4S Lithium-Ion BMS Design
Worked for 4 mins 42 sec
The 1 kΩ sense resistors are now in the schematic and placed beside U1. I’m cleaning stale routing on the affected nets before running the router again.
Worked for 23 mins 17 sec
Implemented the approved RC filter fix.
What changed:
- Added RVC1–RVC4, each 1 kΩ, in series with the four BM3451 cell-sense inputs.
- Rewired the sense inputs so B1–B4 raw taps feed filtered VC2–VC5 nodes through the new resistors.
- Kept the existing 1 µF CVC1–CVC4 capacitors on the filtered side, completing the intended 1 ms RC input filters.
- Placed RVC1–RVC4 next to U1 in the cool logic zone.
- Removed stale routing on the modified nets and re-ran the autorouter.
- Cleaned the overlap and dangling-trace issues introduced during rerouting.
Final audit:
- ERC: no open errors/warnings in the checked categories.
- DRC: no overlapping copper, invalid layers, dangling traces, missing footprints, or protected-intrusion errors.
- Remaining: 2 airwire errors still remain, tied to the same auto-generated fill/stitching artifact seen previously. The normal component routing for the updated RC filter paths has been completed, but the project is still not formally “zero-DRC” until those fill-related airwire artifacts are resolved or accepted.
Fix airwire artifacts
Verify RC routing
Check isolation barrier
Run manufacturability review
Prepare export package
Refine this doc
Ask about this doc
Input Filter Transient Analysis — BM3451 4S BMS
Scope
This report evaluates the BM3451 cell-voltage input filtering and protection-delay behavior for the Toyota Prius 4S CATL NCM module BMS/CMU design.
Datasheet Reference Values
From BM3451 Table 6 external component constants:
Table
| Datasheet Item | Typical | Range / Note |
|---|---|---|
| R1–R5 cell input series resistors | 1 kΩ | 100 Ω to 1 kΩ |
| C1–C5 cell input capacitors | 1 µF | 0.1 µF to 10 µF, voltage rating >50 V |
| RVCC | 1 kΩ | 100 Ω to 1 kΩ |
| CVCC | 10 µF | 10 µF to 100 µF |
| COV / COVD | 0.1 µF | Sets OVP / UVP protection delay |
BM3451 protection delays from the datasheet:
Table
| Delay | Formula / Condition | Result with current values |
|---|---|---|
| OVP delay TOV | 10^7 × COV | 1.0 s typ, datasheet range 0.5–1.5 s |
| UVP delay TOVD | 10^7 × COVD | 1.0 s typ, datasheet range 0.5–1.5 s |
| Overcurrent 1 delay TOC1 | 2 × 10^6 × COC1 | 0.2 s |
| Overcurrent 2 delay TOC2 | 2 × 10^5 × COC2 | 20 ms |
Current Schematic Finding
The schematic currently has the 1 µF cell filter capacitors CVC1–CVC4 across the cell sense nodes, but it does not show the per-cell 1 kΩ series resistors R1–R4 between the M4 sense pads and the BM3451 VC inputs.
Because of that, the current schematic does not implement the full datasheet RC input low-pass filter on VC1–VC4/VC2–VC5. The capacitors still provide local charge storage/noise shunting, but the effective time constant depends on harness impedance, trace impedance, and cell source impedance rather than a controlled resistor value.
Intended Datasheet RC Time Constant
If the datasheet-recommended 1 kΩ series resistor is inserted in each cell sense input path with the existing 1 µF capacitor:
- τ = R × C = 1,000 Ω × 1 µF = 1.0 ms
- f_c = 1 / (2πRC) = 159 Hz
Step/pulse response fraction at the VC input:
Table
| Spike / transient duration | VC input reaches approx. | Interpretation |
|---|---|---|
| 1 µs | 0.10% of spike amplitude | Strong suppression |
| 10 µs | 0.995% of spike amplitude | Strong suppression |
| 100 µs | 9.5% of spike amplitude | Strong attenuation |
| 1 ms | 63.2% of step amplitude | One time constant |
| 5 ms | 99.3% of step amplitude | Essentially settled |
The VCC filter RVCC = 1 kΩ and CVCC = 10 µF gives:
- τ = 10 ms
- f_c = 15.9 Hz
Automotive Transient Interpretation
A 1 ms cell-input RC filter is appropriate for rejecting high-frequency inverter noise and short microsecond-scale regenerative braking or traction inverter spikes. Microsecond events are attenuated by roughly 100× to 1000× before they reach the BM3451 comparator inputs.
This RC delay is still far shorter than the BM3451 OVP/UVP qualification delay of about 1 second. Therefore, the RC filter does not materially slow genuine fault detection: a real OVP/UVP condition lasting hundreds of milliseconds to seconds will pass through the RC filter and then be qualified by the BM3451 internal/external timing network.
Recommendation / Status
As currently drawn, the input filter stage is not fully tuned because the per-cell series resistors are missing. To make the design datasheet-complete and automotive-robust, add 1 kΩ series resistors in each active cell sense path:
- B1 → 1 kΩ → BM3451 VC2 node, with 1 µF capacitor from filtered VC2/B1 sense node to lower cell node
- B2 → 1 kΩ → BM3451 VC3 node, with 1 µF capacitor from filtered VC3/B2 sense node to lower cell node
- B3 → 1 kΩ → BM3451 VC4 node, with 1 µF capacitor from filtered VC4/B3 sense node to lower cell node
- B4 → 1 kΩ → BM3451 VC5/VCC input path, with existing RVCC/CVCC retained for IC supply filtering
After adding those resistors, the filter network will match the BM3451 datasheet typical values and will be suitable for the dynamic Toyota Prius automotive environment: strong microsecond spike rejection, stable inverter-noise filtering, and no meaningful delay penalty against real OVP/UVP faults because BM3451 already uses a ~1 s fault qualification delay.
Welcome 👋
Flux helps you build PCBs faster with an AI teammate!
Create your account to collaborate, stay updated, fork your own version, and get instant answers from our AI agent.