Topic 14 - Mastering Li-ion Cell Balancing and Monitoring

BMS Architecture, Balancing, and State Estimation · Battery Shortcut

Section 4 · Topic 14

BMS Architecture, Balancing, and State Estimation

A multi-cell pack fails catastrophically without a battery management system. The BMS is the brain of the pack — enforcing the safe operating area, managing thermal limits, and estimating the states no sensor reads directly.

Topologies

How the BMS connects to the cells is the first design choice, trading wiring complexity against cost and reliability.

CENTRALIZED BMS 1 board large harness single point of failure MODULAR S S MASTER slave per module cleaner wiring scalable, serviceable DISTRIBUTED CTRL IC per cell (daisy chain) no wire harness max component count
Fig. 4.14 — Three ways to wire a pack. Centralized: one board, a large sense harness, a single point of failure. Modular: a slave board per module reporting to a master — cleaner and serviceable. Distributed: a monitoring IC on each cell over a daisy chain — no harness, but maximum component count.

Balancing

Manufacturing tolerance guarantees that series cells differ in capacity, so the BMS must prevent the lowest-capacity cell from being overcharged as the others fill — and from being driven into over-discharge reversal as they empty. Passive balancing bleeds charge from the highest cells through shunt resistors as heat (50–200 mA); active balancing moves charge from high cells to low cells through capacitors or inductors (1 A+), losing little as heat but costing far more in circuitry.

BalancingMechanismCurrentTrade-off
PassiveShunt resistors bleed charge as heat50–200 mACheap, simple; wastes energy
ActiveCaps/inductors move charge cell-to-cell1 A+Efficient; complex, expensive

State estimation

The charge inside a cell cannot be measured the way a fuel level can — it must be inferred. State of charge by coulomb counting integrates current over time, accurate short-term but drifting as sensor error accumulates; Kalman filtering corrects that drift continuously by blending the count with real-time voltage and internal-resistance measurements against a cell model. State of health tracks degradation by comparing present full capacity against the original nominal capacity, or by following the steady rise in direct-current internal resistance (DCIR). At ~70–80% SoH the cell is treated as end-of-life for its primary application.

Deeper in Section 5

The estimator hardware — four-wire Kelvin shunts, NTC thermistors, pre-charge contactors — and the Extended and Unscented Kalman Filter variants are developed in Section 5, where the model-based charging that depends on them is covered.

Section 4 · Charging & Discharging ProtocolsTopic 14 / 23

Comments