Recipe

Estimate battery life from a duty-cycle table

Use when: a device repeats a cycle of states (sleep, measure, send) and runs from a battery

Goal: an average current and a battery-life estimate, each with its sources and assumptions stated

Taught in Power budgets · extends Estimate a battery's life, best case

1. 1

   Write every state the device passes through in one cycle, with how long it lasts. The sleep state fills the rest of the period.

   Example

   Period 60 s: measure 0.3 s, send 1.5 s, sleep 58.2 s.

   Why: a state you forget is current you never counted, and the forgotten one is often the long, quiet one.

2. 2

   Take each current from a datasheet table you can cite, or from your own measurement, or mark it as an example.

   Example

   Measure 40 mA, send 120 mA, sleep 0.05 mA: all marked example.

   Why: a budget is only as good as its worst number, and a chip's sleep figure is not a board's.

3. 3

   Multiply each current by its time and add the results.

   Example

   40 \times 0.3 + 120 \times 1.5 + 0.05 \times 58.2 = 12 + 180 + 2.91 = 194.9 mA·s.

   Why: charge, not current, is what the battery supplies.

4. 4

   I_{avg} = charge per cycle / T.

   Example

   194.9 / 60 = 3.25 mA.

   Why: the battery sees the average over many cycles.

5. 5

   Life in hours = kC / I_{avg}, with k about 0.7 to 0.8 unless the battery's datasheet supports better.

   Example

   0.8 \times 2000 / 3.25 \approx 492 h, about 20 days.

   • the result is long enough — check the biggest state's numbers again, then measure the real device.
   • it is too short — shrink the biggest charge share first: shorter or rarer sends, or a lower sleep current if sleep dominates.

   Why: derating covers what the label doesn't promise, and attacking the biggest share is the only change that moves the answer much.