Power Draw

[kspn]

Hi,

Sorry if this has already been answered.

I am thinking of replacing a Virtual Machine I am using as a Webserver (apache/php/mysql) with a RasPi and I was wondering if anyone had worked out the Power use over time for a RasPi under some load, No KB/Mouse (SSH), X11 running (via TightVNC) (so cpu use average would probably be at around 50%)

I know that it is kind of a *pick a number form the air* type of question but, in part, I was wondering if anyone had actually dome some tests and checks on this kind of thing.

Also how much of a difference does the power supply used make to the power draw?
1) USB port on router
2) 1a USB 'brick' Blackberry charger for example
3) 1a USB wall adapter HTC For example

Or will I need to get a power meter and check for myself?

[Vindicator]

Lob0426 is running one and he says it is running about 6 watts with the power adapter.

Here's his page on his Raspi http://rich1.dyndns.tv/index.html I think he is intending to post some figures with and without a usb HDD also.

Click on his projects page and it is Raspi #2

[kspn]

Ok, sorry for being dense here, is that 6w per hour/day?

I am unsure exactly how all this stuff is measured

[Vindicator]

That would be per hour.

[pluggy]

No, 6W is a power rating, it is independent of time, if it uses 6W , its all the time, for the time its turned on. How much it uses over time is work, and that is measured in joules(a watt second) or watt hours. Something that uses a lot of power or something that is turned on for a long time will consume kWh (kilowatt hours) or megajoules (millions of joules). Most countries bill their electricity in kWh, some use megajoules. The Pi itself will use around 2 watts, but the inefficiencies of a power adapter and what the peripherals consume will push this up. If it did indeed consume 6 watts (very unlikely IMO), it would use around 1 kWh in a week (somewhere around 9p - 13p per week in the UK or £1 in around 8-11 weeks)

[kspn]

Hmm, it looks like the best bet may be to get a power meter and setup a camera to take an image once per hour so that I can get some idea of the power draw

[pluggy]

Most power meters don't work very well at low powers, they can be very inaccurate. To further complicate the issue, whereas the Pi uses DC (direct current) what you're pulling from the wall socket is AC (alternating current) and is subject to something called power factor. Switch mode power supplies can have very poor power factors. Cheap power meters have a tendency to measure apparent power whereas your electricity meter will measure true power. The better power meters have both and will often calculate the power factor as well. Apparent power is usually measered in VA (volt amps), true power in W (watts). In the DC world, VA = W, but not in AC .The ratio of VA (apparent) to W (true) is the power factor. A good power factor is 1, but usually only occurs in pure resistive loads, small switch mode power supplies can often have a power factor of 2 or more. Feel free to google any of this, but you'll find its close to the mark.

[pluggy]

Whoops, error in the above. Power factor is the ratio of true to apparent power, not the other way around, so a good power factor is 1 and a poor one is less than 1 say 0.5 or less.

[kspn]

Hmm, ok, as a result of that (an I am showing my ignorance here) is it more efficient to increase the power draw on a single device, or to use seperate plugs?

For example, would it consume less power if i used the USB plug on my Router vs a 'Wall Wart'?

[pluggy]

Its generally better for efficiency to have one power supply that will power the whole lot. I have a whole load of networking gear that is turned on all the time and it all uses 12V so I made up an adapter lead to feed them all from a 25 watt power supply, rather than the individual supplies they were supplied with. The compromised power situation on the Pi may dictate otherwise. Micro USB was a very poor choice for feeding the Pi IMO, add to that the polyfuse situation and it gets messy.

I feed my Pi via a serious 5V supply to the GPIO pins and I've short circuited the USB polyfuses. (Now apparently a legitimate thing to do since the new boards don't have USB polyfuses) If you can make do with 2 USBs, short circuiting the poly fuses if you have them has to be better for efficiency than using a powered hub. I wouldn't advise feeding a 2.5" HDD directly from the Pi though. The usual WIfi adapters and kinky keyboards shouldn't be a problem.

[kspn]

I think that most of the issues are form devices being plugged into the USB ports, in my case I won't even have a Keyboard plugged into the RasPi, it will simply be the Pi and Network, so hopefully no issues.

I will however have the RasPi running through a 5000mAh batter so that if I lose power the Pi doens't die with it

[pluggy]

I'd work on 2.5 - 3 watts tops including a not too efficient power supply if you're using it headless without peripherals. The stock answer of 400mA at 5 volts (2 watts) is a 'busy' figure and it typically uses less if its 'just sat there doing nothing'. A kWh every couple of weeks. The losses from a battery will probably add to that since they require a 'trickle' to stay fully charged.