Computing stuff tied to the physical world

Power consumption

In Software on Mar 12, 2011 at 00:01

I’ve been working on some new software to better handle all the messages flying around the house. Took down the old system, and started collecting stuff with the new framework.

First one that always interests me, is the baseline power consumption. So I grabbed some values from last night:

Power 11 03 2011

That’s about 450 watts when all the lights are out and nothing is happening – shocking!

Uh, oh – 4000 kWh of vampire power on a yearly basis. Either my new calculations are way off, or some serious stuff has crept in here at Jee Labs which is drawing lots of power!

Update – Yeah, I was off by a factor of 4 in the low range. Forgot that I had added some fancy bit packing to get more resolution across in the metering sketch:

Screen Shot 2011 03 12 at 11.58.23

The lows are around 110 W. With the 24h “internet presence” (server + router + fiber modem) taking an estimated 30 W. Now that makes sense again!

  1. So the spike at 05:30 would be you making a cup of the Dutch version of Horlicks before bed, 08:15 would be the cup of coffee when you wake up, and 09:00 would be the second cup of coffee to back up the first? :-D

    I’m too frightened to do this on my electric meter, although these days it should be somewhat better now my server is Atom powered instead of the 2.8Ghz P4 beasty I had before… Maybe I will, but I’ll have to build something to watch the magic metal disk go round first!

    • Heh :) – the 5:15 and 9:25 spikes are the kitchen boiler. The 8:15 spike is the shower microwave. I’m guessing that the 7:00 and 9:30 bumps are the fridge.

      The only trouble I have with watching the metal disk, is that at low power levels it takes a lot of time for a single rotation. It’s actually a Heisenberg disk: either you know how much you consume, or you know how long you’re not consuming much!

  2. Indeed, it does turn rather slowly. I guess measurement would depend on recording “time since last dot” instead of “count the dots”.

    Pity I can’t take it apart and add some more dots to the disk.

    Do you have any recommendations for low-cost devices/modules which could be used with a JeeNode to measure the current via an inductive clamp arrangement?

    • measure the current via an inductive clamp arrangement

      There’s a bunch here: – haven’t yet looked into what it takes to do it all, though. For accurate measurement which takes inductive loads into account, you need to rapidly integrate over voltage * current. There are chips for that – hmm, I wonder if these could be used with a current clamp (for isolation) iso direct connection.

  3. Currently, I am counting the dots on the silver disk with a reflex sensor attached to the glas of the counting device (Stromzähler in german :-) ). But you are right – if you only consume low power, the disk turns to slowly to get usefull data. You may have a look at but it’s only german for more detail.

    I am planing to replace the “normal” arduino with a JeeNode soon and maybe a JeeLink for the receiving part…

    • Thanks for the link. Yes, watching the actual disk spin would be fantastic. Other suggestions made so far are: shine a laser beam off tangential to the disk so the little ridges get amplified. Or perhaps even re-purpose an optical mouse sensor, but that needs some optics to get to the disk since such mouse sensors are usually designed for 1..2 mm distance.

      If all else fails, a current clamp could be used for the low values. Still, counting disk revs (or pulses) is essential, since it’s accurate over the long term – regardless of current readings.

  4. how woule you current clamp your house installation? You will need only one of the 230VAC conductors going into the mains distribution, without tampering with it???

    • I must admit I’m a little puzzled by this. I’ve seen the clamps going round the combined, armoured live and neutral cables, and seen them work, but I can’t quite understand how! With the current equal and flowing in opposite directions in the feed and return, how can there be a resultant induced in the clamp to read? :-/

      Obviously there’s something I’ve either missed, or don’t understand!

    • That’s the tricky bit. Yes, as far as I know, you need to place the clamp around single wires (and place it fairly accurately perpendicular to the wire) to get useful readings.

  5. Going to ask for a quote later this week to install 1-fase kwh meter per circuit with S0 bus here in the house. Trying to figure out where my 13.000 kWh comes from (or goes to?)

    Just reading the total consuption is not enough I fear .. and using plugwise modules everywhere is even more expensive.

    And the extra cost of installing these meters might be earned back when I reduce my usage enough :-)


Comments are closed.