BMS-04 — Temperature Measurement

1. Why Measure Temperature?

Temperature affects three things at once: safety (overheating, thermal runaway risk), performance (power drops in the cold), and lifespan (degradation accelerates). That’s why the BMS has to track not just “cell temperature” but the distribution across the pack — the average alone isn’t enough, because a hot spot can be a real risk even while the average looks fine. These differences usually come from cells near vs. far from cooling, or from heat generated at high C-rate not spreading evenly.

2. NTC or RTD?

	NTC Thermistor	RTD (Pt100/Pt1000)
Operating principle	Resistance falls as temperature rises	Resistance rises ~linearly with temperature
Accuracy/linearity	Moderate, needs calibration/lookup tables	High
Cost	Low	Higher
Typical use	Production EV packs	Lab/calibration measurements

An NTC is measured by building a voltage divider with a known series resistor:

V_measured = V_reference × R_NTC / (R_NTC + R_series)

For example, with V_reference = 5 V, R_series = 10 kΩ, and R_NTC = 10 kΩ at 25 °C, V_measured = 2.5 V. The ADC reads this voltage, back-calculates R_NTC, then converts it to temperature using the manufacturer’s resistance-temperature table (or the Steinhart-Hart equation).

ASSUMPTION — Steinhart-Hart coefficients are sensor-specific; this course doesn’t provide a concrete coefficient set — a real design pulls them from the manufacturer’s datasheet.

3. Where Do You Put the Sensor?

Sensor placement determines how well the reading represents the cell’s real internal temperature — the cell core can run hotter than the surface. In a pack like 96s2p with only 1-2 sensors per module, the temperature of cells without their own sensor is only estimated from the nearest one; the fewer sensors you have, the more uncertain that estimate gets, forcing more conservative (earlier) derating. A single sensor isn’t enough in a pack with a gradient — the BMS needs to evaluate the hottest and coldest points separately.

4. Why Sensor Lag Matters

Thermal response describes how quickly a sensor catches up to a change in cell temperature; the thermal contact/mass between sensor and cell sets this lag. During fast events like rapid charging or a sudden load, a lagging sensor can read below the real temperature — which is why the BMS also watches current/voltage anomalies alongside temperature.

5. When Things Go Wrong

Sensor open/short circuits, connector contact issues, and poor thermal contact are the main error sources. An open circuit usually shows up as an “out of range” reading — for instance, an NTC reading -40 °C is probably open, because the voltage-divider output locks near the full-scale reference. The BMS flags such a channel invalid, temporarily substitutes the average of neighboring sensors or the last valid reading, and logs the fault (BMS-13); if an entire module’s data is lost, power gets restricted.

6. How It’s Verified in Production

Temperature channels are calibrated against known reference resistors or in controlled temperature chambers; EOL testing confirms every channel maps to the correct sensor. HIL testing injects open/short-circuit scenarios to verify the BMS’s response (BMS-19).

7. How It Connects to Other Systems

Temperature data directly feeds thermal management’s cooling/heating requests (BMS-12), SOP’s temperature limit (BMS-10), SOH modeling (BMS-09), and balancing decisions (BMS-11).

Summary

• NTC is common for cost/space reasons, RTD wins on accuracy.
• Placement and thermal contact determine how well a reading represents reality.
• A single sensor isn’t enough — the pack’s gradient has to be tracked.
• Calibration and EOL testing guarantee measurement accuracy in production.

Sources

• Gregory L. Plett, Battery Management Systems, Volume I — temperature and thermal modeling.
• Sensor manufacturer datasheets — NTC/RTD characteristics (general concept).