EmonPi PV monitor three-phase

Can the Energy Monitor be used for a 3 Phase PV system ?

Welcome, Wayne, to the OEM forum.

Unfortunately, the emonPi has only one voltage and two current inputs, so on its own it cannot accurately measure three phases. You would need to add one or two emonTx’s in order to do that.

However, the next version of the emonTx will have this capability, with 3 voltage and 6 current inputs - expandable to 12 current inputs. I understand advanced prototypes are now being tested, but there are no firm dates for availability as yet.

1 Like

I have a related question - just visiting someone who has 3 phase supply to the house and we’d like to install an emonPi to monitor usage. We only need the total of the 3 phases at this stage, we don’t need to separately measure each phase (we might add that when emonTX comes back into stock).

It looks like a 4 wire system, there are 3 red wires and 1 black wire into and out of the meter. Presumably the black is a common neutral wire? In that case, if we put the CT on that common neutral, will that give us the total current for all 3 phases?



That’s very likely. The cables should be marked L1, L2, L3 and N, although as you have the old colours, the installation probably predates this practice. Done properly, the 3 reds should be red, yellow & blue.

No. It will give you the sum, but the vector sum. If the loads on the three phases are balanced, which they should be but never will be in practice, the neutral current should be zero.

The only way is to do it properly - measure each phase separately and add the three average powers. You’d be interested in the emonTx V4: AVR-DB: emonTx V4, new hardware in progress - #89 by glyn.hudson

Very likely it predates current practice, the 3 phase is there purely for a lift, which was put in early eighties. If the rest of the house runs off one phase, it is probably only that one we are interested in monitoring. But there are serveal consumers units around the properity, so I’m not sure yet if they are all on the same phase or on different ones.


That’s a very interesting thread! I’ll certainly be interested in that when it is available and also happy to assist with Beta testing when it gets to that stage.

The revelation that there is an alternative AVR processor you can actually buy without waiting a year was also very interesting, and likely relevant to another unrelated project I am looking at.



If the 3-phase supply is solely there for the lift, then you don’t need 3-phase monitoring, because it’s reasonable to assume the lift motor is a balanced load. All you need to do then is monitor it on its own, and multiply by three. If other C.U’s are indeed spread around all 3 phases, which would be a better way to do it (but if they predate the 3-phase supply, it’s unlikely they will have been moved) then you still must measure each phase, because they will never balance exactly.

Further investigation has confirmed that each of the 3 consumer units are on one of the 3 phases, so to get the full picture, we will need to monitor all 3 phases. However, one of the consumer units feeds just the bedrooms, which don’t have a lot of current draw, so for the moment we can just monitor the other 2 phases using the EmonPi. Although all 3 live feeds are red, they are labelled with tape to show the red, yellow and blue phases.

So I’ve installed the EmonPi with the AC and DC mains adapters and configured it.

I created feeds for log and KWh for Power1, Power2 and power1pluspower2 and I’ve created 3 instances of the “My Electric” app on Power1, Power2 and Power1plusPower2 so we can see the power on each of the 2 phases we are monitoring and the total of them both.

This all appears to work fine, but it is hugely under-reading on both phases.

If I turn on charging the EV via the granny charger, which the car reports as 2.2KW, and my EmonPi at home reports the same when I plug it in the same way there, but here the power only goes up by 1KW in the My Electric app. If i put a 3KW kettle on, it only goes up by 1.6KW. There is a computer cabinet here, its UPS reports 270W load, if I turn off the input feed, so the rest of the cabinet is now powered by the UPS, the power reported by EmonPi only drops about 100W.

I have tried the following:

  1. Swap the CTs over to each other’s phases - readings still the same as they were for each phase
  2. Swap the CT connectors over at the EmonPi - readings still the same as they were for each phase
  3. Put both CTs on the same phase, both read the same within 1% or so

I’m not sure I have determined an accurate enough figure to change the scaling factor in the config file.

Does anyone have any clue as to why it might be under-reading by so much?
What should I try next?

Any help greatly appreciated.



Yes, phase. The software is for a single phase, and it’s calculating real power on the assumption that the a.c. adapter and the two c.t’s are all on the same phase, but from the sound of it, they are all on different phases.

If you do step 4, put both c.t’s and the a.c. adapter all on the same phase, I think you will get correct readings.

I think, with a unity power factor load, you’ll be seeing exactly 50% of the true reading. For a reactive load, it will read higher or lower depending on both the phases involved and the load’s power factor. Unfortunately, there cannot be a 3-phase sketch for the emonPi - it can’t run the radio at the same time, and there isn’t an easy answer. You could modify the front-end “emon” sketch where the calculations are done to use current and voltage to report apparent power, but you run the risk of that being overwritten if the emonPi decides to update itself. The best suggestion I have is to sit tight and wait for the emonTx V4.

Thanks for the explanation - at last I have some idea what is going on.

I think you are right, wait for emonTx V4, then we can do it properly.

In the meantime, based on your explanation that, depending on how reactive the loads are, it will be giving something around 50% of the correct value, I have edited the config and put a scaling factor of 2 for those 3 inputs, so at least it is givnig us some idea of the correct readings until we can do it properly.

The figures we are now seeing do look a bit more sensible now.