I meant the would to be emphasised. I was agreeing with you.
1 Like
Sorry - not sure what the question is ?
Yes if you do not want to give up inputs 13 and 14 they can be designated as Voltage references for additional phases (this is how i have one of mine setup) They use the same power packs (but 3 of them) plugged into 3 seperate power points that need to be wired to the 3 phases.
On my other 2 IOTAWATT units i use a derived voltage reference as i know that for about 9 months of the year all of my phases are within 1 volt of each other - they only differ markedly in summer when there is a lot of solar being produced in our area and not all phases are evenly balanced.
[EDIT] - and in response to other posts re PF - my loads are deliberately pretty well balanced across the phases in terms of PF - we do (obviously) see some discrepancies from time to time - particularly with our pool pump which is single phase only
Craig
1 Like
Ah OK, you gave the impression above that you were using derived voltage reference for all off your 3 iotawatt units in your post above-
But if you are using three voltage references in at least one of your iotawatt's Craig, then yes, we are on the same page 
Yep my main unit that monitors my Solar, Grid and Batteries is gettng the voltages from all 3 phases using discrete power packs.
Craig
Which hardware do you actually use for measuring the 3 voltages and how do you connect it ?
What are discrete power packs ?
As per my post above - commonly known as wall warts - essentially a mains level transformer
In the case of each country - the developer of IOTAWATT has specified a number of different sources for these plug in transformers - they produce a known output voltage for a given input voltage and the IOTAWATT uses these as the Grid voltage reference on each segment
Craig
Ok, learned a lot so far 
So I definately want to measure both current and voltage, to have accurate power calculations.
The Iotawatt looks very nice, but not cheap since I need the generic kit + 11 CT clamps + adapter + voltage transformers. When I combine the prices of all these parts, a solution based on Shelly EM seems only a bit more expensive:
This seems also to be a nice device for my use case:
- It is very compact and easy to install, since I don't need any extra transformers, ...
- It would fit very nicely in my electricity closet, when installed with the din-rail clips (I mentioned above in this discussion).
- I would have an all-shelly solution (which improves the maintenance of my home setup), since I also use currently a series of Shelly Plug S devices to measure power of some devices already (simply via the wall outlets).
- It is very modular. I can start with a couple of those devices, and buy new ones when I want to extend my system.
- I can easily get the data into Node-RED via MQTT, which means I can build my own logic in Node-RED (without spending too much time on the hardware). From my hobby point of view, it would be awesome to build some node(s) to integrate this nicely into Node-RED (without needing third-party tools).
I just need to make sure I have wifi available in this area...
How are you going to measure 3 phase power with a Shelly EM??
Are you just intending to measure 1 phase, and derive the other phases from that?
...or just measure a circuit that is fed by one phase?
If I buy minimal 3 of those devices, and I connect each one to a separate phase. Then I can measure two circuits on each phase. For example the total consumption of that phase, and one of the circuits on that phase. Or have I something overlooked perhaps?
Because in areas where I have wifi, I could install Shelly plug S devices in the wall outlets to measure the power consumption of the attached electrical device. Then the plug S can also be used to turn the attached electrical device on or off (as a bonus), to make my house smarter.
Then I only need to buy Shelly EM devices and install those in my electricity closet, to measure the power of electrical devices that cannot be measured via a Shelly plug S:
- Electrical devices that are not plugged into a wall outlet. For example LED spots that are wired directly to my electricity closet.
- Electrical devices that are connected to a wall outlet that is not easy to access, e.g. the freezer that is built-in into my kitchen (so I cannot easily reach the wall outlet behind the freezer).
Does this make sense? Of course I might need to extend my wifi network...
Hhmm, so 3 shelly EM's each measuring the power usage of 1 phase each.
As each shelly will update it's readings individually (instead of all 3 synchronously in a Shelly 3EM), it will be interesting to see how you calculate the totals.
I will try to visualize a bit of what I mean, to make it easier to discuss:
Since I have 3 phases in my house, I would repeat the above setup for each of those 3 phases...
Ok, now I am lost (again) 
The majority of the houses in Belgium only have 1 phase, so they connect all their devices on that single phase. In my house there are 3 phases, so we have tried to distribute the load of the electrical devices as good as possible across those phases. And some devices use the 3 phases.
Can't I simply measure the power on each phase separately, e.g. to display in a chart. And when I want to see my total consumption, I add all those 3 consumptions. Based on your feedback, I asume I am telling nonsense
Yes, it would be good to measure the power of each circuit of your home, but combining them may not be as easy as using a join node.
The timings on the 3 consumptions will arrive at different times, so the chart will not display consistent values, maybe A,B,C, or B,C,A or C,A,B. Each addition will give a different result.
Maybe just academic, but your totals will surely fluctuate.
Hmm you are right. Damn you Paul 
I don't what result the Shelly EM returns: an ever increasing accumulative power consumption value (so I need to calculate the delta since previous reading myself), or the power consumption of the last N seconds, or...
And I also don't know how often a message is send by Shelly EM via MQTT. Or perhaps it is better if I send a http request to all Shelly EM devices myself, so I can more easily aggregate those consumptions?
Need to analyze further...
But my time is up for tonight 
Note that the Voltage wire won't measure the voltage but is for controlling the contactor (= "zekering" in dutch) so you can automatically turn off the circuit. I am not sure if standard contactors support this option (at least mine don't look like the one in the above diagram and should be manually switched on/off )
In order to measure the voltage you need to connect 2 wires from the L and N port of your Shelly to the L and N power line of the phase you want to measure !
Hi @janvda,
Thanks for the update!! I only had a quick look last night, and it seems indeed you are right. Will need to update my diagram...
@Paul-Reed,
I see this in the api documentation of the Shelly EM:
When configured for MQTT Shelly EM reports data on:
shellies/shellyem-/emeter//energy energy counter in Watt-minute
shellies/shellyem-/emeter//returned_energy energy returned to the grid in Watt-minute
shellies/shellyem-/emeter//total total energy in Wh (accumulated in device's non-volatile memory)
shellies/shellyem-/emeter//total_returned total energy returned to the grid in Wh (accumulated in device's non-volatile memory)
shellies/shellyem-/emeter//power instantaneous active power in Watts
shellies/shellyem-/emeter//reactive_power instantaneous reactive power in Watts
shellies/shellyem-/emeter//voltage grid voltage in Volts
shellies/shellyem-/relay/0 reports status: on, off or overpower
When I should measure the total value every minute, and you subtract the previous total value from it (stored e.g. in flow memory). Then you have the power of the last minute? Or do you see something incorrect in my reasoning?
Then you would have the energy used in the last minute.
Power is the instantaneous rate of usage and measured in Watts.
So in your example, keeping track of energy usage (Watt-minutes) is ok, but because power is an instant reading, combining them at a point in time is a bit more problematic as it needs to be at short intervals to give the full picture.
By example, Colin plots his power every 1.3 seconds, I plot mine every 5 seconds.
(Sorry rushing & replying by phone)
Bart,
I think for a 3 phase house - assuming it is a normal one - you would want a a Shelly 3EM rather than the EM - it is designed for 3phase and will give you the readings across all 3 phases at the same time.
Craig
Yep it gets harder @BartButenaers when you start delving into it.
On the iotawatt i poll its api for the average every 3 seconds (the iotawatt does the averaging for me as part of its sampling and storage) to enable my battery and solar control - it is not perfect but over the course of a day i am out less than 1% - and this is in times of sudden spikes (like kettle being turned on_ and this is more because the battery is slower to respond
Craig
2 Likes
Hey @Paul-Reed, @craigcurtin ,
trying to be on the same "wavelength" as you guys 
I tought that by narrowing down the energy measurement interval, you would achieve more and more an accurate 'approximation' of the power? But indeed if the interval becomes very short (like in the case of @Colin), it would become difficult to join the corresponding messages.
Is it so important to have the power instead of the energy? Is that perhaps for measuring peaks for an electricity bill? Although I had expected that even for peak billing, the peak would be measured across a certain time interval...
And measuring power every second seems a hell lot of data to handle in Node-RED for a simple home automation. Never thought people would use such high measurement frequenties for power consumption...
Let's assume that for some reason power is more important than energy, then I assume this can only be solved - in a full Shelly setup - by using:
- One Shelly 3 EM to measure the total power/energy.
- Use one or more extra Shelly EM devices to measure the power/energy of separate circuits on a single phase (of the 3 phases). Because for devices attached to a single phase, there won't be a joining issue?
Is that correct, or still not?
Had a quick view, but don't see how often Shelly pushes messages over mqtt. Not sure whether short push intervals - e.g. 1 second - are possible (if that should be required)...
Another thing that bothers me, is that I need a lot of extra voltage wires (beside the current wires for the clamps). Using thick stiff wires is not easy to install into an existing closet full of din-rail stuff. Does anybody has any idea about how thick these wires should be? I hope the wires can be pretty thing, because the load is not high?
Not sure what you mean by that.
