Automotive BMS signal chain: cells, current, isolation and communication

1. Draw measurement boundaries before choosing parts

Mark each cell module's floating reference, pack negative, low-voltage MCU ground and vehicle-network ground separately. A cell AFE measures a local series-cell segment; it is not a single ADC connected directly across the entire high-voltage pack. Pack-level nodes generally need divider or shunt networks, while communication between ground domains needs its own isolation design.

The table compares responsibilities, not a fixed BOM containing every device. For example, a dedicated isolated daisy chain does not automatically need an additional ISO7741-Q1. Whether logic isolation is needed depends on the MCU and bridge power domains.

Scroll wide tables horizontally; column headers remain visible within the table.

2. Choose cell monitors by module count, temperature and balancing heat

Module boundaries often determine the AFE before total pack cell count does. BQ79612-Q1, BQ79614-Q1 and BQ79616-Q1 have upper limits of 12, 14 and 16 cells, all with a six-cell minimum; TI's family-reuse statements have a defined scope. BQ79606A-Q1 instead supports 3–6 cells and is Grade 2. Do not extend another device's Grade 1 rating or software-reuse claim to it. Check maximum ambient temperature at the installation point before optimizing channel utilization and harness layout.

BQ79718B-Q1 publicly specifies 9–18S, 300mA balancing and typical ±1mV ADC accuracy. A typical value is not a maximum error over temperature. Public document SLVSJM4, dated August 2025, is abridged; the full datasheet requires a request to TI. It is a candidate for an 18S module, but electrical conditions, pin details and diagnostics must be obtained before hardware freeze or any replacement claim for BQ79616-Q1.

The 240mA or 300mA balancing figures are device capability entries, not a guarantee that every channel can remain on simultaneously at any board temperature. Estimate total discharge-path heat per channel using P≈Vcell×Ibalance, then separate resistor loss from internal-switch loss. Validate concurrent channels, copper area, enclosure thermal resistance and thermal-pause duty cycle. Thermal pausing protects the device but can extend the time needed to reach the target cell mismatch.

3. Separate shunt current, pack voltage and isolation ranges

INA240-Q1 measures the differential drop across a shunt, but its input common-mode range is only −4V to 80V and it provides no galvanic isolation. It cannot connect directly to a high-side shunt on a high-voltage traction pack. Even at a low-side or other in-range node, review ground movement, fault transients, output swing and downstream ADC reference. The family has both Grade 1 and Grade 0 variants; temperature and package must be checked against the complete OPN rather than assigning one grade to every variant.

AMC3330-Q1 has a ±1V high-impedance input intended for voltage signals such as a high-voltage resistor divider, not a current amplifier optimized for a small millivolt shunt drop. Its integrated isolated DC/DC permits a single low-side supply, but does not remove divider voltage-rating, dissipation, creepage or fault checks. The current product page's feature and parameter sections disagree on CMTI and offset figures, so this guide does not select one as a guaranteed value. Freeze the datasheet revision and test conditions before building the detailed error budget.

For integrated BJB measurement, BQ79731B-Q1 provides 17 single-ended voltage channels and two independent low-side shunt-current ADCs with a ±275mV current-input range. Measuring high-voltage nodes does not mean its pins connect to the full pack without dividers. Two ADCs also do not establish two validated independent safety channels: common-cause failures from a shared shunt, supply or reference still need analysis.

As a range-sizing illustration only, 500A through a 100µΩ shunt produces 50mV and dissipates 25W. Being below ±275mV only suggests the nominal amplitude may suit BQ79731B-Q1; it does not validate pulse overload, drift, thermal design or layout. After choosing the shunt, also evaluate zero-current offset and low-current resolution, because a suitable high-current range can still accumulate coulomb-counting error during rest.

4. Validate daisy chain, logic isolation, isolated power and CAN separately

BQ79600-Q1 converts MCU UART/SPI traffic into TI battery-monitor daisy-chain communication and returns responses to the MCU. It neither adds cell channels nor performs cell ADC conversions. A monitor's SPI-controller function and a bridge's SPI host interface are different roles; seeing SPI in both parameter lists does not justify connecting them or assuming matching protocols.

In the optional BQ7961x ring, accessibility after a cable break still depends on host fault detection, direction switching and correct addressing in both directions. TI's SLVAEP4 describes direction changes and re-addressing. Test an open at each connector and record reachable modules and data timestamps. A ring does not automatically meet the fault-response deadline or guarantee complete data under every multiple-fault condition.

ISO7741-Q1 is a three-forward/one-reverse digital isolator without integrated isolated power. SN6505B-Q1 is a push-pull driver requiring an external transformer, rectification and associated circuitry, not a self-contained certified isolated DC/DC module. ISO1042-Q1 implements the isolated CAN physical layer; it still needs a CAN controller and supplies on both sides and does not replace BQ79600-Q1. List continuous working voltage, surge, layout spacing and power-loss defaults for each isolation boundary rather than relying only on a withstand-test number.

5. Connect sampling time, diagnostics and the final protective action

An SOC algorithm needs voltage, current and temperature together with acquisition time, filter delay and validity flags, not isolated ADC numbers. BQ79656-Q1 supports an option to synchronize current and cell-voltage measurement; public information for BQ79718B-Q1 and BQ79731B-Q1 also lists voltage/current synchronization capabilities. Verify the selected combination's trigger and communication requirements rather than assuming arbitrary mixed devices synchronize automatically.

Build separate timing budgets for estimation and protection. Low-pass filtering can improve steady readings but cannot substitute for a response to fast overcurrent or an open connection. Include acquisition, filtering, link retries, host scheduling, fault qualification and external-actuator response. Fault injection should demonstrate the required system safe state, not stop when a fault bit is read.

AEC-Q100 is component qualification information. Functional Safety-Compliant and Functional Safety-Capable also represent different documentation and development support. The verified public BQ79731B-Q1 source states AEC-Q100 qualification and −40°C to +125°C but does not explicitly state a numeric Grade; this guide does not infer one from temperature. The ASIL target still requires the safety concept, applicable safety manuals, FMEDA assumptions and system validation.

6. Schematic-review and prototype-acceptance checklist

The following are proposed engineering checks based on the device boundaries above, not a TI compliance promise for an arbitrary BMS. Give each check reviewable inputs, acceptance criteria and measured evidence. Where full documentation is access-controlled, keep the affected design decisions behind a release gate.

• List every input voltage relative to local ground in normal and fault states; distinguish differential range, common-mode range and isolation working voltage.
• Check minimum/maximum cell configuration, unused inputs, NTC wiring and busbar connections against actual modules, not just total pack count divided by channels.
• Record divider/shunt tolerance, drift, fault dissipation and Kelvin sensing; test zero, full scale and expected fault transients separately.
• At the hottest expected condition, validate concurrent balancing channels, board temperature and automatic pause/resume; record actual balancing completion time.
• Open each daisy-chain segment and test reverse communication, CRC/timeouts, stale-data rejection and the fault-wakeup path when the MCU/power system sleeps.
• Verify isolated-power startup/shutdown order, digital-isolator defaults, CAN-side power loss and the safe state of external actuators.
• Freeze full OPNs and document revisions. BQ79718BQPAPRQ1 and BQ79731BQPHPRQ1 map to the B-suffixed GPNs; retain B and do not claim drop-in replacement from similar packages.
• Check safety-manual assumptions, diagnostic intervals and actuator response time individually; sign off component qualification, safety-document availability and system certification separately.

Frequently asked questions

Related parts and complete OPNs

Continue reading