Axon Sense 3 clears bench validation with 21-day battery life

Axon Sense 3 finished bench validation on July 9 after eleven weeks on test. The headline number is battery life: 21 days of continuous monitoring on a single charge, against nine days for Axon Sense 2. The device is the same size, the same weight to within a gram, and runs at the same skin-contact temperature.

None of that came from a bigger cell. The battery is physically identical to the one in the previous generation.

What changed

Three things, in descending order of how much they contributed.

The largest single win was moving continuous signal quality assessment off the application processor and onto the sensor front end. In Axon Sense 2, the main processor woke every 250 milliseconds to decide whether the signal it was seeing was worth keeping. It usually was, which meant we were spending most of the power budget asking a question whose answer was almost always yes. The front end now makes that call itself and only wakes the processor when something is wrong or when there is data to move.

Second, the radio. Axon Sense 2 held a connection open. Axon Sense 3 batches and burst-transmits on a schedule that adapts to how far the device is from a gateway. In a ward with dense Fieldmesh coverage it transmits more often and uses less energy per sample. In a home with one gateway in the hallway, it backs off and buffers.

Third, and least glamorous, we stopped logging things nobody read. About 11 percent of the previous generation's flash writes were diagnostic records that existed because they had been useful once during development.

The number we actually watched

Battery life is the number that gets quoted. Skin temperature is the number that decides whether a device is wearable for three weeks.

A sensor that sits against someone's arm for 21 days has to stay close to ambient, because the failure mode is not a burn, it is irritation, and irritation means the device comes off on day four and the 21-day battery becomes irrelevant. We hold ourselves to a limit well below the regulatory threshold, and the redesign had to hit the new battery target without moving that number at all. It came in at 2.1 °C above ambient at the worst point of the worst unit.

What bench validation is and is not

This was bench work: instrumented units, controlled conditions, simulated and recorded physiological signals, environmental chambers, and a drop rig. It establishes that the device does what the specification says under conditions we chose.

It is not a clinical study, and it does not establish how the device performs on a person going about their week. That work runs next, and the results will be different, because they always are. Adhesive behaves differently on a real arm. People shower, sleep on the device, and put it somewhere we did not anticipate.

What happens next

• Human factors and wear study, starting in August, 140 participants over 28 days.
• Electromagnetic compatibility testing in the Minneapolis chamber, running in parallel.
• Formcloud fleet support for Axon Sense 3 ships alongside the device, not after it.
• Limited availability to existing Axon Sense customers in Q4 2026, general availability in 2027.

Axon Sense 2 remains supported. Devices in the field are not affected by this work, and there is no firmware change coming to them as a result of it.