# 10 years of battery life: what really makes it last How does a battery-powered IoT sensor run for ten years? The role of the lithium cell, transmission frequency, network choice and temperature, explained simply. Publié le 2026-09-20 par DigitalWan — /en/blog/iot-sensor-battery-life/ ## In short - Battery life depends first on how often the sensor measures and transmits, not on the size of the cell. - Lithium thionyl chloride (Li-SOCl2) cells lose very little charge at rest, which makes a ten-year service life realistic. - A LoRa transmission costs far less energy than a cellular one: the network architecture directly shapes battery life. --- “Up to 10 years of battery life”: the promise appears on every industrial sensor datasheet. It is realistic — provided you understand what it depends on. Across a hundred sensors, the gap between a good and a bad configuration is measured in dozens of site visits. ## The cell does not decide, usage does A given cell holds a fixed amount of energy. Battery life is simply the result of a division: that reserve divided by the average power draw. And in an IoT sensor, consumption is spread very unevenly: - **Sleep** takes up more than 99% of the time and a tiny share of the energy; - **Measuring** costs little, except for active sensors such as a radar; - **Radio transmission** is by far the biggest item, in very short but intense current peaks. The direct consequence: **cutting the number of transmissions extends battery life far more effectively than cutting the number of readings.** A sensor that measures every ten minutes but only sends a daily summary, or only sends alerts, will outlast one that transmits every single reading by years. ## Why lithium thionyl chloride Sensors designed to last a decade almost all use a 3.6 V Li-SOCl2 cell — for example an ER18505M in the DigitalWan range. Three reasons: - **High energy density**: a lot of capacity in a small format; - **Very low self-discharge**, around 1% per year: after ten years on a shelf, the cell still holds most of its charge; - **A very stable voltage** for almost the whole service life, which simplifies the electronics — but makes the remaining charge harder to read than on a phone battery. Their quirk: **passivation**. At rest, a thin layer forms on the anode surface and protects the cell. It clears on the first demand, but explains why a long-idle sensor responds slightly slower on its first wake-up. ## The network shapes the energy bill For the same message, not all links are equal. A short LoRa transmission draws noticeably less than a cellular exchange, which means waking a modem, attaching to the network, then transmitting — often at higher power. That is why a well-designed architecture puts the constraint in the right place: - [slave units](/en/products/msu-sd7/) transmit over LoRa, which is very frugal, to a master unit; - the [master unit](/en/products/msu-md7/) gathers the data from several sensors and handles the cellular link on its own; - where power is not an issue, a [solar gateway](/en/products/msu-sg7/) listens continuously and removes the battery constraint from the collection network altogether. The detail that matters: a battery-powered master unit does not listen all the time. It opens scheduled listening windows, which lets it act as a collector without burning through its energy reserve. ## Four settings to get right at deployment 1. **The measurement interval**: the longest one that still meets the real need. A tank level does not need a reading every minute. 2. **The transmission policy**: on threshold and on alert rather than continuously. The vast majority of readings change no decision. 3. **The active detection modes**: every extra mode wakes the sensor more often. 4. **The site temperature**: a buried structure is thermally stable, an enclosure in full sun far less so. ## The real subject: maintenance cost Across a hundred sensors, going from six to ten years of battery life is not gaining four years: it removes an entire replacement campaign, with its travel, its structure openings and its service interruptions. That is where the total cost of a deployment is decided — far more than on the unit price of the sensor. Before ordering, have the battery life estimated **for your configuration**, not in the abstract. ## Frequently asked questions ### What is a Li-SOCl2 cell? A non-rechargeable lithium thionyl chloride cell with a nominal voltage of 3.6 V. It offers high energy density and very low self-discharge, around 1% per year, which makes it the reference for sensors meant to stay in place for ten years without maintenance. ### Why say “up to 10 years, depending on the configuration”? Because battery life depends on how the device is set up: how many readings per day, how many transmissions, which network, and the site temperature. Two identical sensors can last three years or ten years depending on their configuration. ### Does cold weather reduce battery life? Cold temporarily reduces the available power and can increase cell passivation, which shows up as a voltage drop during current peaks. The stored energy is still there, but transmitting in very cold weather is harder on the cell.