Apparent vs Real power

I’ve just moved to a (UK) house with no indoor access to the 100A cable in. I’ve brought my old emonPi (bought in 2015) with me. I don’t have solar yet, but am keen to gather usage information.

The only place I can measure my usage is outdoors, in a cabinet built into the external wall of the house, which houses the electricity meter. There’s no 13A socket there, and I am reluctant to add one given it is exposed to the elements (the cabinet is not sealed).

One option is to keep the emonPi in the cabinet, powered with a long USB cable from the garage. Easily done, but wouldn’t have an AC reference from the AC adapter so I wouldn’t get real power measurements.

The other option is to keep the emonPi in the garage and extend the CT clip (from 1 metre to about 10). I’d be able to use the AC adapter to get real power this way.

Is anyone able to advise what difference it makes to measure apparent vs real power for a household without solar (where current direction is known) but with some non-resistive load? I am making my way through the “Learn” section of the OEM website and there is a great section on AC power theory, but it does not express an opinion on which measure is more useful and whether apparent power is a good enough approximation for a typical household.

Basically: if I only measure apparent power, is the emoncms accumulated kWh value likely to be within 10% of my electricity bill? Many TIA.

There are so many variables that’s it hard to predict and probably not valid to extrapolate from one house to the next, but for what it’s worth, here’s how my whole-house CT measures up in the wee hours when everything is off:

So instead of reading 150W throughout the night, you’d be using 284W, or about a 90% error throughout the night (actually it improves when the fridge kicks on because it adds a bit of inductance and lifts the PF up into the 70s).

But how much impact will a 90% overnight error have on your total bill? That depends a lot on your personal “duck curve”. If 85% of your energy is consumed during the evening peak (when PF is likely a lot closer to 1) then you’re only talking a 90% error in 15% of your bill.

Welcome to OEM, Chris.

I wouldn’t like to say how accurate, i.e. what the overall difference between real and “nominal” apparent power would be, because there are just too many variables, as @dBC says. It will depend on the actual system voltage, which varies almost from minute to minute and day to day, and your power factor will depend on what you’re using from minute to minute.

If you want real power, I think there are three options:
First, the one you’ve mentioned, which is extend the c.t. cable. 10 m shouldn’t be a problem unless you’re running it alongside mains cables with nasty switching spikes on them.
The second is extend the a.c. adapter cable. The current is effectively zero, so volts drop isn’t a problem. It should be less susceptible to interference from adjacent cables, but doing it after the same fashion as the c.t. (see ‘Learn’ for the details) would be a wise precaution.
The third is a bit more “geeky” - it’s been tried but not by me - and it is to use a capacitive pick-up on the line meter tail. It is in the archived forum: Measuring real and reactive power WITHOUT measuring voltage on EmonTX | Archived Forum and Solar Power Divert with no voltage readings | Archived Forum
If the circuit diagram isn’t there, I can supply it. As it stands, your emonPi would require some modification to the analogue input circuitry.

I am guessing the consumer unit is mounted on the inside of the wall where the meter is and the tail to the consumer unit comes through from the rear (possibly run in the cavity)?

This is the option most folk use. Could you extend a socket (add a spur) to be closer to the point the CT cable enters the garage?

Here’s a pic of that:

and even that’s not particularly helpful without knowing how much power was flowing at the time. You can see during the evening peak it settled down a bit and was mostly above 80%.

My guess (and it is only that) is it’s close to unity when you have a decent size heating load, and very poor at light load.

But as I inferred - perhaps I should have spelled it out - without the a.c. adapter, the emonPi assumes a constant mains voltage of 230 V, which is wrong for the UK ‘centre’ or aiming voltage anyway, even before the effect of loading are taken into account.

Thank you all :slight_smile:

Yep - I’m sure the person who designed it thought they were being helpful…!

Very interesting to see the variation in power factor visually, although I take the point that it’s only a multiplier and lacks weighting. In any case, it is clear that apparent power will not be a good enough approximation, so I will need to extend more than simply power over USB.

Provided it’s fine to run the power supply and take an AC reference using an ordinary extension cable, I can move the emonPi to the garage door and have only 5 metres to travel.

That’s an interesting thought. I like the sound of this more than extending the CT sensor if it is less susceptible to interference. Would this work?

Fascinating stuff but sadly far beyond my dexterity level. In fact, it’s a shame there isn’t an off-the-shelf CT sensor with a long cable. If anyone here would be happy to construct and send an extended CT cable for £30, please let me know!

Another quick question - is there a way to know for sure whether the emonPi is using its AC input? I performed a “manual override” on my common sense and ran a 13A extension to the cabinet so I could see what real power measurements looked like, but the emonPi still shows a constant 230 Vrms. Could this be true of Milton Keynes’ supply? I used to live in Kent and saw ~250 Vrms most of the time. I don’t trust the 230V and am wondering if there’s a log somewhere that would tell me the pi didn’t detect the AC.

I don’t think there is any reason not to. If you can put a hole through the wall from the external cabinet to inside the garage, you can thread the CT cable through that (I usually sleeve holes with a bit of round conduit - makes threading cable easier and also you can seal the hole with a bit of silicone). If they have routed the tail in the cavity (you really shouldn’t) be careful drilling the hole.

If it is a meter cabinet that is sunk into the wall, there is very little between that and the inside of the house!

Fix the extension cable to the wall and all is done.

It’s not ideal. For best results, you need two cores screened overall (not two single-core screened cables).
Note that the screen should be earthed at one end, neither of the cores are earthed (they sit approx. 1.65 V d.c. above earth).

Did you connect the a.c. adapter before you powered the emonPi with its 5 V d.c? Because it only checks for the presence of the a.c. at power-up, and if it doesn’t see it, presumes it’s not there and uses the fixed 230 V value. There’s no log entry for that - the decision is made in the analogue front end, long before the numbers get to the Pi part.

Turning the pi off and on did it: “AC wave detected” appeared on the LCD at startup and the voltage is up at 246.8. I had power-cycled the adapter+pi together twice before (switching off and on the extension to which they are both connected), no joy, but this time just cycled the pi with the adapter left on. Perhaps the AC determination is made before the adapter starts to deliver a wave? (seems unlikely…)

Thank you - will try that (presumably 1.65V DC refers to the CT sensor cable, not AC adapter). Is earthing the screen strictly necessary?

No, it applies to both. If you don’t earth the screen, there’s no point having it, and you might as well use two un-screened twisted cores instead. Whether you actually need the screen depends on how much interference is floating through the air where the cable runs. If it’s a tricky cable run, it’s safer to get and run in a screened cable first time, than have to pull it out and replace later.

(No power is taken by the emonPi from the a.c adapter, it is only used to measure the system voltage.)