DIY battery storage solution?

Hello all. Having read some posts here I see there is a wealth of experience. I hope that some can come to bear on generating a possible solution to my problem.

I have 4kWp of solar pv on my roof via 16 off 250W panels, with 16 x Enphase M215 micro-inverters. As luck goes, one of the inverters croaked a few weeks ago, and I bought 2 secondhand replacements, one of which is up there doing the job of the dead one. Enphase have had no qualms about replacing the duff one, so now I will soon have 2 spare inverters with connections via Enphase’s Engage cable.

My proposal is to hook up the two spare inverters to the mains, and supply them with battery power to supplement my overnight electricity use, ie power the fridge and a few lights while the batteries last.

It seems that the primary problem will be to stop the inverters sucking all the power out of the (rather meagre 10Ah @36vdc) batteries as soon as they are connected, and squirting it straight into the arms of the big bag Grid, that is being of no benefit to my lights or fridge after dark.

Clearly I need a smart controller that measures the export current and controls the inverters’ output (2 x 215 = 430 Watt) to give nothing away to the Grid, and this is where I seek your suggestions. How best to achieve this, short of spending big bucks/pounds?

I have built one of Robin Emley’s Mk2 PV diverters from his kit.
I have an un-used, but not new, EmonPi.
I have two spare Raspberry Pis.
I have an EmonTx v3 with two CTs.
I have some small amount of experience with Arduino, but next to nowt on RPi.
I have a willingness to learn, so pointers to others’ similar projects would be welcomed.

My own thoughts are that a philosophy close to that which Robin applied to his Mk2 PVdiverter could work on measuring the exported power after dark, though I am a little ‘in the dark’ regarding the possible techniques to control the inverters’ output, in that they must be connected to the mains in order to synchronise phase and frequency with it, and that they will take about a minute to do this if disconnected from the mains for just a short time (anti-islanding?).

Regards, Phil J

Welcome to the OEM forum, Philip.

I think you are on the right lines with the principle of Robin’s diverter, but you’re going to need a means of controlling the output of the inverter. Unlike a water heater, you can’t operate the inverters in burst fire mode to give an average nett energy flow of zero. That will be crucial to having a working solution. If the inverters have no means, other than knowing when the battery / PV voltage is sufficient, to determine their output power, then I’m afraid I think your idea is dead. Unfortunately, while that facility obviously exists internally, I rather doubt that making it available as an input is part of the design, I can’t find any indication of a means of doing it.

I can think of a way of doing it, which would need experimentation and might work: that is to put a controller between the battery and the inverter so as to limit the input power to the inverter. In concept, it’s a simple rheostat, in practice, it would need to be a step-down switched mode regulator. And then you’d control that to balance the grid energy to zero. But I’ve no idea whether your inverter would even tolerate something like that (which is not what it’s expecting) on its input.

If you could isolate those circuits completely from the grid - by a changeover switch or contactor for example - then it would be a relatively simple matter (technically, given a suitable off-grid inverter). Operationally, I could foresee objections.

All other issues aside, energy production from said 10 Ah, 36 Volt battery will be a whopping 360 Wh.
Sounds like it’s much more trouble than it’s worth for a battery bank of that size.


Thank you Bill and Robert for your input.

@Bill.Thomson. Bill, it’s a concept that will be scaled up to a more practical size if it proves to be a success.

@Robert.Wall. Robert, I have some limited background in electronics and control/automation, but have so far resisted the onslaught of t’interWeb, so I can discuss discrete electronics freely, but I request that WebJargon be explained should it come up!

In a recent Arduino project I used the principles behind Robin’s Mk2 PV diverter to build an energy limiting device for a friend with 6kW of electric heating and a 15A breaker 2 storeys below his flat (this is in Abroadland, hence the low breaker value). The project uses a CT to measure input current (filtering software from Emonlib I am pleased to add), subtracts that from 15A, and burst fire controls the residual power to his electric heater elements via a combination of a relay (for current >13A) and a solid state switch (SSR) with zero volt switching output.

I mention this because I consider that a variation of this (ideally hacking into Robin’s Mk2 hardware for an output pin) should be suitable to drive an SSR to control the inverters’ output so that they ‘fire’ to compensate for electicity imported after dark, ie controlling the imported power down to zero, or thereabouts.
My own control strategy is too basic for this use, in that I chose to burst fire complete mains cycles as a proportion of a two second period (2s = 100 mains cycles at 50Hz and 1% control resolution). A long period, but one suited to the characteristics of a thermal breaker.
Note: He is due to have his supply upgraded to 40A soon. I mention this to avoid sidetracking this thread.

Robin’s technique, in the Mk2, is much more sophisticated - and at the moment is too complicated for me to follow on the Arduino - allowing single mains cycles to be processed; I think that kind of strategy would suit my inverter project better.

Do you agree? If so I shall press on with understanding Robin’s software before returning to ask for help to hone my design; if not perhaps some suggestions as to a better way to proceed?

Kind regards, Phil J

you could try to adapt mk2 software to do what my versions diverter router does . it allows multiple relay control so if you are using multiple small inverters you can step through them. in blocks . the other option there are several cheap Chinese grid tie inverters with limiters on the market that use a potentiometer just change it to digital potentiometer (100k are usually what they are ). or as mentioned a controlling the input going into the inverter… or you could buy the grid tie inverter with limiter that does every thing for you already example 1 or example 2 with example 1 if you figure out the method it sends their rs485 code you could eliminate their limiter and use your own and with example 2 it also possible to eliminate their built in limiter and used your own via a digital pot

The M215s and M250s in my system have a 5 minute anti-islanding delay before restarting.
I don’t know whether or not the units Enphase produces for use outside the US have different AI delays.

I’d go along with @stephen’s suggestion, forget the inverters you have (unless you can find a means of controlling their output) and get an inverter or inverters that can have their output controlled by a voltage that’s proportional to the nett grid energy flow.

I wouldn’t like to think that a normal grid-tie inverter would like being connected and disconnected from the supply at regular intervals. OK, you might reasonably expect them to ride through a missing cycle or two occasionally, but without knowing all the circuit and design details, I wouldn’t expect them to cope with a variable M/S ratio with a period of 2 s.

They wouldn’t. They shut down 20 ms after sensing grid loss, then take 5 mins to resume production.
At least the ones I have behave that way. Variants produced for use outside the US could of course, have different specs.

Thank you for your replies.

@Bill.Thomson. I agree, any anti-islanding delay would render the burst-fire technique unsuitable. Whether UK supplied Enphase inverters have 1 minute or 5 for this does not change that.
I also agree that my variable mark-space burst fire technique is not suitable for controlling the inverters’ output, so I’ll move on to Robert’s ideas…

@Robert.Wall. As a solar inverter I suspect that the Enphase unit (and all other solar inverters) are very much designed to operate on varying voltage and current levels from zero (dark) to about 40vdc (full sunshine, no output load to drive). The Enphase unit requires a minimum of 22vdc before it starts to function, and I should imagine that commercial drivers will dictate that it starts to produce power as soon as possible after the input is above 22vdc.
It is there that I imagine the best route to inverter output control lies; the exact way to achieve that will require further thought…

Apologies if I’ve missed or glossed-over your suggestion, not intentional, just getting a little excited that a possible solution is emerging.

Thank you all, and goodnight for now.

Some interesting info here. Some of it is nothing more than guessing or conjecture, but a couple of posts make good points. Here’s one of them. The person posed the question to Enphase and here’s what they said:

Basically, I would like to know if I can power the Enphase inverter from batteries and fed the grid, i.e. instead of having a direct solar panel as a source, I’ll have solar panels charging batteries and when the voltage is high enough the Enphase inverters will take excess DC power from the batteries and export to the grid.
The Enphase inverter will still be configured to work in a normal grid-tie mode however sourcing power from the batteries. Also, the specifications read that the MIN start voltage is 28v. Is that that below this voltage the inverter is in sleep-mode i.e. not inverting?

Here is the reply from Enphase… :cry:

What you have described below has not been tested by Enphase Energy and is an application that we would not support. The Enphase Energy Microinverters are only to be used specified in the manuals and any deviation would void any Enphase Energy warranties.
As for the minimum voltage, the answer is yes. The Enphase Microinverters must get to 28v for them to turn on. Once the units turn on, they can drop below this voltage and still work in bursts, but they must turn on first. Enphase Energy Microinverters use electrolytic caps in the dc input to filter and regulate changes in voltage, but only Enphase has added the logic to control the discharge of these caps to the utility during low light conditions, which causes the Microinverter to pulse or burst energy out.

I did a quick search and found this:

Interposed between the battery and inverter, it looks as if this might allow you to drop the voltage and convince the inverter to reduce its output power.

Control by a knob obviously won’t work (except maybe to establish whether it’s worth putting more effort into the idea and developing the control strategy and implementing it), but the “efficiency … up to 94%” caught my eye. In other words, you’re guaranteed losses of worse than 6%, and that’s on top of the inverter losses.

It’s not making a lot of sense.

Which are ~3.5%


I found using these type of devices to control the input to grid tie inverter in either buck or boost - you also need one that limits the current as well… as the MPPT will just pull more current. either burning the fuse or the device
I have use one of these on an mppt device for 2 years with out problems

That falls in line with one of the posts in the thread I referenced above.

MPPT = Maximum Power Point Tracker

Thank you @stephen, an idea worth pursuing later. First I must get the basics together. Having salvaged three 12v 12Ah sealed lead acid batteries from an old electric bike I found this morning that the supplied charger is dead, the batteries have 5v, 2v and 5v respectively, and my posh desulphating 12v charger is refusing to attempt to revive two of them. This project may have a short delay while I purloin replacements!
My plan has changed somewhat:-
First, secure the services of three 12v lead acid car batteries to make a 39vdc pack.
Second, prove that each inverter will work and provide useful output for the period that the batteries remain charged, using a DC-DC buck converter to limit the input to 30vdc and the current to 7.5Adc (about 215 W).
Step three; once that is proven, move on to measuring the output characteristics of the inverter pack with reduced voltage, and its recovery time from OFF when battery voltage increases above its lower ‘switch-on’ threshold.
Step 4; investigate the modifications necessary to the DC-DC converter to enable it to provide automatic output voltage adjustment in order ultimately to match inverter output to household consumption.

Currently on step 1…

What I’ve been assuming all along is you realise that a battery is a close-ish approximation to a voltage source, and a PV array is a close-ish approximation to a current source, and the two behave entirely differently. You must bear that in mind and tread very carefully.

I will, and thank you all for your contributions.
P.S. One 12Ah battery restored to reasonable health so far.

I have to admit I’ve only really skimmed this thread, so apologies if I’ve missed something important! However, it seems to me you might be over complicating things a bit… the fact that you’re using micro-inverters gives you some degree of control already. Without any current limiting ( which as mentioned would be needed to truly control the output ) you can simply switch a given battery on or off to give you stepped outputs, so if grid import goes above x you enable one, 2x you enable two, etc.

Given that you’re unlikely not to use the whole capacity of the batteries over a 24hr period, you could also install fixed current limiters between the battery and inverter if you really wanted to, so you get smaller steps ( but longer time on battery. ) This would reduce the efficiency a bit though, both in the inverter and by adding the current limiter.

You might need to seed the inverters to keep them “on” so they remain synced to the grid - that way you can control the output pretty instantly. You could do this from the grid itself I suppose rather than use the batteries ( i.e. feed in a current which is above the inverter power up threshold but below the point where output is generated. ) I’ve not really thought that one through though, so would need some investigation.

Edit: … and thinking about it, just in case you haven’t investigated already… as mentioned above, solar panels are not really voltage sources as such, in fact a solar inverter will normally try to adjust the voltage across the panel to achieve the best power transfer using MPPT, if it tries this with a battery things could go badly wrong depending on how it’s implemented, so having a current limiter in circuit might well be a wise thing anyway.

Interesting thread. A quick question/comment; how will you stop the solar panels and battery set-up from islanding itself in the event the grid goes down?

20 milliseconds after the grid dies, the output from the microinverters dies as well.
That’s what the IEEE 1547 and UL 1741 certifcations mean. Enphase microinverters have both
certifications. In that respect, there’s essentially no difference between feeding the inverters
from PV modules or batteries. If the grid goes down during daylight hours, the inverters go down too,
but the PV modules don’t stop production.