After messing about with solar panels and lead acid battery charging for a arduino project in the UK. I managed to easily keep it all running on a 15W solar panel. Lessons learnt however is to get the most from the battery it is better to charge a 12v battery and use a step down buck converter for a 5v project, because from 13.6v to 11.2v is the useable stored energy. If you just use a lipo at 4.7v on an esp32 running at 3v you only get 4.7 to 3.2v useful energy. However some one suggested Super capacitors. This is a concept i never considered before, but after my experience of solar , I think this is a brilliant solution. So from what i have learnt i would suggest a 16V 20F(because most quality solar panels are 12V(18Vmppt) and 16V packs are cheap and pre built) super capacitor pack using a MPPT module (maximum power point tracking solar charger ) that is designed for 14.2v lead acid charging. this will prevent over voltage on the capacitors. And for the voltage to the ESP32 use a step up step /down 5v buck converter for usb or direct to the board set at 3V. The advantage to the Super capacitors is fast charge up in sunlight . Not effected by low temperatures and effective discharge voltage from 14.2V down to 0V because of no deep cycle damage. I have ordered a Super capacitor pack and i already have the rest of the kit on the shelf. Let you know how it turns out.
@M1BDK Do you have a link to the Super Capacitor you are using? I would be interested in seeing the complete setup when you have all the pieces put together.
MPPT
Super Caps
DC-DC buck step down
Found this calculation for Ah for Super capacitors.
Ah= (Cap V*F/3600)
The system of using MPPT module to charge Super caps or batteries. Works well while the MPPT module is able to maintain 18V at the solar panel. This works well in good sunlight situations, however I have noticed MPPT systems suffer on dull or if the sun is low in the sky and not at a good angle to the solar panel. The voltage output from the Solar panel is unable to reach 18V so the MPPT will not pump any power to the Super caps and no charging takes place. Not built one as I am not an expert in developing electronic circuits. But my principle in this low light situation. An intelligent charge controller would recognise the panel is unable to reach it’s maximum power point of 18V and then would operate a circuit to hold the panel voltage as close to maximum power point to charge a Super cap to as high as possible and then operate a step up buck converter to lift the voltage from the solar storage cap to 16V to the final Super cap power bank.
Interesting. The last time I looked into super capacitors they had super prices. I see some tech companies have been able to bring the prices down. A few problems you might run into (though I am not familiar with that MPPT), because that is designed to charge a ~12V battery, it wont be expecting the voltage to be below 10V. So when the sun comes back up in the morning, there may not be enough voltage left to re-start the charge. The MPPT might need to be temporarily powered to restart. And then as you mentioned, cloudy days where the voltage is only 11v you wont be fully charging the cells. The other concern is the buck converter from 16v to 5V, you will have a few losses at the higher voltages but also might become unstable below 5V.
OK so looking at that super capacitor you linked, there are 6 2.7V 100F capacitors. The energy at 16V 20F (16.7 Farad Actual, Caps in series) would be the same as if you had 2.7V 600F (All in Parallel).
If you were able to take your solar panel and limit the voltage to 2.5V (%93 Full) you would never over volt your capacitor and thus would not need the protection circuit. By design they will stop taking current once full so we dont have to worry about overcharging them either. Now from the capacitor you use a boost converter DC-DC to a USB port to power the radio. You can take the 2.5V down to 0.9 volt. if you need to add more capacity, add another Super Cap in parallel. This setup has some drawbacks too, like only being about to go down to 0.9V on the capacitors is ~%33 down from %93. So here too we are only using %60 of the total capacity.
I like the concept, I may have to order some new toys to try it out. I did manage to find some parts. I may place an order and play with it. I did see a couple of protection boards earlier, it might be possible to put 2 caps in series for a 5V setup, then adjust the voltage from the panel to 5.2V and run the group down to 0.9V. This would bring each cell down closer to 0.45V (%16). You will need a protection circuit if you installing them in series to make up 5V, since 1 cell could fail sending >2.7V to the remaining cell and causing it to fail (violently?)
Some Reference Math Hidden Here.
Just for reference. Using the capacitors listed below. If you install Capacitors in Parallel, the voltage will remain 2.7V but you can just straight up add the Farads.
500F+500F = 1000F.
If we install them in series you increase the voltage.
2.7V+2.7V = 5.4V
but the Farads are added as
1/((1/F)+(1/F)) = 1/((1/500)+(1/500)) = 1/(0.002+0.002) = 1/(0.004) = 250 Farads.
And 2.7V @ 1000F = 5.4V @ 250F.
I just skimmed through this thread and didn’t see this here, forgive me if it’s been mentioned already. But I came across this on the Adafruit website. seems like it might have a place in this discussion. I like the USB C port…
Adafruit Universal USB / DC / Solar Lithium Ion/Polymer charger - bq24074
Nice find. The BQ2407x series was what I landed on as the best option for linear regulator based solar charging.
This looks like a fun project, but if you want to buy something pre-assembled that just works then these are good: https://www.tindie.com/products/xorbit/lifepo4weredsolar1/
You have found the perfect electronic board 
I am amazed that it has everything you need for an independent node.
It doesn’t even cost that much.
Thanks so much for recommending it
I’m about to setup a T-Beam as a repeater. I had the cell holders and cells on hand, so I am starting with this for power. I’ll be using a 12v/10w pv panel and a cheap 10A charge controller.
Very clean install. I have used similar controllers (cost about $16) , they work well. What kind of 12v battery do you plan to use ?
I’m actually hoping to not use a 12v battery. The USB cable in the photo is soldered to the +/- lines of the USB jack on the reverse of the board. The TBeam has two 18650 cells in parallel on the back of it. My hope (and it really is hope, because I’m just starting the testing) is that the panel and controller can provide 5v to the TBeam for enough hours every day to keep the batteries topped up.
I’m here to learn so if anyone sees anything that leads to trouble, please chime in. This is the first time I have used one of these controllers. It came with this wee panel: 10w pv panel. We get a lot of sun here, but I probably should have purchased the 20w panel.
I believe I have the same controller, if I recal the controller will not power up on solar alone. It has to sense at least 10V from the battery even under full sun.
If you are only charging a small amount of AH through the t-beam I’d think you are not capturing excess energy on the good days to carry you though the bad.
Maybe something like this?
I can calculate estimate the amount of mAh of consumption of the board, multiple it for the amount of day that you need to run on battery.
Example:
10Ah = 10000mAh battery 1 cell lipo
Board consumption 10mA
10000mAh / 10mA = 1000h
1000h / 24h = 41 day approximately
Also NMHi are not a good choice it’s a tipe of chemistry that prefer to been cycled not keep it charged.
How are you getting the 10 mA consumption?
I haven’t seen any new numbers for the recent updates.
I ask because transmit, GPS fix, Bluetooth activity can all make a big difference in daily mA usage.
Is the 10mA for a relay that rarely transmits?
Also 10AH (10000mAh) is a lot bigger that what most are attaching directly to devices.
In case it helps this is the spreadsheet I use:
I think for an esp32 based board the average power draw in normal mode (case 2 in spreadsheet) should be about 15mA.
I’ve remeasured and confirmed these values recently for NRF52 but the last time I’ve measured a esp32 was on 10/8. See line 51 for that info.
I have made same assumption for the calculation, but 10Ah i think a vaible capacity for a long term use, it’s only 3 or 4 good 18650 cells.
I’m currently testig 4 recycled cell’s
For my use case; deploy solar hopefully not revisit them about them for 5-10 years I’ve concluded that Lifepo4 are a better choice for me that Lithium-Ion.
For me, it was the 1500-2000 charge cycles
But the non-explosive failure and better charging at low temperatures sealed the deal.
