Battery Runtime

AD7UF, Sep 2025

When proposing to run a Meshtastic node from batteries, whether a temporary portable or a permanent solar power installation, power budget becomes an important question. Online and among friends, I have gathered many anecdotes and figures thrown out for battery life and PV panel requirements, covering a range far too wide for mere optimism vs. pessimism to explain.

One crucial factor that almost none of them address is usage profile.  Sure, your node might stay up on a single 18650 for two weeks if there's absolutely no traffic, but what about under an actual load, esp. a heavy load as you might expect if it's being used for emergency traffic that other media are unable to carry?  Merely driving a 0.1W radio makes a big difference, even without the increased CPU activity.

I set out to compare the maximum and minimum expected runtime of a node, partly because I was planning to build a series of standalone solar-power nodes, and needed to determine two things for a successful node:

1. How many batteries are needed to reliably carry it through dark periods?

2. What size PV panel is needed to keep them charged during limited daylight hours?

Test hardware

The meshtastic node consisted of a power-efficient RAK 19003, powered by a single used 18650 of unknown age/condition (pulled from a dead GPS). I'm guessing its effective capacity to be 1-1.5 Ah. My goal wasn't to establish an absolute certifiable rating, but simply to get a rough range of real-life usage.

Methodology

All tests were performed in my office, with room temperature generally staying between 65-75°F.

For the "minimum drain" test, I ensured that no other nodes were nearby to minimize traffic, and only activated my phone's Bluetooth radio to take a battery measurement. I took no other power-saving measures, leaving most settings at default.

For the "maximum drain" test, I used the "Range Test" module set for 1 second interval to broadcast as often as permitted, maintaining 25% "air utilization". It automatically stops after 8 hrs, so I ensured that I manually shut it off and then back on before that.  (I tried various other methods of generating traffic, including using a shell script with the python client to generate traffic on other nodes for it to relay, but nothing provided significantly increased load over time, and the Range Test was hard to beat for simplicity.)

I also performed two charge tests, using a USB charger rated for 800mA @ 5V -- high enough not to be the limiting factor as the battery took roughly 10 hrs to charge, meaning that it was absorbing charge at a rate of 50-300mA if its capacity was 500-3000mAh.

I've noticed that the node's telemetry report often seems to hold onto previous values and then jump to a new one. Much variance might be explained by luck of timing: how much current was it drawing (which would drop the voltage) at the instant the measurement was taken? The final readings on the low-drain test were interesting. I actually stopped the test with the node still alive, though I doubt that it would've run much longer. The node actually died at the end of the high-drain test.

So how long did it last on a single cell? 2-11 days depending on load.  You'd better figure on 2-3.

The two tests tracked pretty closely.  I went to sleep during the last part of the 1st test; thus the long gap between readings. I slept longer than expected, and it was done when I next checked it.