Chasing the Power

Every once in a while I get a good dose of greeniness, and look around my little corner of geekiness and sort of wonder “How much juice is this actually using?” After asking Cat what our monthly electric bill was ($300!), I decided this question needed a closer look.
A month or two ago I had picked up a Kill-a-watt (terrible name, ain’t it?) power monitor. This little gadget plugs into a wall outlet, and tells how much power is being used by things drawing through it. Today I jacked it into the single outlet that feeds my nest o machines, and powered things up.
The meter dutifully reported the load as things came online, and steadied out at about 280 watts. All in all, that’s not too bad for 3 computers, 3 lamps, and associated peripherals, but I was curious how that load was distributed. What was actually pulling all that juice?
Unsurprisingly, the single largest draw is yawl, my 2.2gig P4 Linux box. It accounted for about 85watts of power (without monitor). The second biggest draw was, oddly enough, lights. I have 2 compact flourescent desk lamps (about 15watts each), and a single halogen desk lamp (35 watts). I knew the halogen light was pretty dreadful, and this pretty much confirms it. That chalks another 65 or so watts. Which leaves me with 140 unaccounted for.
Well, the two laptops were about 30 watts each (pretty nice considering the horsepower in clipper and hunter). Down to 80 now. This last chunk was pretty much the combined load of the LCD monitors, various chargers and other desktop doodads, a pair of external USB drives, and the like.
So what’s to be done about it? Well, I’ve been considering moving to LED based lights for a while. They’re small, cool, draw -very- little power, but have the current drawback of being ridiculously expensive. A single bulb equivelent to a 100 watt incandescent bulb would cost around $52. The equivelent compact flourescent bulb costs around $5. The advantage to using LED is the current draw is miniscule. For the equivelent amount of light, the bulb would only consume about 2watts of power, AND have the advantage of being dimmable – something impossible with CF bulbs.
If I replaced my 3 desktop lamps with LED lamps, I could cut my power consumption by a third. I also have 4 other lamps in the room that could be replaced as well. The question is, is it worth it?
I’m still puzzling this one out. If anyone has suggestions for good sources for inexpensive LED fixtures and lamps, please let me know!

About

A wandering geek. Toys, shiny things, pursuits and distractions.

View all posts by

6 thoughts on “Chasing the Power

  1. if you’re seriously concerned about how much power you’re drawing, try not running BOINC for a month.
    sure, your computer is still burning some power while it’s idle, but powering up the floating point functional units and chunking workunits draws a heckuva lot more. this point is made rather salient to me by how much more heat gets generated by my laptop when i am/am not running BOINC in the background.
    i can’t tell you exactly how much power you’d save by doing that but i’m genuinely curious to know 🙂

  2. Well one of the experiments I”ll do is see what the change in draw is running BOINC vs not running it. It’s very easy to see the load difference just starting and stopping the client.
    The heavy load from yawl was on bootup – basically as busy as it gets. I’ll try various “doing stuff” vs “not doing stuff” bits to see what the delta is.
    Fun though 🙂

  3. Welll, the monitors aren’t really drawing a lot of current, even when on. Here, lemme test. Currently,the meter is bouncing around 340 watts (I set up another machine – prepping for event 🙂
    Unplugging one of the 17″ monitors changes the load to 310. So the monitors use about 30watts each, and I keep 2 active all the time. They timeout after about a half an hour, and at night I physically turn them off (cuz I generally go right from desk to bed,and don’t want to wait for them to go dark 🙂

Leave a Reply

Your email address will not be published. Required fields are marked *