A data sheet says the microcontroller sleeps at 2 µA. The board in the field drains a cell in four months and nobody knows why. The gap is usually a measurement problem first and a design problem second.
Measuring without fooling yourself
A bench multimeter on its microamp range puts a shunt in series with the board, and the voltage across that shunt, the burden voltage, can reach hundreds of millivolts on some meters. A board running from 3.3 V that sees 3.0 V may brown out or sleep differently. Worse, the range changes when the board wakes and the meter can drop out.
Better methods:
- A known shunt, from 10 Ω to 1 kΩ for sleep and a few ohms for bursts, with an oscilloscope or a precision amplifier reading the drop. Pick it so the drop stays under about 10 mV at the current you expect.
- A source measure unit that supplies the board and measures current at the same time, with low burden.
- A current profiler, for example the Nordic Power Profiler Kit II or the Qoitech Otii, which logs a trace over time and changes range automatically as the current swings from microamps to hundreds of milliamps.
Power the board through the profiler at the actual battery voltage, since regulators and sleep modes behave differently at 3.0 V and at 4.2 V.
Wake bursts set the average
Sleep current is only part of the budget. Say a device sleeps at 10 µA and wakes every 15 minutes for 2 s at 100 mA to join a network and send a packet. Over 900 s the burst takes 0.2 As and the sleep 0.009 As, so the average is about 230 µA. A 2000 mAh cell then lasts about 360 days on paper. Halving the sleep current barely moves that figure, while a 1 s shorter burst helps a lot. A profile trace shows how long each radio attach really takes.
What the data sheet leaves out
The number on page one does not include a regulator’s quiescent current, the leakage of a sensor or divider left connected, a pull-up that is held low, or a floating input. These often total more than the microcontroller itself. We probe each rail with the profiler and disable branches one at a time.
Limits
A bench trace at room temperature overstates the capacity you will see outdoors. Cold cuts usable capacity of many chemistries, cells self-discharge, and a weak signal makes the radio retry and the burst grow. We treat the bench number as a ceiling and test a few units in the field before promising a battery life.
If you are designing a battery device and want its power budget measured, book a discovery call and see our embedded firmware and PCB design work.
