Measuring Battery Capacity: Coulomb Counting, Offset, and Cutoff

Begin With a Reproducible Test Boundary

The Battery Capacity and Pack Sizing Guide explains how to use nominal Ah and Wh in a design. This guide answers a different question: how much charge did an assembled battery actually deliver in a specified test? Record chemistry and part identification, charge procedure, rest period if applicable, initial state, discharge current profile, cell and pack temperatures, cell-level or pack-level voltage cutoff, BMS trip condition, and measurement boundary. A pack stopped by its BMS before the cell-level limit has measured system-delivered capacity for that condition, not necessarily the cell's laboratory capacity.

Use a qualified test setup and manufacturer-specified voltage, current, and thermal limits. Do not force a cell below its approved cutoff to obtain a larger number. A constant-current discharge makes comparison straightforward; a product load profile may better represent actual service. State which one was used. Compare repeated tests only when starting and ending conditions are equivalent.

Integrate Current, Not a Capacity Label

For a genuinely constant 2 A discharge lasting 3 h, delivered charge is 6 Ah, or 21,600 C. If voltage falls through the test, energy is not exactly 6 Ah times either the initial or final voltage; integrate V(t)I(t) or use a justified average voltage. The Battery Capacity Converter translates a measured charge into Ah, mAh, microampere-hours, and coulombs. It cannot perform the current measurement, integrate a waveform, or validate the cutoff.

Build the Current-Measurement Chain

A shunt-based system measures a small voltage and converts it with a characterized resistance; Hall or other sensors have their own offset and gain behavior. Choose sensor range so peak current does not clip while low-current resolution remains useful. Document whether charger, BMS, balancing, auxiliary electronics, and external loads pass through the same sensor. Any unmeasured current path makes the integrated charge incomplete. Check shunt heating and temperature coefficient, amplifier offset and drift, ADC reference, current direction, timestamp source, and missing samples.

Take a zero-current reading under a verified no-current condition and repeat over the intended temperature range. Zero adjustment removes an observed offset at a moment, not all temperature drift or time variation. A static 1 mA offset integrated for 24 h creates a 24 mAh error; for a 100 mAh device this is material. Conversely, measuring only fast high-current pulses with a slow or poorly synchronized logger can miss charge even when the average displayed current looks plausible. Integrate using actual timestamps and characterize the instrument's bandwidth relative to the load profile.

Example Error Ledger

Suppose a 1.0 A constant-current test ends after 2 h. Ideal integrated charge is 2.0 Ah. A +2 mA offset adds 0.004 Ah, or 4 mAh, while a 1% gain error changes it by about 20 mAh. These are independent first-order examples, not a complete uncertainty estimate. A wrong cutoff or incomplete current path may dominate both. Preserve raw samples and timestamps so any correction is auditable.

Separate Capacity, State of Charge, and Runtime

Delivered Ah between two endpoints is not automatically a full rated capacity. A test starting at 80% state of charge or stopping at a conservative application cutoff measures only that interval. Coulomb counting can track *change* in charge relative to an initial estimate, but it does not independently establish the initial state of charge, usable capacity at another temperature, or aging state. Measurement offset accumulates over time, self-discharge may be outside the sensed path, and capacity changes with operating conditions. Gauge systems therefore need qualified initialization, periodic correction, and validation against the intended cell and pack behavior.

The Charging and Discharge Validation Guide treats SOC windows, charge/discharge efficiency, and runtime modeling. Keep measured delivered charge, measured energy, modelled runtime, and nominal label values in separate report columns. A unit converter does not make these physical quantities interchangeable.

Practical Test Sequence

1. 1. Define the product's safe start, load, temperature, and stop conditions.
2. 2. Confirm all battery current crosses the measurement boundary.
3. 3. Calibrate current gain and characterize zero offset at relevant temperatures.
4. 4. Log synchronized current, terminal voltage, temperature, and timestamps.
5. 5. Integrate actual sample intervals without rounding intermediate values.
6. 6. Record whether cell voltage, pack voltage, BMS, or elapsed time ended the test.
7. 7. Convert the final charge units only after integration.
8. 8. Repeat and compare results at controlled rate, temperature, and age.

Engineering Sources

Texas Instruments current-measurement discussion explains charge integration and accumulating offset error. Analog Devices layout note shows that physical sense routing can affect current offset. These sources describe measurement mechanisms, not universal cell discharge limits; use the selected manufacturer's cell and BMS instructions for safe test conditions.