Powering an Electron in subzero temperatures


I have a field trial running an Electron 24/7 (no sleeping) and a groundwater monitor using the 12V panel/controller that I linked to, and a 7Ah 12V SLA battery.

I sized it using the following design conditions:

  • Assume 50% battery capacity to extend the life
  • Assume 50% total efficiency for Solar Recharge
  • Assume 1 week of cloudy/rainy weather (little Solar Recharge)

The Electron monitors the SLA voltage using a simple voltage divider.
We are currently experiencing 11 days of Rain and cloud cover (verses the 7 day “design”).

I prefer empirical data when available over equations, so here it is:
This is a screenshot of the Battery/Solar Performance for the past month:

As you can see, the lowest dip is 12.00 Volts during the past 11 days of cloud cover.

Lithium chemistries have many uses and I love them. But it’s hard to beat the watt hours (>10x compared to 2Ah, 3.7V Electron Li-Po) provided by a simple 7Ah SLA when size & weight are not design constraints.
Cost and Local availability for replacement batteries are usually much better for SLA verses Li-Po.

If you need longer cloudy weather runtime or more reserve capacity, bump up to a 12Ah battery for $25.


I faced the same quandary in January 2018, when I setup my first device to run in the winter. Up until then, my particular application (agricultural-related) had no use in the winter, so I typically removed the devices from the field in the early fall. I did some reading on LiPO’s in cold weather, and concluded that it might be okay to discharge them at freezing temps, it is apparently not good at all to charge.

I went with the following setup:

  • 5 Ah 12V sealed-lead-acid battery, feeding the Vin pin on the electron, with no LiPO connected to the electron’s battery connector. (About $25)
  • 5 W 17 V solar panel ($20 on Amazon). The mounting bracket that I use puts the panel right at 45 degrees, which is almost ideal for my 40 degree latitude. I think I paid another $15 for the mounting bracket, which seems like too much. I’m sure there’s a better/cheaper solution for this.
  • “Battery Tender” solar battery charger as the charge controller ($25)

The electron is always cloud-connected (always breathing cyan). No lower-power or sleep modes in use.

Normal operating voltage is about 12.5-12.8 V. In the middle of the day when the charge controller is “topping off” the battery, I see as high as 13.8 volts. This means that I’m somewhat violating the Vin spec on the electron, though not violating the TI power management IC (PMIC) voltage spec. My understanding (based on other threads here) is that the electron’s limit of 12V is related to thermal concerns with the power dissipation of the PMIC.

My assumption is that there is not likely a thermal problem in my particular application, since my current demands are always low on average. I.e., I’m not charging a LiPO from Vin, so the only “high current” (and therefore higher power disspation) ever seen by the power management IC are the short spikes that occur when the cellular modem is actually transmitting. Since these are short spikes, the PMIC should not overheat. Though, I haven’t done measurements or calculations to prove this, so take it with a grain of salt…

The battery voltage is connected (through a voltage divider) to one of the electron’s ADC inputs, so I have a log of battery voltages over the 10 months that the unit has been out in the field. I’ve seen the battery voltage drop as low as about 11.8 when it’s really cold and cloudy/snowy for multiple days, but this occurs rarely where I am in Colorado. Even in the winter, most days I see the charge controller top off the battery by late afternoon.

So far so good on this setup. I think I plan to switch to this in the long run for all of my deployments because, though it’s a bit more pricey than using the supplied LiPO, it is really a much more robust power solution than a small solar cell plus the 2 Ah LiPO that comes with the electron. And, it seems simpler to me than trying to use heaters, insulation, fans, etc, to keep a LiPO in a safe operating temp range.



@RWB Going to be testing that LTO charging chip after Thanksgiving, so I’ll try to provide everyone with my results. LTO cells are certainly the way to go in literally ALL temperatures, with the only drawback being that you need to design two custom circuits to manage the LTO cell (a charge controller + boost converter). The booster converter is needed because the LTO cell voltage isn’t high enough to power the Electron.

My application will be unique because I’m using low-voltage / high-current solar cells, so my charge controller circuit will be different than most people, which will use a higher voltage solar panels.


We have several hundred deployed battery-powered units using solar for recharging. We have chosen 12V SLA for simplicity, cost, availability, and environmental reasons.

  1. Simplicity: Everyone can charge a SLA, so that means we can use TaskRabbits efficiently because we don’t run the risk of burning down their house if our training isn’t watertight. Additionally, the chargers are inexpensive and the batteries are (almost) invulnerable to charging abuse. You can even plug solar panels directly into SLAs without any charging circuitry (not saying you should, though).
  2. Cost: We understand that cycle cost across the full battery lifetime is cheaper for LiFe batteries, but that assumes the product is fully baked and won’t change. The calculated risk is to go with the cheaper battery and see if we need to worry about years 5-10.
  3. Availability: The 12V12Ah battery is ubiquitous, and can be found at Amazon, local stores, or shipped from overseas. Great for buying a ton in a pinch, or picking up an extra one if urgently needed (esp. because airlines won’t let you fly batteries of any size past ~100Whr).
  4. Environmentals: while there are industrial lithium batteries which can be charged down to -40C, the standard batteries cannot be charged below 0C. Cheap SLAs handle real-world temperature swings far better than cheap LiPo/LiFe batteries.

Regarding physical differences between lithium-based and lead-based technologies, while it’s true that the lithium greatly eclipses lead in terms of weight, in terms of volume it’s a much closer ratio. If you’re driving batteries out to your site, and your stuff is installed immobile on the ground, then the volume difference isn’t really relevant. We have enough empty space to install 8 batteries in our product, and yet we only put in two.

Another important factor is that we have allowed in our business model a certain amount of manual swaps per year. This keeps solar panel size and cost reasonable, as if you’re zero maintenance and 100% off-grid either your solar panel is so large you have tons of waste in the summer, or it’s too small to keep you going 100% of the time in the winter. The best route was middle-of-the-road, where we address individual underperformers, perhaps by swapping batteries occasionally or perhaps by cramming in every battery we can so that one device will survive winter.


After reading this thread and worried about my solar & LiPo powered electrons in northern Minnesota (so far fine @ -22C last week), I decided to experiment with lead-acid batteries.

I had a 4V SLA around so i wired on a jst plug and plugged it into the jst battery port on the Electron. So far seems to be working. It charged some today with a little sunshine and the voltage and SOC read off the Electron (fuel.getVCell() and fuel.getSoC()) seem to be reasonable.

@rickkas7 seemed to feel that charging and the battery meter wouldn’t work. I’d rather not break one of my Electrons, so what problems should I be looking for?


He just thinks the SOC levels will not be accurate since it’s a different battery chemistry. It will not damage the Electron.


The Charging Profiles for Lead Acid and Lithium-Ion are completely different.
I’m not sure if you risk damaging the Electron, but I don’t think you should do this.

Your 4V nominal SLA will never get a chance to saturate.
I’m guessing you will end up with a sulfated battery pretty soon.
And you wont have anywhere near the rated capacity of the battery either with the Electron terminating charging at ~4.1V.

You would get much better performance using a 12V SLA (to Vin) and recharging it with a solar panel.


Rftop is correct. There are two issues: the bq24195 charge controller/PMIC and the MAX17043 fuel gauge.

The charger in the bq24195 is only designed to charge 3.7V LiPo batteries. It’s probably not a good idea to charge any other chemistry of battery with it, though if you are using an external charger you can use the voltage regulator only and disable charging. It should be fine with charging disabled.

The fuel gauge is only designed to monitor LiPo batteries as well. You’re within the specs of the chip, so the chip won’t be damaged, but it won’t return correct SoC, either.


Has anyone finished a charging design to solve this -40C? Preferably with the 4V SLA and maybe a BQ24650 (which I wish was on the E0… :frowning: ) @Darmitage If you haven’t got everything finished yet, we’re starting to look into this design for the B0 integration.


Space was not a constraint for us, so we went with SLA.


How did you work around the BQ24195 limitations? Did you have to replace it with a different charge controller circuit?


We are using a separate SLA charger. Solar --> charger --> SLA --> 6vdc --> e-series (or electron). When we don’t need to equip for extreme cold we just jun the solar straight into the particle and use a 18650 lithium.