Archived Forum

You could try each circuit breaker in turn, until they're all off - that might identify which (lighting, ring mains, kitchen, boiler) is causing the drain?

Yes, I have tried a few, unfortunately I guess I then need to move my setup to a fully portable setup so i can also switch of the power sockets circuit breakers.

You don't say what your system voltage is, so I don't know what 100 W means in terms of current, nor do I know what your maximum current is, nor whether you are reading real or apparent power.

With that background, take a look here:
http://openenergymonitor.org/emon/buildingblocks/measurement-implication...
http://openenergymonitor.org/emon/emontx/accuracy
http://openenergymonitor.org/emon/node/1385
http://openenergymonitor.org/emon/node/2311

I have yet to run a full set of tests on an emonTx V3 - all the above links refer to the V2. The V3 incorporates some design improvements that should make it perform better at low current, and my initial impression is that is indeed the case.

It may well be that you do not actually have any unexplained loads, and the "power" you are reading is noise pickup.
 

As Robert said, it is very important to establish your baseline reference. People have differing view points on how to do this. I'll offer what I did and say that it seems to correlate pretty well with other independent measurements that I have made in my house with more accurate instrumentation.

The first thing that I did was to ensure that the mains supply voltage is being measured correctly. Voltage is a more tangible thing to measure than current. Assuming that you are using a voltage transformer (perhaps the one available through the shop) you need to connect this to a socket which is close to your consumer unit. Try to avoid connecting it on a multi way extension lead. It's always better to connect it directly to a wall socket near the consumer unit. It's amazing how the sensed voltage drops when the voltage monitor is connected to a poor performing trailing socket which is already power a number of energy hungry devices because your ring mail has a much lower contact resistance than a load of daisy chained trailing cables. Really you need to verify what you are reading using your sketch with an independent  instrument. A reasonable cost multi meter from Maplin will suffice. Equally, and probably better (even if you have a meter) is to buy a plug in energy monitor for about £20. You need one which can measure volts and true power

http://www.maplin.co.uk/p/15a-plug-in-energy-saving-monitor-l61aq

the above is the one that I have.

you need to tweak your voltage calibration in the sketch so that the voltage measured lines up with you meter. In my sketch I print the voltage measured as well as the power and I even send this data to my GLCD display constantly. It's amazing in our house how, even with a properly sited voltage sensor, the house load affects the voltage measured (our ovens are electric).

having got this within a percent or so, you can now attack the current measurement. Many of us, myself included, use the coiled wire and lamp approach. I have a 100w bulb attached to a light fitting and this is fed via a long length of multi core wire. I have this wire looped 1, 5, 7, 10 and 25 times. The wire needs to be thin so that the current clamp can fit around it. I then use the plug in monitor to establish the base power that the bulb is taking and I attach the current clamp to various coils so as to establish the perception of increasing load. With my rig I can easily simulate about 4KW of power usage. It's then a case of tweaking the current calibration so as to correlate what the sketch measures with the load power from the meter geared by the number of wire loops inside the current clamp.

as Robert said, there is non linearity with low power and my V2 units suffer from this. I have accepted that readings become inaccurate below a few 100w. However, the base load in our house is about 400W anyway so I can live with this.

generally speaking, as power increases the load in many households becomes more resistive. It's often down to ovens, electric fires, kettles, coffee machines, etc. When operating at base power it's often standby loads, trickle chargers and such like which consume the power. These tend to have complicated power profiles and don't make power monitoring any easier with the emon units.  I have chosen to accept that it's a compromise in measuring at low power and have tweaked my immersion dump software so as to give me a decent safety margin (80W or so). 

One final thing for my rigs: I also have an independent pulse monitoring system which I have correlated very accurately against the network operator metering equipment and my generation meter over a period of several months. I can't tweak this equipment but it stands up well. I have found that that, over extended periods, my emonTX measures to within 0.5% of my pulse monitoring on the solar  channel. It's more difficult on the grid channel because the pulse equipment cannot measure export. However, point visualisations under a wide variety of load conditions suggest to me that it is in the same ball park as the solar channel.

If, having got everything as good as you can, you still find a huge low power error, as has been said, you need to work your way around trying to find out where the power is being consumed. If you have a friendly neighbour and a long extension lead, borrow a socket from his house, connect up your gear on the end of a cable reel, turn off the power at the consumer unit and monitor your rig. That should give you an indication of the offset error. 

You should be able to isolate most of the rings in your house aside from one and get a pretty good idea where the power is being consumed. Could it, for example, be a fridge left connected or something in the garage or shed.?

Hello,

Thanks for te replies, this is insightful stuff, and I am very newbie on this. I did a full test where only only plug was left connected which had my internet modem (15W), wireless ap (10W), network switch (25W) and router (10W), The lowest reading i got last night was 80W, which gives us about 80-35W = 45W (i am discounting the switch as it was on ups during the test). 

My measurement was done directly under the meter, I have 2 medium size cables that seem to report the same load regardless of which one I clamp the sensor to. I can send a picture if that helps. I live in Ireland and am using an old meter from Electric Ireland (ESB). We get 230V/50hz over here. Very similar to the UK I believe. 

I am slightly considering getting a digital clamp to measure stuff but heard quality varies widely depending on price.

So I guess 45W is something I can live with. I have measured a lot of equipment in the house and correct reading is achieved, my questions is about these 50W remaining. This is more me not trusting myself to have done the setup right. 

I have no garage and nothing else running apart from the aforementioned equipment when doing the tests. This is a picture of the test i did last night.