Robins solar diverter and Sofar ME3000SP

Some years ago now, I built 2 of Robins Arduino based solar diverters. They have worked, and are still working, very well. Thanks Robin! However, I have recently installed a Sofar ME3000SP ac coupled inverter - charger and 4 (48 volts) lead acid batteries. I was hoping that the diverter would see any excess power the unit exported and send it to the emersion heater, but this has proved not to be the case. The Sofar inverter seems to see the the diverter as part of the house load, and instead of just using the excess power, sends a lot of power straight to the diverter and rapidly depletes the batteries. If the Sofar is turned off the diverter works flawlessly. I am trying to work out what is happening and would welcome any ideas or suggestions anyone on the forum may have.

I am also trying to use the Solarman monitoring software suggested for use with the Sofar,but find it unreliable and it only seems to update the data at random intervals sometimes with many hours between updates. Any suggestions please?

I have a similar problem with a different diverter and battery, it’s a bit tricky and I have not yet come up with a suitable way of working that doesn’t involve me switching the diverter on when the battery is full and there’s enough sun to max out the diverter without invoking the battery (like right now!).

Thanks for the reply Christian.

We are doing the same thing at the moment, otherwise we end up exporting, when we could use the surplus that is being exported. Will keep thinking about it.

Gentlemen, take a look at the Learn section - in there, there’s a piece about multiple diverters and priorities.

Thanks Robert

I have read the article a few times, and I think I have got my head around it now and will give it a go tomorrow . Just got to work out how to actually wire it. Once again many thanks.

Thanks everybody.

Robins diverter is now working as I hoped it would. It only sends power to the immersion heater when the sofar is exporting to the grid and does not drain the batteries.

I am not sure if this is the right forum, but I have now finished installing the Sofar ME3000SP and it is working as expected. I wondered if anyone would be interested in how the installation went and the problems encountered. Just thought that I could pass on my experiences, in case anyone else was thinking of installing one.

Perhaps you could let ne know if anyone is interested.

1 Like

If there’s anything generic that could usefully be added to “Using Multiple Diverters”, let me know and I’ll add it.

Or if you find somewhere else to post your thoughts about the installation then I for one would be interested to learn the location.


Due to the recent, and ongoing increases in energy costs, I wanted to see if I could make more efficient use of my 4kw roof mounted solar system…

As a result,I have recently purchased and installed a Sofar ME3000SP AC coupled inverter charger, along with 4 Rolls 85 amp lead acid batteries. The Sofar was second hand but in good condition, the batteries were new.

I had no problems connecting the Sofar up as per instructions. The batteries were connected in series, to give the required 48 volts, using 25mm cable along with a fuse and isolating switch.

2 CT were clipped to the AC solar connections and the incoming mains supply.

I followed the instructions for switching on. Switch everything, including solar off. Switch batteries on and wait 3 minutes. Then switch on the mains to the Sofar and it should come to life, after a countdown of about 180 seconds. Then reconnected the solar system to the Sofar. Then input the setting for country battery etc. YOU MUST DO THIS OR IT WILL NOT WORK CORRECTLY.

The unit then displayed a fault condition, in that the the batteries were overheating. I realised that I did not have the battery temperature sensor fitted, as one was not supplied with the unit I contacted the supplier, who were good enough to send me one. I am informed that a 100k resistor across the terminals, will get the unit started and show 25 c until the sensor can be obtained.

The sensor was later fitted, and the unit stared up in “standby” but would not operate as normal. I found that if I turned off the “Logic Interface Mode”, (on by default) everything stared to work fine.

The battery default setting were not correct for my batteries, so I entered suitable ones (available if required) and it worked well. Beware, even if you just want to look at the battery settings, the units resets them to default, and you have to re-enter all 12 of them again.

The Sofar has been working for a week now, and I would like to make the following points.

You will see above the problems I had with one of Robins diverts, I had on the system, and how that was rectified. Thank you.

When the batteries are nearly full, say 80%, the charge rate falls very low, only a few amps until fully charged. This is to be expected I suppose, but it does cause a lot of electricity to be exported rather than being saved. Fortunately Robins diverter is now picking this up and sending it to the water.

On setup the unit was continually exporting over 100 watts. But going to “CT calibration”, by trial

l and error this could be reduced to about 5 watts. This does not sound much, but it is enough to trigger Robins diverter.

On startup, the unit automatically sets the CTs as required, but resets them on every other startup. If you have the unit running well, you can go to “Freeze CT” and they will remain as you have set them.

The Sofar does come with Wi-Fi monitoring, using the Solarman app. However this does not work well. It is very difficult connect to the server, and it only works when it feels like it. I can find no way to update the readings, and many hours may pass between updates. If anyone knows anything better please let me know.

Sorry for the length of this post, but I thought my experiences may save someone a lot of work.

I would be happy to answer any questions, if I can.

1 Like

Hi Solar (Roy),

To get your Mk2 Router to work correctly alongside your new battery charging system, did you follow the guidance in Robert’s article Learn | OpenEnergyMonitor

Hi Robin.

Thanks for the reply.

Yes, I did follow Roberts article on multiple diverts, and it worked well on the directly linked immersion heater. I took the live supply cable and routed if in the reverse direction through the Sofar CT as outlined, and it worked as it should.

However, I also have a remote radio controlled system which activates after the water heater is satisfied. It is not practical to take this feed through the Sofar CT, so unable to use this at the moment, unless anyone else has any good ideas.

Just to be clear, I am still using your Arduino based original system which I built with you help several ? years ago and it is still working well.

Many thanks.

1 Like

I have a similar setup with a s/h Sofar 3000, lead acid batteries and a non-Robin diverter. I use a similar CT trick to Robin’s. All works well except for:

  1. Due to the double CT some of the energy values reported are false. Can only be corrected by getting the software to deduct the second CT’s value off.

  2. Often when there is sun the battery will discharge at say 100 - 150W into the immersion with the grid at around 30 W export. Surely the Sofar should settle at much nearer to 0W? What is its accuracy at achieving close to zero import/export? Rather it was charging slightly, and thus letting the immersion power reduce, than discharging.

Don’t think the second problem is due to the double CT but not completely sure.


Point 1. I don’t seem to have this problem. I believe that the theory is that the current in the live conductor going to the water heater, cancels out the current in the main conductor by going in the reverse current flow direction, through the CT. Is it possible that you have them going in the same direction?. Just a thought and I hope someone will correct me if I have got it wrong.

I do however have the same problem as point 2, as its seems do other people. I could reduce the effect by going to “CT calibration” and adjusting the values by trail and error. I eventually arrived at a value off minus 200. This has minimised the export to between 0 and 5 watts, which is the best I could do.

Hope this helps.

Nope, you haven’t got the theory wrong. The object of taking the wire the wrong way through the c.t. is to subtract that current - what happens is exactly the same as when you put the c.t. on a 2-core cable - the current going one way exactly cancels the current going the other way, and it reads zero.
The objective is to ensure the c.t. in question doesn’t see that current, so it doesn’t attempt to respond to the other diverter. There are big problems when each diverter attempts to respond to what the other one is doing, and making sure you have the correct cascade with the monitoring in the right places is essential.

Maybe not. Most if not all diverters (with the notable exception of Robin’s MK 2) are designed to bleed a “small” (could be 100 - 150 W mentioned) amount of power back into the grid. I believe it’s solely to prevent user complaints that they’re still importing when they expect not to be.

I think you’re probably right, but until we know the exact arrangement of the connections - where each item connects in, where the monitoring is in relation to those, and what’s meant by “double c.t.”, it’s hard to say.

1 Like

To clarify, when I say double CT I mean I am using 2 CTs to be functionally equivalent to the 2 wires through one CT as in Learn | OpenEnergyMonitor I don’t have the room in my CU to put 2 wires into one CT. If one wire is through one CT and one through another and the 2 CT outputs are anti-paralleled it’s functionally the same (I hope). But, if the 2 “identical” CTs are in fact 5% apart, when there is 2000W in both, one will measure 100W more than the other so the subtracted total is 100W.

That though does not explain the 100ish W effect as it’s a pretty steady 100ish even when the 2000 becomes 1000 or 500. A 5% error would lead to the 100W ish changing as the immersion power level charges. And a fixed 100ish offset is not explainable by a CT offset as they can’t produce an fixed offset on a small current.

It’s the tail end of the solar day that is the problem when the battery can’t take much charge ,as its fairly full. The diverter is capable of cutting it to 20-30W so why does it stick at 100?

Diverter balanced powers are a bit of a minefield I think as, with such a distorted waveform being generated by phase control or burst fire diverters are they really delivering the claimed export power? Their software may be thinking it’s measuring something very small and writing it on their screen/log, but is it really true? Only a very accurate true RMS power meter gives the true figure. That said my Smart Meter and the diverter generally are agreeing on the 30W within 20%.

Any log of immersion or Sofar load that uses that double CT /double threaded single CT in its calculation is going to be wrong as it’s actually measuring immersion minus Sofar. E.g. my Sofar LCD is giving the incorrect export to grid measurement when immersion current is >0.

Can’t work out how to quote part of a previous post but in reply to my own “The diverter is capable of cutting it to 20-30W so why does it stick at 100?” a friend suggests 2 possibilities:

  1. The Sofar does not want to switch on and off from the grid too often as it wears out its grid relay. So it may be hanging on with the relay closed for a while and consequently delivering a small current to the house/immersion. Can’t though explain why it’s not less than 100W unless the Sofar behaves like a diverter that does not deliver that close to zero. Nor explain why it’s hanging on for up to an hour in this state.

  2. Dirt between the faces of the CTs can cause small errors. Clean them all carefully and check they are closing tightly.

To quote some real figs (at least what the Sofar and diverter were reporting as real):

“Generation” was 610 solar + 110 out of battery = 720W

Consumption = 325 to immersion + 366 to rest of house + 29W to grid = 720W

What I would have expected is :

“Generation” was 610 solar + 0ish out of battery = 610W

Consumption = 215W to immersion + 366 to rest of house + 29W to grid = 610W

Or even 100W ish of battery charging, with 100 ish less to immersion, as the SOC was 95% at the time.

Not knowing your Sofar, I can’t help with that. If there is a genuine relay in it with mechanical contacts, then your friend is quite likely right - they want to minimise mechanical wear on the contacts.

To quote a post, drag a highlight over the part you want to quote, and click “Quote”:

Screenshot 2022-03-23 11:24:35

And this gets inserted:


Sorry, carn’t help with the 2 CTs issue, I don’t have enough knowledge on that subject. But I have checked today and found that on my setup, if the solar generation meets the house needs, their is no drain or contribution from the battery. I have managed to get the export down to 0 most of the time, but occasionally it does go to 5 watts. Therefore, I think your second example would be correct.

If the solar generation is more than the house needs and the battery is nearly fully charged, the unit will export to grid, but this is picked up by Robins diverter and sent to the immersion heater, as expected. I have noticed that the unit can export any amount to grid and does not seem to stick at a minimum of 100 watts as I think you suggest.

Hope this helps.

Sorry, I missed answering this point:

Yes, that’s right. But do you really mean that there isn’t enough slack in the wire or wires to allow both to go through the one c.t? If that’s the case, can one not be extended to allow that? I don’t know of a rule that says you can’t, as long as the wire and any connector you use has an adequate current-carrying capacity, I don’t see any objection.

It is possible but means running the diverter live wire 2m back from CU to the meter where my export CT is. Or throwing out my Eco Eye monitor’s export CT from the CU to make space for the Sofar/Diverter CT.

The Eco Eye is pretty useless now as is can’t correctly report powers with all the wring complexities of battery, Open EVSE charger, two solar PV inverters, etc. I’d need to massively extend its CT wires to clip to wires that do have the right current flow and/or double thread its CTs.

Not sure I quite understand you but on mine the battery is sometimes adding 100W ish to the spare PV and driving the immersion with that total (325W in above example). As PV reduces the immersion reduces but the battery stays at that 100W ish and the export at 30W. It will revert to the “correct” zero battery power once PV is too low to drive the immersion.