Battery Build Checklist

Summary

The steps that decide whether a battery-powered sensor lasts months or days. Written after hallway-display-build, where two of these were learned the expensive way.

Before ordering anything

• Decide the wake interval from the data, not from feel. How often does the underlying value actually change? Every-minute wakes cost roughly 15× the battery of every-15-minute wakes and usually show the same number.
• Budget the power from the wake interval, then pick the board. Not the other way round.
• Check the display or sensor holds no lock that blocks deep sleep.

On the breadboard, before the enclosure

• Measure sleep current with a meter. Not the logs. The logs will look correct on a board that never sleeps.
• Compare the measurement against the datasheet figure. Expect to be several times worse — the regulator and peripherals draw too.
• Project the battery life from the measured number and confirm it clears the target before building an enclosure around it.

After it’s running

• Pin the firmware version. Record the pin and the reason.
• Re-measure sleep current after any firmware upgrade, before trusting it.
• Expose battery voltage as a sensor so decline is visible before it’s a failure.

The two learned the hard way

Sleep current is invisible in logs. A firmware upgrade raised it 30× on hallway-display and nothing in the logs changed. It surfaced as a battery reading a week later. Any upgrade on a battery device needs a meter, not a log check.

The wake interval is worth more than the board choice. Going from 5 to 15 minutes turned six weeks into five months — a bigger gain than any hardware decision in the build, and free.