Help sizing and configuring of emonPi

Hello all,

I’m new to the community and I am trying to wrap my head around what equipment I need to monitor my house’s power consumption.

I have been looking around the site and haven’t been able to find the answer to a couple (basic? in my head at least) questions.

  1. What’s the maximum number of clip-on CT Sensors supported by a single emonPi base station?
  2. Can you hook multiple emonPi base stations together so that you can view their combined data in the same dashboard?

The closest answer I have found for #1 was form the store page of the emonTx that mentioned.

" More than 2x emonTx units can be connected to a single emonPi / emonBase with manual change of RF nodeID. This can be done This can be done via serial node ID config"

However, it doesn’t mention the maximum amount of emonTx units that can be connected.

I am looking to monitor at least 23 (ideally 37) circuits (mix of 30A/20A/15A).

I assume this means I need 23 CT channels connected to an emonPi?
or multiple EmonPi’s?
Is this possible? Or stupid?

I have started to questioning my sizing/understanding of the emonPi when I started looking at the IotaWatt because it supports 14 CT channels right out of the box without any expansion like the emonTx.

Anyways, any help or point to the right documentation would be much appreciated.

of note: I’m in the US and have two Eaton 200A breaker boxes for my main power. I also have a solar array.


Yes you can.

You will struggle with that many circuits. It implies 2 circuits to the emonPi itself, plus 6 or 9 emonTx’s with 4 circuits each, leaving three or one spare. The emonTx’s will normally share the same radio channel to send the data to the emonPi, they are not ‘polled’ nor controlled in any way by a master station, therefore the possibility of two transmitting at the same time and jamming each other is significant.

Any questions about the iotawatt are best addressed to it’s creator, Bob Lemaire at, where he has set up a dedicated forum for IotaWatt, if you post your question there I’m sure he will get back to you quickly.

Thanks Robert!

That really helps clarify everything for me.

I was reading up on some other residential monitors, and it seems a lot of them only use 2 CT sensors on the main incoming power lines to the main breaker box.

Is that main the design intent for the emonPi?
I was originally thinking I could monitor each individual breaker inside the panel.
It seems I might have started under the wrong premise?

One last question if you don’t mind.
I don’t think the emonPi will support all the channels I would like to monitor.
Would you be aware of any other solution that would be worth investigating?


No - in the UK, we have a 240 V single phase TN-C-S system, and the two inputs are primarily intended for the Grid and PV infeeds.

Yes you can, but our system was conceived with whole-house (UK) in mind, so it’s not a good fit for your usage.

EmonCMS is capable - you can run that on any server (including a Raspberry Pi) and feed it data via your LAN. Boredom projects has a design based on the emonTx but using a more powerful processor (still in the Arduino range).

If you find something, let us know. As Robert says, the software part of the system will support many channels, but the measuring part does not easily. I believe (and this is discussed elsewhere recently) that using CTs for measuring small loads can be error prone. - If I am wrong @Robert.Wall will correct me :smile:.

The alternative is to use an invasive system but that requires physical rewiring.

May I ask what you hope to gain by monitoring each circuit? Measuring some devices and picking the big hitters often has significant payback and is relatively simple to do.

No, you are right. This comes up quite often, and there’s a point in the FAQ about it. Every current transformer has increasing errors at “small loads” - it is inherent in the way they work. “Small” of course is relative to the rated current of the c.t. - and the amount of money you paid for it.

A c.t. should always be rated as close as possible to but above the maximum current anticipated.

1 Like

Tell me more… I don’t mind rewiring things. I haven’t run across any pass-through systems, I assume that’s what you’r referring to?

Just seemed more useful. Systems like try and use machine learning to figure out what appliances are turning on and I have read some reviews of people not liking them.

I ran across today.
They have a system that will monitor up to 92 different circuits.
Does anyone have any insight into the company? They are a lot more expensive, but support what I am looking for and seem legit.

I Haven’t done much additional research on them yet, and I’m not sure if it’s possible to pump the data to EmonCMS from their monitoring but I’ll do some additional digging.

Hi Oli,

Are you sure that’s not 32 circuits?

This is from their order page.
The bundle is called System 4-D that has 96 channels.
However, it looks like the smaller one would fit me better.
I’ll list the bullet points below of the two systems.

System 4-D (Donut CTs)

  • Ideal for most large homes.
  • 96 Energy monitoring channels!
  • Includes DashBox Server for Data Logging and Dashboard.
  • Comes with 54 Donut CTs for breakers up to 40A.
  • Add more CTs based on your panel’s requirements.

System 3-D (Donut CTs)

  • Ideal for most medium sized homes.
  • 64 Energy monitoring channels!
  • Includes DashBox Server for Data Logging and Dashboard.
  • Comes with 36 Donut CTs for breakers up to 40A.
  • Add more CTs based on your panel’s requirements.

Something like the Sonoff POW is a switch with builtin power measurment capability but it is only 16A.

Yikes! At 1145 bucks, they’re mighty proud of that bad boy.

You’d save even more if you went with a System-2-D and bought any additional CTs separately.
The “2-D” comes with 18 CTs. Their Micro-40 donut CT is only 5 bucks. Since branch cicuits are
usually rated at 20 Amps, the Micro-40 would be the best choice.

One thing to keep in mind. They recommend 2 CTs for an unbalanced 240 Volt load.
Typically the 120 Volt part of those loads will be a clock or other small device that uses
a minimal amount of energy. i.e. only a few Watts. If you’re OK with a small error in total energy
consumed for those loads, you can use one CT for each of them, vice two, and save a few more bucks.

I don’t have any of the emon hardware, but I have been using emoncms for several years. I have (but no longer use) a Brultech ecm-1240. It is their 2nd iteration of power monitor. The Green Eye Monitor (GEM) is their 3rd generation. When I bought the 1240 many years ago, it was the best reasonably priced solution available. It has RS-232 output and requires something to consume the data. I used a PC and sent the data to Smart Energy Groups and later emoncms. The system worked pretty well and gave me some insight into what was using power. But I am not sure how accurate it was. When main power was out and the house was running on the generator, only the two main channels would show any power usage.

While having 5 extra channels to monitor additional circuits seems like a lot (when the emonpi has zero extra) it turns out to be really hard to really figure out where all that power is going. My brother has a Sense on his house, his feedback is that it doesn’t really provide that insight.

When I first heard about the IoTaWatt, I was interested. I had been looking at the GEM, but was not sure about its accuracy and didn’t like that it was still RS-232. I also didn’t like that it was closed source. The IoTaWatt addressed these issues for me. There are some here that question the accuracy of it, and I am sure for some loads it is not as accurate as some other devices. But, I am not using it to bill anyone for usage. And, it does provide information about utilization that I can use for various purposes.

The Sonoff Pow family are also useful. I have one on my dehumidifier to measure how effective different control strategies are. I don’t believe the POW to be particularly accurate. I believe the POW2 is much better, but still not that great, ie the error is usually significantly greater than 1%.

Bottom line, I am a fan of more channels and even low accuracy (1-10% error for the readings shown) channels can provide useful information.

PS. Each GEM box has 32 channels. You get 96 channels by having three of them. So, buying the 32 channel one and extra CTs won’t get you the same thing. As far as the price goes, what would an emonpi system that can handle 96 channels cost?

Quite right. But the package I referenced is supplied with only 18 CTs.
Hence my comment about buying additional CTs.

Unless one’s home is large, the typical US residence is wired with 200 A service.
(a large home will have 320 or 400 A service)
The largest 200 Amp load centers (circuit breaker box) in common use, have a maximum of 40 positions and some of those positons will likely be occupied by double pole breakers.

The bottom line is 32 channels is very likely to be enough for the majority of users wishing to do monitoring at the circuit level.

For reference, the OP and I are both in the US.

What about the drum motor on a US dryer? Aren’t they usually 120V?

Indeed they are. But when you compare that load to the ~6 kW the heating element uses and the
fact the motor is not heavily loaded… (~15:1 difference in speed between the motor and the drum)

Having used EmonPi for some time on a standard PV setup, i’m looking to add energy storage and extending the monitoring to the energy store connection.

There was a reference above to the ability to utilise multiple EmonPi’s together and use their combined data in a dashboard - in my case this would be extending an existing dashboard to include the energy store feed.

Any pointers to documentation (or even a hint of the direction to take) would be most welcome…my Linux is ever so slightly rusty, and whilst i’ve worked on IoT, that was on the commercial side rather than the technical).

Apologies in advance if this is the wrong thread - thought it easier to mention again where it was originally raised, rather than start something new elsewhere.



Welcome, Chris, to the OEM forum.

The question I’d ask is why a second emonPi instead of an emonTx? Have you considered that option?

And what do you have in mind for your “energy store” and what exactly are you looking to measure? The reason I’m asking is if it’s a battery and inverter/charger, you can only (easily) monitor the a.c. side unless the inverter/charger makes data available and in a format that we can use. There are problems with safety and isolation that are solvable but at significant cost when you start monitoring the d.c. side.

The emonTx will give you 4 current transformer inputs (vs 2 on the emonPi) and if that’s all you’re using it for and you’re using the radio to send the data to your emonPi, it only needs the one a.c. adapter. If the radio isn’t likely to work (too distant) and you’ve got Wi-Fi, you can add the ESP8266 module to an emonTx, but you’ll then need a 5 V USB adapter to power it.

If you do decide to go with a second emonPi, then sending the data to the first is (AFAIK) just a matter of putting the correct URL into the configuration file - just like sending the data to

Thanks for the welcome Robert. Now I have more time available, i’m relishing the thought of getting back to my electronics & computing background. “Rusty” doesn’t cover it though!

In terms of the questions you pose;

1/ i’m looking to monitor the (AC side) in a similar way as the existing grid connection to log energy in/out. Primarily to continue the data gathering i’ve had in place for a while with the existing EmonPi, and also to use as modelling input for considering something like the Octopus Agile wholesale rate tariffs (both for import & export).

2/ the new CT would be mounted in the same distribution board as the existing grid & PV CTs, with the monitoring hardware very close by : there are no distance issues.

3/ I had considered EmonTX : I may have misunderstood, but I think I read somewhere that it transmitted data at (presumably configurable) intervals. Possibly unnecessary but i had rather got used to the real time updating available from the existing EmonPi. maybe I should loose that obsession…



As the two are going to be close together, you could use a serial connection from an emonTx to the emonPi. It won’t affect the way it works, but there have occasionally been issues because the signal was too high. It’s something to bear in mind.

The emonPi by default reports every 5 s, the emonTx every 10. When you look at the sketches in both, they both use the same library routines, and they both function pretty much identically, other than the reporting interval. It’s a very simple change to make the emonTx report every 5 s. They won’t be synchronous, but with something that changes relatively slowly, that shouldn’t be much of an issue.
(I’ve had an emonTx reporting every 0.1 s with the new emonLibCM - but I don’t recommend that, you’d be swamped with data before you knew what had hit you.)

There’s no problem with measuring the a.c. side, inverter etc losses will show up as output falling short of input. You won’t however be able to separate out inverter/charger losses from battery losses.

Thanks Robert I will probably go down that direct connect route…will report back. I have some time to get it sorted before the energy store is installed.

The data from the existing EmonPi setup was invaluable in modelling the sizing of the store, but be nice to move away from the Excel…