The engineering decision
Define measurement conditions and acceptance bands for capacity, resistance, self-discharge and state of charge, then verify assembled-group behavior.
Have these inputs ready
- Cell identity and lot records
- Controlled SOC and temperature
- Capacity and resistance test method
- Storage interval and balancing capability
Equal voltage can hide unequal behavior
Scroll horizontally to view the full diagram →
State of charge is only one dimension
Cells can have similar open-circuit voltage while differing in capacity, internal resistance or self-discharge. On a flat voltage plateau, a small voltage difference may correspond to a substantial SOC uncertainty. Use voltage only within a defined rest, temperature and SOC procedure.
In series, a weak group can reach a voltage limit before the others. Parallel connections tend to equalize terminal voltage but do not guarantee equal branch current. Resistance in the cells and interconnections, plus temperature differences, affects sharing.
| Parameter | Control conditions | Why it matters |
|---|---|---|
| Capacity | Charge profile, discharge rate, cutoff, temperature | Available charge before group limits |
| DCIR | SOC, temperature, pulse amplitude and sample time | Loaded voltage and heat |
| ACIR | Frequency, fixture, temperature and SOC | Consistent screening metric |
| OCV | SOC history, rest interval and temperature | Assembly and screening state |
| Self-discharge | Rest/storage interval and repeated measurements | Abnormal charge loss |
| Identity | Model, lot, date and revision | Traceability and substitution control |
Estimate whether balancing can finish in time
For a hypothetical 30 Ah cell with a 2% charge imbalance, the charge difference is 0.6 Ah. A passive bleed path removing this 0.6 Ah excess at an effective 100 mA needs at least six hours with the bleed path enabled. Thermal scheduling or balancing enabled only near the top of charge makes elapsed time longer.
Passive balancing dissipates energy from selected cells. Voltage-based algorithms commonly select higher-voltage cells; their SOC interpretation depends on load, rest, temperature and chemistry. Balancing cannot restore lost capacity or reduce cell resistance. Active balancing transfers charge with finite current, efficiency and operating-window limits.
Build tolerances from the pack requirement
Choose screening bands after relating measured spread to minimum group voltage, temperature and usable energy under the mission. An instrument resolution is not an acceptance tolerance. Include repeatability, fixture resistance and measurement uncertainty before rejecting or accepting cells near the boundary.
Preserve the distribution and the rejected outliers. A batch average can hide a small tail of unsuitable cells. Group assignment should remain traceable to the source records so a later field outlier can be investigated.
Validate the matching method on assembled packs from more than one lot. Define what happens when the incoming distribution shifts: quarantine, additional tests, revised grouping or a formal design review.
Avoid misleading screening shortcuts
| Shortcut | What remains unknown | Corrective evidence |
|---|---|---|
| OCV only | Capacity and load behavior | Capacity and pulse tests |
| ACIR used as DCIR | Time-dependent voltage drop | Defined DC pulse measurement |
| Batch mean only | Outlier population | Distribution and individual records |
| Balancing used to fix all spread | Capacity or resistance defect | Root-cause testing before assembly |
Measurement & validation procedure
| Test / review | Record | Acceptance condition |
|---|---|---|
| Confirm method repeatability | Repeated readings with controlled fixture and conditions | Uncertainty is small enough for the chosen decision band |
| Check incoming distributions | Individual values, lot and rejected records | Approved matching limits are met |
| Validate assembled groups | Voltage/temperature spread over the duty cycle | No group reaches a limit prematurely |
Continue this engineering task
Related engineering guidance
Prepare the evidence for review
- Individual cell measurements and lot IDs
- Test procedure and fixture details
- Proposed matching tolerances
- Balancing current and available time
Technical sources & claim support
References support the stated engineering principles. Worked examples use their explicitly stated assumptions.
- Analog Devices: Passive Battery Cell Balancing ↗
Capacity, internal resistance, self-discharge and passive balancing
Why equal state of charge does not establish equal capacity or resistance.
- Analog Devices: Active Battery Cell Balancing ↗
Charge redistribution during discharge and charge
Active balancing can improve utilization within its current and energy-transfer limits.
- Keysight: How to Test Coin Cell Batteries ↗
Coin Cell Battery Testing and Modeling: four-wire sensing, pulse resistance, pulse width/rest time and defined cutoffs
Measurement principles illustrated with coin cells. Select the current profile, fixture, sampling, temperature and cutoff conditions for the actual cell or pack.
- NIST: Metrology and process control, dealing with measurement uncertainty ↗
Measurement uncertainty and manufacturing risk
Basis for considering uncertainty in acceptance decisions.
- Texas Instruments BQ76952 Technical Reference Manual, SLUUBY2B ↗
Protections; overcurrent in discharge; short circuit; FET control
Threshold, delay, recovery and configuration behavior must be checked for the actual device revision.
- Analog Devices: Battery Stack Monitor Maximizes Performance of Li-Ion Batteries ↗
Cell voltage, current and temperature monitoring; wiring, EMI and connector reliability
Architecture guidance; validation remains specific to the pack, harness and operating environment.
Key claims and evidence
| Claim | Basis | Reference |
|---|---|---|
| Capacity, resistance and self-discharge can differ between similar cells | documentation | Analog Devices: Passive Battery Cell Balancing |
| Active balancing redistributes charge | documentation | Analog Devices: Active Battery Cell Balancing |
| 0.6 Ah at 0.1 A takes at least 6 hours | calculation | calculation |
| Matching bands must be derived from application and measurement capability | workflow | NIST: Metrology and process control, dealing with measurement uncertainty |