A device that reports beautifully and dies in four months is a failed product. On a battery-powered field device the radio is the easy part; the life is decided in microamps of sleep current, in how seldom the device wakes, and in an energy budget where the datasheet's headline figure barely matters.
| Contributor | Why it matters | How it is controlled |
|---|---|---|
| Deep-sleep current | Runs continuously for years | Hunt leakage to microamps; power peripherals fully down |
| Wake overhead | Every wake has a fixed cost | Wake less often; justify each wake |
| Radio transmit | Short but high peak | Send rarely, compress, batch |
| Location fix | Often the biggest single event | Fix only when needed; use cheaper estimates |
| Modem paging (PSM/eDRX) | Reachability costs energy | Tune sleep against required latency |
| Temperature derating | Cold cuts capacity, sags pulses | Model the cell under cold pulse load |
The intuition that a wireless device spends its battery talking is almost always wrong. A field device — a tracker, a sensor, a meter — transmits for seconds and sleeps for hours or days, so the current it draws while doing nothing dominates the total. A few microamps of avoidable sleep current, drawn continuously, outweighs the entire energy of a year's transmissions. That is why the datasheet's active-mode figure barely matters and its omissions — real quiescent current, brown-out behaviour, how cleanly a peripheral actually powers down — matter enormously.
So the design starts as an energy budget, not a feature list. Every event — a wake, a sensor read, a location fix, a transmission, a return to sleep — is accounted in microamp-hours, and the sum against the battery capacity is what sets the achievable reporting rate and life. Location is usually the surprise: a satellite fix can cost more energy than a day of everything else, which turns when and whether to acquire one into a battery decision rather than a feature toggle. And every wake carries a fixed overhead, so waking twice as often can cost far more than twice the energy.
The battery itself is not a constant, which is where lab results and field results diverge. Capacity falls in the cold and sags under pulse load, so a cell that comfortably lasts on the bench can fail in winter if the current pulses were not modelled. And because a three-year life cannot be tested in three years, it has to be proven by measuring the energy of each event, combining it with the duty cycle and a temperature-aware battery model, and projecting — a projection that is only as honest as the measured sleep current behind it.
Because a field device spends almost all of its life asleep and only seconds transmitting. The link works; what decides whether the product lasts is the current it draws while doing nothing. A few microamps of avoidable sleep current, running continuously, outweighs the entire energy of a year's transmissions, so the design is an energy budget dominated by the quiescent state rather than by the radio.
Because acquiring a satellite position is energy-expensive — it can cost more than a day of everything else the device does. That makes when and whether to take a fix a battery decision rather than a simple feature: fixing on every report may be unaffordable, so the design decides when a fresh fix is genuinely worth its cost and when a cheaper estimate will do, which is often the difference between meeting the life target and missing it.
By measurement and modelling. The energy of every event — wake, sensor read, location fix, transmission, sleep — is measured in microamp-hours, combined with the duty cycle and a battery model that accounts for temperature and pulse load, to project the life. Accelerated testing and, above all, the measured sleep current make that projection defensible rather than optimistic, because the continuous quiescent draw is what dominates the total.
Written for engineers. Share it with one.
Tell us the specification, the constraint and the deadline. Programmes that cross silicon, radio, embedded and AI are where Faststream is strongest.