I've heard that RAK boards will get stuck in a deep sleep mode or something if the voltage goes below a certain threshold, and need to be power cycled or reset. Haven't tested myself, just something my local mesh group has said they experienced.
this post was submitted on 19 Sep 2025
14 points (93.8% liked)
Meshtastic
1179 readers
60 users here now
A community to discuss Meshtastic (https://meshtastic.org/docs/introduction)
founded 2 years ago
MODERATORS
I was all geared up to make my first solar node, got a rak and a solar panel, put an 18650 in it. I was doing some testing to see how long it would last on my bench without sun. I got distracted and the voltage got lower than the rak liked, and then it just... Never came back. Wouldn't wake up again, not with solar power or adding additional external power. I'd probably need to disconnect and reconnect the battery. But I never tried that, because the point was that I wouldn't be able to.
Before this, I had already been researching and thinking about different methods, and I had already wasted some money on some ill-planned hybrid lithium capacitors. I also had some disappointing range tests from the local hill. Plus I was trying to get my buddies on board and they were dragging their feet.
So this was the last nail in the coffin. You can't have a device enter a non recoverable state when it's 40 feet up in a tree at the top of the tallest hill in town. Just bad design.
That stupid thing is still sitting on my desk, almost a year now. I lost all interest, unfortunately.
I'm sure I'll come back to it, eventually.
But yes, in my experience, rak doesn't like low voltages, and will likely get stuck in an unrecoverable state.
So, I design battery packs for a living. 115F ambient operating conditions are doable assuming you dont add solar loading. Can you paint your enclosures white? Or cover them? A high quality 18650 will operate fine up to 50C especially with a very light load and a low charging current. Do you have more details on your setup, chipset and average load?
I developed a simple circuit using 1 Farad supercapacitors and a power management IC, no battery. It buffers a 5v solar panel and provides smoother power when the aun is up. It even trickle charges the supercapacitors from just a bright security light outside the window at night.
The effect is that the capacitors store the trickle until they are full, then deliver a burst of steady full power, enough to power a node for 30 minites when busy. So with this circuit, a node cleverly placed near an outdoor light would operate continuously in daylight and intermittently at night with some artificial light nearby.
I would love to have a copy of this circuit!
I imagine adding more capacitance not only increases the run time at night under a light, but also the charge time. How did you decide on a good value?
Very cool!
Have you looked into lithium capacitors? Here's one built into a charging circuit. Kind of interesting
I don’t have the circuit diagramed out, I just built it on a breadboard using what I had on hand already. I dug out the breadboard to at least give you what info I can.
It is an old project so I pulled some of the passive components for use elsewhere. But I can at least give some guidance.
I mis-spoke on the capacitance. I used two 4F supercapacitors in parallel, chosen based solely on voltage and economy for small purchase quantities.
I used a dual voltage comparator LM393 to detect full charge and the cutoff voltage and NPN switching transistor 2N222a to turn things off and on.
I used trimpots to set and tweak the on and off threshold voltages.
I didn’t bother with any output voltage conditioning because the application was to run a small air pump to aerate a tiny aquarium intermittently. The air pumps operate on a wide voltage range (4-6v are fine) so I opted for simplicity and lower losses. It ran for five months without a hitch, as I recall. The pump drew about 500ma steady state.
The research that I am really interested in is into biodegradable polymers for super capacitors. One of the main problems with batteries is the need for exotic materials that are expensive and harder to recycle when they inevitably wear out. There are so many applications that can be powered intermittently and eliminating batteries when possible can lower cost and circuit part count significantly.
Hope that helps.
Hey thanks this is helpful!
I ran a PirateBox on a Pi 0w by connecting it to a pass-through USB battery pack, then the battery pack into one of those tri-fold Anker (or was it Aukey? Been a few years) solar panels. I hung this up in a window that got good sun most of the day (being in San Diego was of benefit here) and that thing would run off battery after sundown and the battery would charge once the sun was back. I wasn’t measuring any voltage or other metrics, but it worked fine.
Have often thought about taking one of the Heltecs in my drawer and firing up a node that way. I still have that same battery in another drawer. I have many drawers.
Edit: fkn typos.
A bit of a buffer battery is good.
I've got the opposite problem of lithium batteries freezing. But three NiMH batteries in series is close enoigh to drop-in. They're ok with cold, maybe also heat?
Or, capacitor to even out the load. Or a super-capacitor if you want to last all night long. They'll consume everything to charge up from empty, though.
I would say a small bank of supercapacitors and like an inexpensive 10 watt USB-C solar panel from ali Express would be a pretty good way to accomplish this. Supercapacitors, if they're protected from rain and UV, will last a very long time and from what I understand are not vastly affected by environmental temperatures.
Yeah, it's completely possible to just buy one of those solar panels with a USB connector and plug it right in. I've done it before. But you're going to experience multiple daily outages if there's so much as a cloud going over it (unless you install a giant panel), not to mention entire days where it's cloudy.
Thanks, i may give it a try.
Im in one of the sunniest places on the planet. Almost no rain or clouds....ever. So it may work out :)
Would be a nice way of communicating :). You can communicate with everyone around, but when come evening it's automatically quiet time.
Im thinking its just more reliable and one less part to fail in a years time. And less chance of a spark in high fire areas. I dont necessarily need to communicate 24x7 if its just for fun. And it makes the local mesh just a bit bigger in the day.
Put a daylight sensor in line with one of the leads with a cutoff some percentage higher than the cutoff of the device.
Out of curiosity, what about a supercapacitor? Since it doesn't store energy chemically, it might work better and not become a spicy pillow.
Was just watching a video about it yesterday. I don't think this is the link I watched but there are dozens of builds if you search YouTube.
LiFePo4 are way more stable than Lithium Polymers, even LiPos, which are used in every EV, sitting in the hot sun for days on end...
LiPos are obviously a stupid choice. They should only really be used when you need extreme discharge rates, or really is sizing.
“daytime only” node
There's lots of ways to store energy besides batteries...
Depending on how much you'd want it at night, or just how bored you are, you could do something with a simple motor/generator combo, have it pump water up to a mini tower, and then when solar kicks off it flows back down turning the pump into a generator and powering the solar till the tanks empty.
Almost definitely not worth the hassle, but would be a pretty fun project to fuck around with.
I’m just nervous that it will under voltage the device over the long run if that makes sense.
There definitely should be some kind of capacitor. Likely built into your panel, but you want to make sure there's one somewhere