BMS-03 — Cell Voltage Measurement

1. Why You Have to Measure Every Single Cell

In a 96-series-cell pack, total voltage sits around 355 V. Even if one cell breaches its 4.2 V limit and climbs to 4.3 V, that’s only a ~0.1 V shift (about 0.03%) at the pack level — something a pack-level measurement will never catch. Overvoltage/undervoltage protection, SOC estimation (via OCV), and cell balancing all need per-cell voltage data; the pack average simply isn’t enough.

2. The Measurement Chain: From Cell to BMU

Cell → sense wire → AFE (MUX + ADC) → digital data → BMU

Each cell’s terminals connect to an AFE (Analog Front End) / Cell Monitoring IC through thin sense wires. Putting a dedicated ADC on every cell would be expensive; instead, a MUX (multiplexer) routes every cell channel to a single ADC in sequence. An AFE typically monitors 6–18 cells — for a 96s pack with 12-cell AFEs, you’d need 96/12 = 8 AFEs, each measuring its own region and forwarding the data to the BMU.

3. How Accurate Does This Need to Be?

Overcharge protection typically needs precision in the tens-of-millivolts range. Say you want to hold the 4.2 V limit within ±10 mV; the measurement chain’s total error (offset + gain + thermal drift + noise) has to fit inside that budget:

Total error ≈ |Offset| + |Gain error| + |Thermal drift| + |Noise|

If offset is ±4 mV and gain error is ±3 mV, that leaves only ~±3 mV for thermal drift and noise combined — meaning every error source has to be budgeted individually.

ASSUMPTION — Real designs usually combine these errors statistically (RSS — root-sum-of-squares); the simple addition above is only meant to illustrate a conceptual upper bound.

4. Common-Mode Voltage and Isolation

In a 96s pack, the topmost cell’s reference point can sit close to the full pack voltage (~350 V). Every AFE effectively “floats” relative to its own region’s reference — the bottom one operates near 0 V, the top one near 340 V. Data from these different reference levels has to reach the BMU’s common (LV) reference safely, which is why AFE-to-AFE communication has to be isolated. Isolation does two jobs at once: it keeps high voltage from leaking into LV electronics and the chassis, and it keeps the measurement accurate.

5. Why Timing Matters

All cell voltages need to be sampled at the same instant while current is changing. If one AFE samples cell 1 at t=0 and a neighboring AFE samples cell 13 at t=5 ms, and current shifts suddenly within that 5 ms window, the measured difference between the two cells may partly reflect timing skew rather than a real SOC/internal-resistance difference — which can lead to incorrect balancing decisions.

6. The Cost of Daisy-Chaining

Multiple CMUs are typically chained together with daisy-chain wiring — it cuts down on cabling, but a break anywhere in the chain takes out data from every CMU downstream of it. That’s why fault handling and timeout design are essential.

7. When Things Go Wrong

Sense wire resistance, connector contact resistance, ADC offset/thermal drift, and channel-to-channel crosstalk are the main error sources. When a sense wire goes open, the affected channel usually shows an implausible value (0 V, full-scale voltage, or excessive noise); the BMS catches this by comparing it against neighboring cells (plausibility, BMS-13). If a channel gets flagged “no data,” the BMS assumes the worst case for that cell, flags the affected chain as a warning, and limits power or disconnects HV if needed.

8. How It’s Verified in Production

Every AFE channel is calibrated against known reference voltages; EOL testing confirms that every channel maps to the correct cell and meets the accuracy budget (BMS-20). HIL testing injects scenarios like a broken sense wire or lost AFE communication to verify the BMS reacts correctly (BMS-19).

9. How It Connects to Other Systems

Cell voltage data feeds directly into SOC estimation (BMS-08, as the OCV input), cell balancing (BMS-11), and fault management (BMS-13, overvoltage/undervoltage detection). SOP calculation (BMS-10) also relies on instantaneous cell voltages.

Summary

• Cell voltage is critical for protection and SOC — the pack average isn’t enough.
• The AFE measures, the ADC digitizes, and the daisy-chain relays the data.
• Common-mode handling and isolation are mandatory for HV/LV safety.
• The accuracy budget is set by how precise the protection limits need to be; calibration and EOL testing guarantee it in production.

Sources

• Gregory L. Plett, Battery Management Systems, Volume I.
• AFE/Cell Monitoring IC manufacturer application notes — general measurement concepts.