How Do You Calculate Battery Life for an Embedded Device?

Detailed Explanation

Battery life calculation is one of the most important steps in designing a battery-powered embedded device. Getting it wrong by a factor of 2–5× is common when designers use peak current figures instead of average current, or forget that quiescent currents add up.

The Fundamental Formula

Battery life (hours) = Battery capacity (mAh) ÷ Average current draw (mA)

This is the starting point. The critical variable is the average current draw — not the peak current, not the idle current, but the actual time-weighted average over a full duty cycle.

Example: A 2000 mAh Li-ion cell powering a device that draws 50 mA:

Battery life = 2000 mAh ÷ 50 mA = 40 hours

Simple when the device runs at constant current. The engineering work comes in calculating the true average for a device with variable states.

Calculating Average Current for a Duty-Cycle Device

Most battery-powered embedded systems cycle between at least two states: active and sleep. The average current is:

I_avg = (I_active × t_active + I_sleep × t_sleep) / (t_active + t_sleep)

Or equivalently, using duty cycle D = t_active / (t_active + t_sleep):

I_avg = I_active × D + I_sleep × (1 − D)

Example: IoT sensor node

• Active (MCU awake, sensor reading, transmit): 15 mA for 200 ms
• Sleep (MCU in deep sleep, radio off): 20 µA for 9.8 s
• Cycle period: 10 s → 10 cycles per minute → 86,400 cycles/day
• Duty cycle: 200 ms / 10,000 ms = 2%

I_avg = 15 mA × 0.02 + 0.02 mA × 0.98 = 0.30 mA + 0.020 mA = 0.32 mA
Battery life (2000 mAh) = 2000 ÷ 0.32 = 6250 hours ≈ 260 days

Building a Full Current Budget

For any real device, the current model must include every current-consuming component:

The quiescent currents add up: a system with three LDOs at 20 µA each, a flash in standby at 3 µA, and a protection IC at 5 µA already has 68 µA of standing load before the MCU's own sleep current. For a device targeting sub-100 µA sleep current, this is a major fraction of the budget.

The Peukert Effect and C-Rate Derating

The stated capacity of a Li-ion battery (e.g. 2000 mAh) is typically measured at a slow discharge rate, often C/5 (0.2C). At higher discharge rates:

• Internal resistance causes additional voltage drop
• The cell hits its cutoff voltage sooner
• The actual delivered charge is less than the rated mAh

Practical impact for embedded designs: for most IoT devices with average currents well below 0.1C, this effect is negligible. For devices with sustained high-current loads (a cellular modem transmitting at 1–2A bursts, a motor driver), the available capacity at those high rates may be 80–90% of the rated capacity. Factor this in for accurate predictions on high-drain designs.

The Effect of Temperature on Capacity

Li-ion cells lose effective capacity at low temperatures:

For devices deployed in outdoor, automotive, or cold-chain environments, use worst-case temperature capacity when sizing the battery. A device designed to last one year at room temperature may only last six months in a cold warehouse.

Self-Discharge as a Current Budget Item

Li-ion self-discharge is approximately 2–3% of rated capacity per month at 25°C. Converting to a current:

I_self = 2% × 2000 mAh ÷ (30 days × 24 hours) = 0.056 mA = 56 µA

For devices targeting multi-year battery life at very low active duty cycles, 56 µA of effective self-discharge is significant — comparable to the MCU's sleep current. Add it as a constant term in the current budget for long-life designs.

Adding a Design Margin

Theoretical calculations are optimistic. Apply a 20–30% derating factor to account for:

• Cell capacity tolerance (±10% between units is common)
• Capacity fade over the product's service life (expect 80% capacity after 500 cycles)
• Temperature derating in the worst-case deployment environment
• Unmodelled standby currents and leakage paths

Practical runtime = Theoretical runtime × (1 − margin)

For a calculated 260-day runtime with 30% margin: 260 × 0.70 = 182 days as the design target.

Design Considerations

• Measure, don't estimate: manufacturer datasheet current figures are often under typical conditions, not worst-case. Use a power analyser to measure actual current in every operating state — including boot, OTA update, and error recovery paths — before finalising the battery size.
• Optimise sleep current first: in low-duty-cycle designs, the sleep current dominates the average. Cutting sleep current from 50 µA to 10 µA saves more battery life than cutting active current from 20 mA to 15 mA.
• Account for voltage regulator quiescent current across the full voltage range: LDO IQ typically increases slightly at low input voltage. Measure it at the battery's minimum operating voltage (around 3.0–3.3V), not only at 3.7V nominal.
• Know your Li-ion battery's voltage range: the full usable voltage window is approximately 3.0–4.2V. Your power supply circuit must keep the output rail stable across this range — check that your switching regulator or LDO can maintain regulation down to 3.0V input.
• Power profiling as a hardware validation step: including a precision power measurement break-out (a 0.1Ω shunt resistor and test points) in the prototype design makes current profiling straightforward. Removing it in production costs only the unpopulated pads.
• Low-power circuit design: architecting an embedded system for multi-year battery life involves choosing the right MCU sleep mode for each idle period, clock gating peripherals, and coordinating radio duty cycles with power domains — work that Zeus Design's firmware and hardware teams approach systematically for IoT and wearable products.

Common Mistakes

• Using peak current as the "typical" draw: a BLE radio transmitting at 15 mA for 5 ms per second has an average current of 75 µA from radio alone — but if the designer uses 15 mA as the current draw, they underestimate battery life by 200×.
• Forgetting regulator and protection quiescent currents: these are always present, even in deep sleep. For ultra-low-power designs, every component in the power chain has an idle current that contributes to the battery drain floor.
• Assuming rated capacity equals usable capacity: a 2000 mAh cell derated for temperature, aged to 80% capacity, and operating at a higher-than-test C-rate may deliver only 1200–1400 mAh in the field.
• Not profiling edge case states: the "device performing OTA firmware update" state often draws 5–10× the typical active current for 10–60 seconds. If OTA happens daily, this can represent a non-trivial fraction of total energy consumed and must be included in the budget.