Open Source MPPT Solar Charge Controller- Wifi enabled

[evjim]

Hi, I am working on an open source solar charge controller using the TI SM72442 IC. I am designing for a 100VDC maximum open circuit input and a 65 amp current limit. It can charge batteries from 12 to 48 volts. It will incorporate the ESP8266 wifi chip for remote status view and programming.

I originally started this project using the BQ24650 because it is an all in one chip for MPPT. It does not need a separate H bridge driver. But it is limited to 30volts open circuit and cannot communicate with a microcontroller. The SM72442 has I2C and readable registers that show the status of Vin, Vout, Iin, & Iout. Most often solar charge controllers are located in a not so commonly accessed area. So the homeowner may not know the battery state of charge is low or there is an error. Having wifi connectivity allows the user to more easily monitor the system.

I'm working on the project in KiCad. It is my first project in KiCad as I usually use a closed source EDA tool.
Here is my github page of the project:  https://github.com/junglejim8/opensolar
Would be cool to get feedback or collaborators.
The schematic is nearly complete, I am working on the BOM, will move to layout in a week or so.

Question:
-I am trying to source an inductor for the project. It must be at least 0.8uH, But Ideally more like 3-10uH to keep input and output capacitance values lower. The design documents allow for a 60% peak to peak ripple current. So I need to find a power inductor with Irms 65 and Isat >85amps. I like coilcraft because they have interactive tools for predicting inductor temperature. But unfortunately none of their inductors offer this current rating for prolonged use.

Can I reliably parallel power inductors to double the current? If I use two of the same part will they share current equally?

[Codebird]

Can I reliably parallel power inductors to double the current? If I use two of the same part will they share current equally?

No. They are low resistance parts, at least the good ones, so a variation of even less than an ohm could result in seriously unequal sharing. Remember that at low load, when you aren't drawing enough current to bring the panels down to MPPT, your buck converter is going to be continually on - and your inductors become basically resistors.

You might be able to use lower-inductance, higher-current parts and put them in series? But 85A is a lot to ask of a buck converter, you might have to look into some sort of multiphase buck, or else a specially-wound giant of a toroid.

[evjim]

Thanks, yea Coilcraft also replied saying they did not recommend it. They also suggested putting higher current capable inductors in series.

I have found this inductor  on digikey. It seems a lot heftier than the coilcraft ones I found. But they do not show any temp rise information. I would still need to limit Ipeak to about 70 amps. Might be worth lowering my ambitions and set a rms current of 50 amps or so.

[w1ntermute0]

I just started thinking about building one of these.  It doesn't look like there's a reasonable off the shelf version with WiFi.  I was originally thinking of building one using a Raspberry Pi since it has built in WiFi and would allow for a lot of flexibility but maybe I can work with you on yours.  I'm mainly a mechanical engineer but I have some experience with software/electronics well.  Maybe I can help with an enclosure among other things.

Here's two projects I was looking at:

Looks Great but does not use MPPT:
http://www.instructables.com/file/FZIT4MII2TQVGML/

Looks promising but doesn't include a circuit to control/monitor the load:
http://www.freechargecontroller.com/

I'd like to be able to monitor Solar volts/amps, battery volts, load amps (with ability to switch load off when battery gets low).  Having cool charts/graphs/stats in the local web interface and/or the ability to relay to one of the online IoT sites that does graphs would also be good.  Although these are often installed out of the way, I kind of like the LCD screen also.

Let me know what you think,
wm

[Codebird]

There's an old project of mine from a year or so ago. It actually kind of sucks - it'll only do 10A, and it's hardly professional, but perhaps the design might give you something of an inspiration. It's not a charge controller, it's purely the MPPT part - something you're supposed to be able to put inline between your panel and a PWM charge controller or passive load device. The intention was for it to be cheap enough that you could fit one to every 100W panel. In practice it's unpredictable and unstable, but it does work under the right conditions.

https://birds-are-nice.me/ipfs/QmeHTKsAWhaKeygAqRBkAUkt5t2e9NPZcHT2ZcuyZcXcwy/

There are certain fundamental laws of physics that force you to trade between inductance, current handling, and physical size. You can't have all three: If you want a 65A inductor with lots of inductance, it's going to have to be huge. No getting around it.

[evjim]

That's a cool project you were working on codebird.

wm, yea I agree there are no good wifi options on the market. And the open source options so for have limited specs. My goal is to have the specs of this design be pretty all around useful (up to 2500watt installs). But it wont come cheap. At low volume (25qty), this project's BOM is ~$65

Enclosure is something I have researched a bunch but not found something I am set on yet. I am thinking extruded aluminum. I would like to find some design that is readily available and consistent. It is kind of a barrier before I can pick the PCB dimensions and mosfet locations. The board needs to dissipate a lot of heat, >50watts. It needs to be ~10cmx10cm, but that is flexible. Staying under 10cm has pcb manufacturing discounts in low volume.

[Codebird]

I can't say I'm much of a buck-expert, but I think the way to go will be multiphase - which probably means your microcontroller will be generating the buck switching signals. Forget an atmega, you'll need an ARM chip if you go that route, as it involves generating multiple PWM outputs via bit-banging. You'll also need syncronous drivers if you want to avoid wasting a lot of power as heat.

The alternative would be to use one really, really big inductor.

[evjim]

The TI IC (SM72442) handles the buck/boost algorithm and switching signals in this design. It adds a bit of cost, but it also handles reading current in and out.

[w1ntermute0]

$65 sounds more than reasonable.

For the enclosure, do you expect it will need to help dissipate a significant amount of heat? You could start with something simple made from flat plate/sheet, stand offs and acrylic, kind of like this.

More custom solutions could be made from bent sheet metal or extrusion like you mentioned.

If the heat isn't too great we could start with a 3d printed enclosure.  When you have a board layout with mounting holes I can take a swing at some boxes.

wm

[martinj]

Hey, your project looks very interesting (and ambitious).

Just had a look to your github repo. First thought: The selected MOSFETs IPI076N12N3 might be slightly under-dimensioned. For high duty cycles when input and output voltage are similar, the High Side MOSFETs is almost always on. In that case you have to dissipate (50A)^2 * 12 mOhm = 25 W plus lots of switching losses in only one MOSFET. (I chose 12 mOhm instead of the rated 7.6 mOhm, as the Rdson increases with temperature)

I assume your switching frequency will be very high, considering the small inductor. Which frequency do you want to use?

Do you know the TI reference design for a two-phase buck converter also using the SM72295? This could be a good starting point for a two-phase buck design: http://www.ti.com/tool/TIDA-00120

I am also working on an open source MPPT solar charge controller. With 55V max input and 20A output at 12V/24V it just fits for one or two standard roof-top PV panels (approx. 280W). You can find it here: https://github.com/LibreSolar/MPPT-Charger_20A

The Libre Solar MPPT contains an extension port called UEXT. You can directly plug an ESP8266 from Olimex to get a wifi connection. I successfully integrated it into an OpenEnergyMonitor system (emoncms.org) to generate nice graphs.

Martin

[Inverted18650]

Are there any updates for this project? I'd love to get onboard

[martinj]

Are there any updates for this project? I'd love to get onboard

Yes, just finished a new revision and ordered two boards. If tests are OK, I want to produce it in a small batch for interested developers. I am currently waiting for a quotation.

It's not a finished product yet, but a previous version is already running since approx. 1 year without problems in a friend's home.

You are more than welcome to get involved. The hardware should be more or less finished now, but there are plenty of ideas for optimization of the software (especially regarding communication and user interface). Just send me an email if you are interested to get one of the boards.

[evjim]

Awesome, thanks for sharing. It has been many months since I have worked on this project. Yours looks very far along. I think I won't have time to get back to it until next summer.

[kicker22004]

I know it's been a while since you posted this. I'm curious where I can find a parts list for your project. I'd like to build or buy one for my small solar array to compare it to my current PWM setup.