Best approach for 4 mains circuits, 1 PV inverter

So I want to monitor my house’s energy usage and PV production. It appears I have 4 circuits: lights, air conditioner, power 1, power 2. I would definitely like to have visibility of the air con consumption; lights would be nice to have, but not vital. I’m planning on using an emonPi (or emon base) in the house and a TX in the meter box.

Seeing as the TX is limited to 4 CT clamps, what is my best approach?

Do I have to put a clamp on each circuit and the PV (requiring 5 clamps/2 TXs)?

Or, can I put one on the whole house, one on the PV and one on the air con circuit (and maybe one on the lights circuit)?

Other info: Australia, 240v, single phase

I think you could get away with:

  1. whole house net (i.e. feed from street)
  2. PV
  3. aircon
  4. lights

Provided you also use a VT (AC wall-wart for voltage reference) then the first one will be signed (i.e. you’ll know whether you’re currently importing or exporting). Then I think you can do all the remaining maths in emoncms. The only thing you wouldn’t be measuring directly is power1 and power2, but you could calculate their total in emoncms from the others.

Another trick you can consider… if your power1 and power2 breakers sit next to each other on the DIN rail, it’s usually pretty easy to wrap one of your CTs around both of their wires. Then you’d be measuring power1+power2 with that CT… you might consider that instead of lights, in which case you’d then calculate lights from the others. In theory you’d get the same answer either way, subject to accuracy constraints.

Thanks! Yeah, I’d wondered about the 2 wires/1 clamp idea myself. Either way, good to know that emon cms can work with the whole house/sub circuits approach.

And there’s a handy AC socket nearby for the VT, so I was planning on that anyway.

There was a third option I was wondering about though… the TX would use its 4 CT clamps on the air con, lights, power 1 and power 2, and I could run a 5th CT from my emonPi in the house (with a long cable). In this scenario, would both the Pi and TX need their own AC-AC adaptors for voltage reference, or would just the TX one do?

I’m not sure about that one, but hopefully others here will know the answer.

Yes - the multiplication of current and voltage to get real power is done 50 times per cycle in the front end processing. But if you’re happy with apparent power, then no, you don’t need the second a.c. adapter. There’s little or no difference between real and apparent power if you can guarantee a load that’s always close to unity power factor, e.g. only things like immersion or storage heaters, but otherwise the discrepancy could be large (my washing machine on the spin cycle: power factor ≈ 0.3, meaning apparent power is more than 3 times real power).

On the question of how many c.t’s, does your PV have its own meter - and does it have a flashing LED? If so, you could use the 4 c.t’s for the outgoing circuits and count pulses for the PV, and you’d know the energy generated that way (but current and power only by inference). Pulse counting is by definition historical, so you wouldn’t have an immediate display of the true nett power - you’d only be able to know nett energy transferred over the period from the last pulse to the one that’s just arrived.

Good to know. Thanks Robert!

Ah… I guess I could have answered that. I interpreted the question slightly differently. I thought the OP was proposing splicing the output of the VT near TX and running that (alongside the 5th CT) back to the emonPi, so they’d both hang off the same VT. I didn’t feel confident answering that interpretation because I had some vague recollection of there being loading constraints, and GND loops issues.

In any case, it sounds like your interpretation was the correct one.

I didn’t think that forking the a.c. adapter voltage was the intention. It would be possible but, as you suggest, it would need careful checks for isolation because of the possibility of a ground loop if ever other things were connected simultaneously to the emonTx and the emonPi. And the danger there would be in the future when the details had been forgotten. Loading would only be a problem in the sense that the emonPi would see the distorted voltage wave that the emonTx uses - the emonPi doesn’t draw power from the a.c. adapter like the emonTx does (when it doesn’t use the 5 V USB supply). And that feed would need screening, but nothing special as the outer of the connector is GND at both ends (unlike the c.t. where the plug sleeve is actually the input signal!).