Bob
December 27, 2006 7:33:57 PM

Quote:
1Tanker: If you hurry up you can get to your local CT before closing :lol: 

The exact model is UPM Energy Meter EM 100. I found it in the electrical supplies (extension cords, etc.) section.

No geek should be without one :wink:

-Bob
Thx Bob. :D 
a b à CPUs
December 27, 2006 8:08:04 PM

Quote:
Europe used to vary between 220V and 240V in the various countries. The UK for example was 240V±6% (225.6V-254.4V).

Then the EU rationalised all of Europe to 230V±10% (207V-253V).

As the existing ranges were all inside this new range, no supplies were ever actually changed. All that happened was that the manufacturers of appliances were forced to make devices to a 230V±10% tolerance, so that they could be sold across Europe.


eh, well that was convenient :D 
December 28, 2006 4:30:31 AM

Hi Bob. Sounds like you know your way around these meters. I'm still learning how to use mine. Like all boys eager to use their new toys, I didn't read the instructions before using it, and I just noticed something odd. This KAW will measure wattage in what they call "Active Power", which they don't explain much about, and "Apparent Power" which would be the traditional Volt x Amp = Watt calculation. It then takes these two wattage measurements to come up with its "Power Factor" (active power / apparent power = power factor). My question is, what is this active power, how is it calculated, and does the power factor indicate? All the previous measurements were taken in this active power setting, which appears to be lower than the traditional apparent power calculation (sometimes quite a significant difference). I fear I'll have to redo all my measurements. :( 

Below are the (active power) wattage measurements for the Pentium D system:

System Specs:
Dell Dimension 4500 case, OEM 250 watt PSU
Asrock 775i65G
2x256MB pc2100 DDR ram
D 805, stock speed, undervolted to 1.15 vcore
GeForce3 Ti200, 64MB
Leadtek Winfast TV2000 XP Expert video capture card
Aureal Vortex 2 audio card
Seagate Barracuda IV ATA 100 80GB hdd
Seagate Barracuda 7200.7 ATA 100 200GB hdd
12x DVD-Rom drive
16x DVDRW drive
floppy drive
15" Samsung SyncMaster 515v LCD

Samsung Monitor: 14 watt active, 1 watt standby

System power consumption, in Bios: 115 watt
System power consumption, idle in Windows desktop: 100 watt
1 core fully loaded: 125 watt
Both cores fully loaded: 150 watt
Disk Defragment: 107 watt

*What happens when I remove the Ti200 and use Intel's 865G IGP?:
System power consumption, idle: 90 watt

*IGP, and using AC97 instead of audio card:
System power consumption, idle: 85 watt

*IGP, AC97, and removal of capture card:
System power consumption, idle: 80 watt
System in S1 standby mode (basically hard drives in standby and no output to monitor): 67 watt

A CD-ROM spinning continuously at 48x increases power consumption by 20 watts!

System power consumption means only the tower (no monitor) was measured. As I stated earlier, all these readings are the "active power" measurements, meaning that the power draw from the outlet should be higher. For instance, the 15" Samsung LCD consumed 14 watts active power, and 22 watts apparent power, giving a power factor rating of .64, or 64%. I believe the apparent power for the system with the IGP, AC97, and no capture card configuration was somewhere around 130 watt, which is significantly higher than the 80 watts of "active power" measured. As for another reference, a lamp with a 40 watt light bulb measured 42 watt active power, and 42 watt apparent power.
If you can shed some light on this Bob, I'd be much obliged.
December 28, 2006 9:41:22 AM

A normal resistive load will have a power factor of 1, and its actual power will match its apparent power.

With an AC circuit you have a sinusoidal waveform of both voltage and current flow. A capacitive or inductive load will shift the phase of the current waveform, causing the apparent power to be higher than the actual power. The phase difference determines the so called "power factor".

The problem arises in that the supply cables etc have to be rated for the higher apparent power. This isnt a problem for you in your home but would be a major headache for the supply company if everyone ran, say, a highly capacitive load. This is why large companies with alot of power draw are forced to pay for apparent power or pay a surcharge based on the type of load they run, they use power factor correction (a big capacitor for an inductive load, a big induction coil for a capacitive load) to bring it back to being a purely resistive load.

This is also why it is often pretty easy for them to catch people growing canabis/weed : high pressure sodium lighting is a highly inductive load, and coupled with the 12/12 duty cycle it sticks out like a sore thumb.

Recent regulatory requirements have also forced PC PSUs to have Power Factor Correction. Its another thing people see in the spec sheets ("oooh Active PFC!") and get all excited about, when actually it helps the power company not them.
December 28, 2006 10:57:56 AM

Same system as earlier, but e6700@ 401*9 = 3.6GHz, 1.625 vCore (vCore higher than needed, was trying 401*10 and forgot to drop it down :D )

Prime95 on both cores : 477W
TAT 100% load both cores: 512W

3Dmark06 Return to Proxycon: 506W

My CPU is now using more than my GPU :D  lol....
December 28, 2006 11:40:36 AM

I actually bought my cousin a Kill-a-watt for Christmas... I wasn't sure if we were having a family get-together or not and I bought gifts that I'd want in case we didn't... well, damnit... we DID so he has it now. I thought it was a creative gift idea.

Since I got my laptop I believe my electric bill has dipped slightly because 90% of my computing chores on done on it while my desktop sits in the corner powered off.
December 28, 2006 1:26:00 PM

Joefriday: reading my instruction manual, it defines wattage (which it calls "real power") as voltage x current x power factor. So, the wattage reading I am getting would be active power if I understand this correctly. My meter does not actually show the power factor number, just the result of the calculation.

I'm no expert in electricity concepts, but I think your measurements at the active power concept are what is wanted.

-Bob
December 28, 2006 1:42:54 PM

Quote:
A normal resistive load will have a power factor of 1, and its actual power will match its apparent power.

With an AC circuit you have a sinusoidal waveform of both voltage and current flow. A capacitive or inductive load will shift the phase of the current waveform, causing the apparent power to be higher than the actual power. The phase difference determines the so called "power factor".

The problem arises in that the supply cables etc have to be rated for the higher apparent power. This isnt a problem for you in your home but would be a major headache for the supply company if everyone ran, say, a highly capacitive load. This is why large companies with alot of power draw are forced to pay for apparent power or pay a surcharge based on the type of load they run, they use power factor correction (a big capacitor for an inductive load, a big induction coil for a capacitive load) to bring it back to being a purely resistive load.

This is also why it is often pretty easy for them to catch people growing canabis/weed : high pressure sodium lighting is a highly inductive load, and coupled with the 12/12 duty cycle it sticks out like a sore thumb.

Recent regulatory requirements have also forced PC PSUs to have Power Factor Correction. Its another thing people see in the spec sheets ("oooh Active PFC!") and get all excited about, when actually it helps the power company not them.


So some sort of capacitive system would negate the tell-tale electricity footprint?

Interesting, :lol: 
December 28, 2006 2:03:52 PM

Yeah, ideally you'd have two sets of lights on alternate 12/12 cycles, and power factor correction in the form of a highly capacitive load.

Hell, if you can make the power factor exactly 1 (pure resistive load) then you could more than likely get away with bypassing the meter and they would put it down to resistive losses on cables/transformers :p 

I should probably shut up now however...
December 28, 2006 3:56:44 PM

Thanks for the reply. So, for my home, do I pay for active power, or apparent power (in the USA)? I would think that when we're talking about computers, the power factor is analogous to the efficiency rating of the PSU, so if I have a low PF, I have an inefficient PSU? Does this sound correct to you?
December 28, 2006 4:18:18 PM

Just off the top of my head this is how I understand a PSU's efficiency rating... a 100% efficient PSU (this doesn't exist in reality to my knowledge) would take in X number of watts of alternating current and output X number of watts of direct current to power your PC. In reality, 85% efficiency is about as good as we can hope for... so in that case your computer pulls in X watts of AC and the PSU outputs X(.85) watts to power your computer. So yes, a higher efficiency PSU would save you on your electric bill as well as put out less heat.
December 28, 2006 7:30:19 PM

Efficiency is different from Power Factor.

If the PSU has a power factor of 0.90, as does mine, and takes 360W of actual power from the wall, then its "apparent power" is 396VA (voltamps).

Coming out of the other side of the PSU however, would be less than 360W. I cant measure this figure, but I'd guess it is around 90% efficient.

Power factor correction does not make the PC more efficient. It is not about reducing (or increasing) the actual power, only about reducing the apparent power, and increasing the power factor.

Your (Domestic) electricity meter will charge you for actual power, not apparent power. As such, active or passive PFC (or even a PSU with no PFC) shouldn't bother you, unless you feel like helping out the local power company, out of the goodness of your heart ;) 

An interesting article is Here
December 28, 2006 7:51:16 PM

Thanks. That clears up a lot of questions. :wink:

One thing I'd like to point out about my current power measurements, is that despite the tremendous performance of my Pentium D rig compared to the elderly K6-2, the K6 rig actually consumes MORE power than the Pentium D, due to the incredibly power hungry CRT monitor.
December 28, 2006 10:22:44 PM

I've tested a few other items:

Belkin 54G wireless router: 6 watt
Motorolla Surfboard Modem: 6 watt, regardless of "ON" or "Standby
Labtec 2.1 speaker system: 1 to 3 watt, depending on volume.
13" CRT television: 30-35 watt
27" Zenith CRT Television: 80-100 watt
humidifier: 180 watt on low, 360 watt on high.
27" Insignia HDTV, CRT: 100-175 watt
VCR: 6 watt standby, 14 watt on, 17 watt playing VHS tape.
Our Christmas tree: 50 watt
My Parents' Christmas tree: 90 watt
December 28, 2006 11:02:08 PM

Quote:
the K6 rig actually consumes MORE power than the Pentium D, due to the incredibly power hungry CRT monitor.


My 20.1" LCD uses 52W at max brightness, that seems excessive for an LCD to me :/ 
!