A small project needing some advice/confirmations, please!

I have a small project. I am no beginner to electricity or monitoring (among other things I design datacentres), but I am new to the emon range. What I would like to do is to monitor each individual circuit on my consumer unit. I have 10 MCBs on a test circuit and I would like to monitor the load through each circuit, as well as one on the incoming feed for reference and calibration purposes.
Obviously the first challenge is space. A MCB is some 18mm wide and most CT sensors are way bigger than that. Plus the CT sensor needs to fit behind the plastic cover. Then I need to ensure that the low voltage wires from the CT sensor can pass safely through the consumer unit without either getting any crosstalk from the high voltage cables, and are insulated from them as well.
Unless anyone has done this already, I am not sure if there will be any crosstalk-type effects of having so many CT sensors so close to one another, so this is another part of the experiment.
So assuming I have found an appropriate CT sensor and have managed to safely fit them and pass the LV wires out of the consumer unit, I have the following questions:

There doesn’t seem to be a simple guide to explain the differences and the advantages/disadvantages of using either the emonPi or the emonBase. What I have found seems to indicate they do pretty much the same job, so could someone much more knowledgeable than I, explain which is going to be the best for my project?
The emonTx has 4 ports. 1 seems to be for an extra sensitive probe, but I assume that it can it be used with a normal CT sensor, as long as the current on that circuit is less than 18.8A
The CT sensors available are too big for my project, so I assume that these will do an adequate job;
Reading up I can see that I can connect more than 2 emonTx units to an emonBase unit, as long as I do some manual config work. I assume I can connect more than 2 to an emonPi if this is deemed a better option for my project?
My project would therefore seem to require 3 x emonTx units? 10xMCBs and 1 x incoming feed. All three emonTx units and the (emonPi or emonBase) will be within a few centimeters of each other.
With regards to the AC-AC PSU, would I need one for each emonTx, or could I use one PSU and split it to run 3 emonTx units? Or if I use one PSU connected to just one of the emonTx units, will that give me the reference that I can use across all 3 once I get the data into a database? OR, if I have an emonPi, can I use the PSU to connect to that and not fit one to any of the emonTx units?
Other than a programming cable to configure the emonTx units, have I missed anything?
Sorry if these questions have been asked before, but I have spent quite a lot of time reading through the various areas and the manuals but haven’t managed to answer these specific questions (although many others have been answered….)
I also plan to try out the IoTaWatt system as well, although I’m not a fan of WiFi, I prefer hard wired connections, mostly because there are too many bored teenagers trying to hack every WiFi and Bluetooth device they can find.

Welcome, Ashley, to the OEM forum.

There are so many points to answer, please forgive me if I miss one or two.

Are you looking to buy “finished” units, or would you be prepared to build something? The reason I mention that is there’s project that heavily borrows from OEM which uses a high-powered Arduino and a “DIY” front end for the analogue conditioning circuitry. It’s
It has many more channels than an emonTx, but I doubt it can successfully run “continuous monitoring” as the emonTx does.

With our standard split-core YHDC c.t’s, there is a tiny amount of crosstalk. I have never tested ring-core ones, but I’d certainly expect it to not be worse.
With so many c.t’s in a normal C.U, you must go for the smallest ring-core one that will do the job. And yes, those RS ones will be suitable.

An emonPi is a Raspberry Pi with a two-channel emonTx attached, in a nice aluminium case with a LCD display. An emonBase is a Raspberry Pi with a radio attached, but no facility for voltage or current monitoring, no case, no display. And one you might not have spotted, A Raspberry Pi Shield is a Raspberry Pi with only the two-channel emonTx attached, nothing else.

The only difference is the value of the burden resistor. If you add a second burden resistor in parallel with the fitted SMD one - there are holes ready - you can make it what you like above that. As your biggest ring-core c.t. is likely to be 50 A and all require a 33Ω burden, I’d advise removing all the fitted SMD burden resistors and replacing them with 1% (or better) 33 Ω wire-ended ones - except probably for 1 channel for the main incomer.

In that respect, the emonPi and emonBase are identical. Connection is by 433 MHz ISM band radio.

Now this is where it gets tricky.
If you want 11 inputs, there’s no advantage in having an emonPi. 3 × emonTx will give you 12, plus an emonBase on which to run emonCMS should do what you need.
Normally, you can power a single emonTx from the a.c. adapter which is also its voltage reference for accurately measuring real power. Current is strictly limited because of the dent it makes in the waveform it’s trying to measure. In your case, the most economical will be to power the emonBase and the emonTx’s with a good clean 5 V d.c., and provide the a.c. voltage reference to all 3 from a single a.c. adapter. (You might want to consider a separate 5 V d.c. for the emonBase, because of noise considerations.) Provided you observe polarity (because one side of the a.c. input is grounded to a.c.), that will be fine as there’s almost no current drawn.
The downside to running 3 emonTx’s into one emonBase (and there’s no solution to this) is they all use the same radio channel, so you might get two jamming each other on occasions.

A handful of 3.5 mm plugs to solder on the c.t. leads, d.c. plugs for the a.c. reference, and USB for the 5 V d.c.?

I am unable to speak for that system. The designer is a programmer by profession, living and working in the USA. He’s not an electrical engineer. The last time I checked, the claimed accuracy for the system overall was better than the accuracy tolerance of the input device. There are very few of us here who have any knowledge of iotawatt, it has its own support forum.

Awesome reply, thanks Robert.
Sorry for the late reply, work got the better of me.
Thank you very much for the link from the other guy, I am indeed looking to do something very similar, although I am surprised that with using so many large CT sensors there hasn’t been a fire!
In an ideal world of course, the MCB would have this functionality built in. But even more so, do you know why the emonTx cannot have more than 4 inputs?
Good news on those small CT sensors. I plan to get 10 to try them out. The CU I have has 10 MCBs but rarely draws more than 2KW through the whole CU, so I’m not going to run into trouble
Thanks for the explanation on the differences between the emonPi and emonBase. So an emonBase with multiple emonTx units will give me the same functionality, but with more input channels?
So given that my CU never draws more than 4KW, so about 18Amps, it seems that I can use the 4th CT input without modification, especially if I use it on the smaller MCB (6A).
Tell me more about these burden resistors and why you think it is best to replace them all. Surely if this is a good idea, this should be part of the emonTx design? Not wanting to start a fight, just trying to understand. I’d really like to get a project off the ground that makes this available to the masses.
OK, so my plan is to 3D print an enclosure for the 3 x emonTx and emonBase unit. I’ll feed the 9V PSU into the case and split it through to the 3 x emonTx units. I’ll have a damned good 5V supply that will feed all 4 devices, probably powered from my UPS as that should also give a perfectly clean and steady supply.
Yes, the 3.5mm jack plugs are on the purchase spreadsheet :smiley:
The only reason I am interested in the IoTaWatt is the 14 CT inputs. However, it would be a useful and interesting project to run the IoTaWatt and emonXXXX systems side by side to see the output data!
So although I am new here, I would be interested to know why the emonTx systems are radio only to the emonBase/Pi and not wired. Also, why there are only 4 CT sensor connections to the emonTx and not more (although you di allude to processing power above).

In short, no. Probably history - the Atmel 328P has 6 analogue inputs, so it could have 5 current plus 1 voltage. Robin Emley uses the 6 inputs as 3 × 2 for his 3-phase diverter.

If you’re using the 25 A c.t, that input is now 4.5A approx. Inputs 1 - 3 are whatever the rated current of your c.t. is if you don’t change the burden.

Whatever you do, you’ll need to change the calibration to suit the c.t. ratio and burden value. The calibration constant is the current that gives you 1.0 V across the burden.

Basically, yes.

The choice of burden is a compromise. The c.t. performs best into a dead short (it’s not developing a voltage, it’s got no load). The ADC input has a full-scale input of just over 1.1 V to get the maximum resolution - and lack of resolution shows up at low currents, where c.t’s are inherently struggling with errors. So despite what some say, you should choose the c.t. with the lowest rating above the maximum current you’re likely to encounter. Having done that, you juggle the burden value to get the lowest voltage that gives an acceptable resolution but remains clear of noise.

I suggested a cuttable track to the SMD burden, but there wasn’t room. There are holes for a wire-ended resistor, so it’s not hard to flip the SMD resistor off with a hot iron and solder in a replacement wire-ended component.

The Raspberry Pi has only one serial port available. To have more than one emonTx connected would mean a multi-point protocol and dedicated drivers etc (RS485 or similar). Then of course most people don’t want to monitor each MCB separately, I think the emonTx was conceived basically as a ‘whole-house’ device. The emonPi was certainly intended for the house + PV, so 2 current inputs are adequate for that in the UK.

Hopefully, my colleague @Bill.Thomson might chip in with his experience of iotawart. He had one, tried it and sent it back.

Don’t have much to say about the unit itself. My issue was with the developer vice the device.
I had it for only a few weeks, so didn’t really get much of a chance to “wring it out” so to speak.

And just in case the OP is unaware, if you have any big AC motors on any of your circuits you may want to factor in in-rush current in that calculation. Initially I simplistically assumed a 20A CT would be fine on my pool circuit which sits at about 5A but that resulted in a whole bunch of clipping for quite a few cycles during start-up. Moving that circuit to a 50A CT solved that:

1 Like

Thanks for all the help and advice guys,
I’m probably going to go a bit quiet for a few weeks whilst I think more about this and finish off a couple of other projects I have on the go (Finishing my English teacher’s course, Drone Pilot’s licence, building a JAMMA game cabinet and designing and 3D printing some bespoke camera mounts for one of the drones). After that lot, this is my next project

1 Like

Whilst reading this topic, and the discussion on burden resistors, I got the impression that there is no burden resistor fitted to the split core CT as supplied for the EmonPi/EmonBase. Is that so? If it is then surely there will be a high risk of electrical shock from the jack plug before it is connected to the Pi/Base?
If burden resistors are fitted as standard to the supplied CT, then perhaps @SKiller should be warned that his RS solid core CT will present that danger if used with just a jack plug attached.

Or did I miss something fundamental?


I’m afraid you did. The SCT-013-000 does not indeed have an internal burden, but it has a transient voltage suppressor that clips the output voltage - there’s a picture in the test report in the “Learn” section. And you can see the clipping in this picture:

(That is an older model, clipping at ~22 V.)

Nevertheless, I always advise users never to unplug the c.t. while the cable is energised, and to short the c.t. if it must remain on the current-carrying cable without an external burden.

I think there’s some risk, but as the voltage appears to be falling off at around 3 V, then the VA rating can only be about 0.2 VA, the risk is not all that great as the primary current would need to be 60% of rated current for the 30 mA that a normal RCD would trip at to be reached - and that’s ignoring skin resistance. So painful - possibly.

Adding pair of zener diodes or a 9 V TVS inside the plug would remove what danger there might be.

The bigger risk might actually be the possibility of either flash-over inside the windings or of magnetising the c.t. core. The data sheet doesn’t give the core material, but if it’s electrical steel, then magnetisation is certainly possible if it sees a substantial current without a burden, even if the insulation doesn’t break down.