
A day in the life of a battery node
How long a battery node lasts is a sum of sleep, awake time, transmit airtime, receive windows and regulator loss. This lesson builds it for one probe, shows which part dominates and qualifies the answer in years.
A sum, not a figure
A battery estimate is current times time for each thing the node does, divided into the charge the cell can really give. LoRa is one line of it. The bench builds the sum for one soil probe from example values, all written on the panel.
The five parts
The node sleeps for nearly the whole day, and sleep is a current of its own: 20 microamps for the whole node here, not the radio chip's figure. Around each report the microcontroller and sensor are awake, at 15 milliamps for 400 milliseconds. The transmitter draws 45 milliamps for the packet's airtime. After the uplink a Class A device opens short receive windows, at 11 milliamps. The regulator that feeds all of it loses some of what passes through.
With a report every 15 minutes at the course's reference setting, those add up to about 0.94 milliamp hours a day. Sleep is the largest part, at 0.48. The cell is 2400 milliamp hours, of which the bench treats 70 percent as usable, and that gives about 4.9 years.
Which part moves
Reporting half as often does not double the life, because sleep stays. Moving to spreading factor 12 makes the same payload take over a second on the air, the transmit part becomes the largest, and the estimate falls to about 1.8 years. Listening longer costs a little charge every report, whether or not anyone spoke.
The number that is too good
Sleep current alone, 20 microamps against 1680 usable milliamp hours, would promise about nine and a half years. It leaves out everything the node does when it is awake. A radio chip's datasheet sleep figure is not the node's either.
What the years assume
Cold, ageing and self discharge take charge back from a cell, so the usable share is itself an example. The years are a calculation for these inputs, not a property of LoRa. The next lesson steps back from the radio to the network around it.
Common mistakes
- Quoting years without the inputs. An estimate with no interval, currents and capacity beside it is not a specification.
- Using the chip's sleep figure as the node's. The microcontroller, sensor and regulator draw current too.
- Forgetting that listening is paid for every report. The receive windows open after each uplink, whether or not the network has anything to say.
Edit this page — content/fundamentals/lora/a-day-in-the-life-of-a-battery-node.mdx
Discuss this article
Ask about this page. The answer stays here, on the page it belongs to, for whoever hits the same wall next.