I am designing a solar power system which I would take deep into the Amazon forest to an indio village. They have no energy, and they like to have internet (StarLink) and a fridge would be really useful. So I am thinking to start with a system terminating in a 2000 W inverter.

I think I will use LiFePO4 batteries. And I would like a charge controller setup which can provide reasonable load capacity in addition to charging the battery. Battery longevity would be important because it's very difficult to get replacement things shipped to this remove place.

I am thinking that a Priority Charger feature should be included where the voltage and current characteristics to the battery can be provided while possibly limiting the load current.

In my understanding either the charge controller needs to have this load current limiter feature or the inverter should have a load current control input. I don't know what is typically done? In the DIY setups I have seen on YouTube everyone seems to hook the load directly to the batteries, and I haven't even seen an ammeter on the load wire so that the charge controller could at least know what the load current is to then be able to regulate the battery charge current by taking the load current into account.

I could draw some block schema diagrams, but I thought maybe what I am describing in words is already clear to those familiar with the subject matter?

Since there is no answer, I am going to continue to think aloud until someone bites. This question here is actually getting to the same point as I am: Best practice: Should I charge a battery while also pulling load?

So the charge controller thinks it's outputting 14.8V to charge the battery in absorption phase, and let's say that during that time 20 A rush into the batteries. The loads at that time will pull, say 100 A sustained peak at times, but the charge controller has a maximum of 60 A. So now what happens?

  1. The charge controller would blow up
  2. The charge controller would reduce voltage to limit current to 60 A, ending the absorption phase
  3. The charge controller would shut off as if a fuse had tripped

I think these situations should be fairly common. So what I am proposing is to put a current limiting device into the load circuit.


simulate this circuit – Schematic created using CircuitLab

And if we are closing the load current limiting MOSFET, then we have part of the 14.8 V lost over it, so only the remainder would be available for the inverter, pushing it out of it's input operating range, and thus essentially shutting off the inverter. This would mean that a MOSFET may not even be useful here, that a simple relay might do the trick. We just trip the switch to shut off the load.

But that is not great either, because all we would need to turn off is whatever pulls the 100 A, there may be a few LED lights and a laptop, and especially the StarLink internet device plugged in, which we would like to continue running, and which could easily be sustained with the remaining 40 A to the inverter.

How would that be best managed by the charge controller? Given that we can program our charge controllers based on an Arduino core, it seems to me that I would like to program handling of this issue into my system. Someone says that this is beyond a DIY project, but then how does a non-DIY device handle this situation?

Here is what I could imagine doing:


simulate this circuit

I can designate a Critical (low) Load branch, which has critical round-the-clock usage, mainly the StarLink device, WiFi, laptop connection (but ideally in such a way that one could tell the laptops not to charge their batteries for a moment), maybe some critical lights. This load would be always on and always sustainable by the charge controller given the battery charge current. Then there would be a high-load branch which would have the inverter for the refrigerator. It would possibly turn off the refrigerator for 3 hours straight as needed to sustain the battery charging absorption phase.

This would be fine if the high load would be for something like electrical tools. But it would not be fine for a fridge when we are in a hot climate, as in 3 hours it is possible that the freezer compartment would start to thaw! So how could we avoid that? One easy way would be to just dimension the charge controller and solar panels such as to be able to sustain the fridge in the critical (low) load branch. Or to have some nifty circuit of smaller batteries which would be available during the charge phase of the main batteries and just sustain the compressor cycles of the fridge for those 3 hours that the main batteries were to charge. But now that is starting to get complex.

The point of all of this was just I like to know how "priority charging" works and what we can do with loads that have a higher current demand than can be granted during the absorption phase of the charge cycle?

  • 2
    \$\begingroup\$ This is not a DIY-level project. Sorry. \$\endgroup\$ Commented Apr 19 at 0:55
  • \$\begingroup\$ It might be easiest to have separate "priority" invertors, say 3, and sequence them based on demand and solar capacity. You would have one invertor for high-priority that's always on. A medium priority for "nice to haves" and low-priority for things that don't matter. Then, using logic in the Arduino to enable which invertors are online. \$\endgroup\$
    – MOSFET
    Commented Apr 19 at 19:47
  • \$\begingroup\$ In your previous post on the same topic, I presented some simple system diagrams from which to get ideas. The systems are designed to provide 9kW, but reducing the number of panels and suitable inverters solves the problem. It's not useful? \$\endgroup\$
    – Franc
    Commented Apr 20 at 8:02

1 Answer 1


All of this can easily be done with off the shelf hardware, which will be a lot more reliable than taking an arduino into the jungle.

Here's how I'm doing it:

PV -> Hybrid inverter with 48V battery -> Mains output

The inverter has a communication interface, they all do. Usually it's modbus. Same for solar charge controllers, if you get decent gear (like Victron) they will have a communication port. Therefore, it is very easy to talk to it with a low power computer. I use an Orange Pi Lite with isolated USB-RS485 interfaces, then Python.

Various variables can be queried: PV power, load power, battery state of charge and charging/discharging power, smartmeter data...

Based on these, the Pi can control a bunch of relays, so it is easy to switch loads at the output of the inverter. For example, when the battery is charged and there is excess PV power available, I use it to turn on the water heater, along with several other heaters in the house.

You could also use several inverters, of course.

For an offgrid application it is very important to pay attention to the idle power use of inverters.

So the charge controller thinks it's outputting 14.8V to charge the battery in absorption phase, and let's say that during that time 20 A rush into the batteries. The loads at that time will pull, say 100 A sustained peak at times, but the charge controller has a maximum of 60 A. So now what happens?

Nothing special. Solar charge controllers are designed for this, so it will dump current into the battery from PV while the inverter takes current from the battery. When battery voltage drops a little bit, the charge controller will output more current.

More details:

If you have DC loads it's better to use DC rather than double conversion (battery->AC->DC) especially if that allows you to switch off the inverter and get rid of its idle power use, which can be high: 50-70W for a 3-6kW low cost inverter. Some are much better (Victron) but you need to check the specs carefully.

But you will need to distribute that DC efficiently and safely. 12V over not so long distances is a problem due to high current and thus voltage drop. If you want to light one house, OK. Light a village, probably not. So you would need to distribute 48V, which means DC rated protections (fuses, breakers...) and DC-DC converters at the point of load. These are not hard to find, here's an example 48V to 5V converter which I'm going to use to power my Pi from the solar battery.

You can't put LED lightbulbs in series as the one which uses less current would get too much voltage.

Note switches rated for AC are always rated for a much lower DC voltage. This is because AC arcs are extinguished easily due to AC voltage going to zero twice per cycle, but DC does not, so DC arcs will keep burning. For example a 600V AC breaker used to switch 300V DC solar panels will pretty much burst into flames instantly when turned off.

If you manage to run with no inverters at night, just DC, it would be interesting to put a temperature sensor in the fridge, so you can disable the inverter when the fridge doesn't need power anyway while guaranteeing it won't get too hot.

There needs to be a way to disconnect everything if battery SOC is too low to avoid killing the battery from over-discharge. If you use Lithium, the BMS will take care of that.

If the charger tries to charge the battery, the load can't take anything out, that would interrupt the charge cycle, wouldn't it?

That wouldn't be usable, so solar chargers are designed to work in this situation. It's possible to charge a battery while pulling current from it: battery current is the difference between charging current and load current. Depending on current direction, the battery will either charge or discharge. The charger simply needs to not freak out when it sees battery voltage dropping due to load current, so it's a software issue.

  • \$\begingroup\$ Thanks for this idea, good point. Take data from the charge controller and then turn on various different loads, inverters, controlling it all with a Raspberry PI. And this allows us to turn off inverters that can't be used anyway, so we save on their idle power drain. The only thing I didn't get is your last paragraph. If the charger tries to charge the battery, the load can't take anything out, that would interrupt the charge cycle, wouldn't it? \$\endgroup\$ Commented Apr 20 at 2:30
  • \$\begingroup\$ One thing I wonder with your single inverter solution: isn't it better to use DC where DC is desired instead of going through an inverter and then back through a power supply to DC? For example, the StarLink box wants 48 V DC anyway, the laptop would be good with 24 V, and the cell phones want to be charged with 5 V. While light might be done with 12 V DC (or just put 4 in series onto the 48 V DC line). It would be less waste, right? \$\endgroup\$ Commented Apr 20 at 2:37
  • 1
    \$\begingroup\$ I put the answers to that in the answers, because comments often get cleaned up. \$\endgroup\$
    – bobflux
    Commented Apr 20 at 7:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.