← Back to: Battery Charging / Gauging / Protection / BMS
Role in the Charging Branch (not Protection)
In the charging branch of a BMS, the isolated stack communication link exists to carry high-voltage, multi-cell measurement data from the floating battery stack to the low-voltage charging/power-path controller. It must survive dv/dt, common-mode noise and balancing activity around the charger, so we treat it as a dedicated communication layer, not as a generic vehicle bus and not as a pack-FET control line.
The service scope of this link is limited to the charging state machine — pre-charge, fast-charge, CV/taper and recharge. It enables the charger to decide whether to continue, to derate or to pause. It does not directly drive pack-FETs and it does not define system-wide alarm priorities.
In this page we describe only the bolded path of the full data flow: cell / AFE → isolated stack comm (high-CMTI, transient-suppressed) → charger / power-path controller → upper-level BMS. The AFE belongs to the measurement domain and exposes structured, synchronous, multi-cell data. The isolated comm layer only makes it safely available inside the noisy charging zone.
We emphasize “high-CMTI comms with transient suppression” because charger MOSFETs, power-path switches and balancing FETs all introduce sharp transitions. If we used low-CMTI digital isolators or non-protected CAN/SPI fronts here, the chain would show packet loss, false node-missing events or even total link stalls.
Out of scope for this page: pack-FET opening/closing logic, vehicle-level CAN/diagnostics, HV insulation algorithms and global BMS alarm tables.
Topology Options: isoSPI-like vs Isolated CAN-FD
Inside the charging branch we need an isolated link that is practical to build with TI, ST, NXP, Renesas, onsemi, Microchip and Melexis parts, and that still works when the charger area is noisy. For this reason we compare three topologies: (1) daisy chain for simple stacks, (2) dual-path/ring for higher availability, and (3) an isolated CAN-FD backbone when procurement and cross-brand replacement are priorities.
isoSPI-like solutions sit closer to the AFE side — they move “SPI-meaningful” frames across isolation. Isolated CAN-FD sits closer to the controller side — it reuses the mature automotive CAN-FD ecosystem (transceivers, TVS, surge, diagnostics) and is often easier to source in small batches.
Not covered here: wireless BMS schemes, vehicle/zone gateways, multi-master CAN arbitration or high-voltage comm isolation certifications. This page stays focused on isolated stack links for the charging domain.
Noise / Transient / CMTI Design Hooks for Charger Area
The charging zone is the noisiest place for this isolated stack link: charger MOSFETs, power-path transitions and even cell-balancing FETs create sharp dv/dt that easily pollute communication. Here we are not improving HV safety distances — we are protecting communication availability so the charger can keep running its CC/CV/taper logic.
Typical charging/balancing events can produce tens of kV/µs. A generic 25 kV/µs digital isolator may pass basic tests but will show sporadic frame errors here. For that reason, add a clear BOM note such as “High-CMTI (≥50 kV/µs) for charger dv/dt zone – do not substitute.” This prevents low-cost, low-CMTI parts from being dropped in by purchasing.
Transient/surge/ESD protection should be placed in three layers: (1) at the connector to dump energy, (2) at the transceiver or differential input to keep the comm device alive, and (3) before the isolator — often with an RC or CM choke — to remove the common-mode energy that would otherwise hit the high-CMTI isolator right where dv/dt is worst.
Out of scope for this page: full EMC/ESD certification flow, HV insulation test injection methods, and pack-level grounding/shielding strategy — those are handled in the system-level documents.
Chain-Integrity & Node-Presence Checks
A stack-style, daisy/ring isolated link inside the charging branch must always know who is online. When one node does not respond during CV or taper, the charger should first degrade charging and report upstream, not immediately trip the pack. That is how we keep this logic inside the charging domain and avoid taking over the protection page.
Integrity must be checked at two levels: (1) power-on discovery to bind logical order to physical order, and (2) runtime periodic checks (heartbeat, poll, echo) to catch temporary outages caused by noise or by balancing windows. CAN-FD has richer diagnostic fields; isoSPI-like stacks can do round-trip/echo to test whether the line is still continuous.
Not covered in this page: vehicle “limp-home” strategy, SOC/SOH derating logic and multi-pack master arbitration — those belong to system-level BMS pages, not to this charging-branch communication link.
Sync with Cell AFE / Balancing Activity
The cell-monitor AFE belongs to the measurement domain. It converts multi-cell, high-voltage, time-aligned data into a structured stream so the isolated stack link can hand it to the charging controller. This page stays inside the charging branch — we do not modify how the AFE measures and we do not take over pack-FET decisions.
Problem: balancing FETs create brief cell-voltage disturbances and the charger zone is already noisy (dv/dt from power-path, inrush, pre-charge). If the isolated chain reports during those disturbances, the charger may see fake OV/UV or fake missing-node. So we align balancing → sampling → comm slot.
Key alignment rules (charging branch only)
- Define a grouped polling schedule so not every module reports during balancing.
- During fast-charge, raise comm priority; during CV/taper, stretch the slot to avoid FET noise.
- Frames captured inside a balancing window → tag as soft-fault and forward with timestamp.
- Keep this logic in the charging controller; don’t let it promote itself to pack-FET control.
IC selection matrix — charging-domain isolated stack comm
Pick parts with ≥50 kV/µs CMTI, CAN-FD up to 5 Mbps, and automotive grade (AEC-Q100 / -Q1) whenever possible. Tell purchasing “do not down-bin to low-CMTI isolators”.
TI
Isolated CAN-FD, high CMTI, separate iso power.
- ISO1042BQDWVRQ1 — isolated CAN FD, 5 Mbps, AEC-Q100.
- ISO1044BDWVRQ1 — reinforced, improved transient.
- TCAN1042HGV-Q1 — HS CAN FD front for ring/daisy.
- SN6505BDBVR — push-pull iso supply for remote node.
Remark: “High-CMTI (≥50 kV/µs) isolator required; do not substitute.”
ST
PMIC + CAN-FD good for charger-side modules.
- SPSB0815-TR — SBC with CAN FD, LIN, watchdog. :contentReference[oaicite:0]{index=0}
- SPSB0813-TR — same family, CAN FD, fits compact boards. :contentReference[oaicite:1]{index=1}
- Use with ST automotive TVS (SMxJ series) close to connector.
Remark: place in low-noise corner; PMIC dissipation + isolator heat must not stack.
NXP
Mantis family CAN FD, easy to buy, automotive.
- TJA1042T,118 — CAN FD 5 Mbps, ISO 11898-2. :contentReference[oaicite:2]{index=2}
- TJA1044GT/3Z — CAN FD with standby/wakeup. :contentReference[oaicite:3]{index=3}
- TJA1044V — newer rev, same bus immunity. :contentReference[oaicite:4]{index=4}
Remark: add digital isolator in front when the AFE side is floating → becomes charging-domain safe.
Renesas
GMR isolator + CAN transceiver pair.
- ISL71710M — 50 kV/µs digital isolator. :contentReference[oaicite:5]{index=5}
- ISL71026M — CAN transceiver for harsh / isolated link. :contentReference[oaicite:6]{index=6}
- ISL72026SEH — up to 5 Mbps CAN, tolerant. :contentReference[oaicite:7]{index=7}
Remark: good where charger MCU is already Renesas and you want same-vendor isolation chain.
onsemi
Strong CAN FD + SBC parts for noisy zones.
- NCV7344 — CAN FD, low power. :contentReference[oaicite:8]{index=8}
- NCV7357-D — high-speed CAN FD for 5 Mbps. :contentReference[oaicite:9]{index=9}
- NCV7450 / NCV7446 — SBC with CAN FD, diagnostics. :contentReference[oaicite:10]{index=10}
Remark: pair with onsemi ESD/TVS at connector to survive charger dv/dt zone.
Microchip
Very available CAN FD, good for small batches.
- ATA6560-GAQW — CAN FD 5 Mbps, automotive. :contentReference[oaicite:11]{index=11}
- ATA6561 — standby/silent modes. :contentReference[oaicite:12]{index=12}
- Add MCP2562FD if you don’t need SBC features.
Remark: state in BOM: “place surge/ESD at connector due to charger dv/dt zone”.
Melexis
Use for node ID / presence / low-speed diag.
- MLX80002 — dual-channel LIN master for local diag. :contentReference[oaicite:13]{index=13}
- MLX80004 — 4-channel LIN for multi-module stacks. :contentReference[oaicite:14]{index=14}
- TH8056 — single-wire CAN for simple presence lines. :contentReference[oaicite:15]{index=15}
Remark: position as “who’s online” side-band, not the main high-rate charger link.
Small-Batch Procurement & Cross-Brand Alternatives
This chapter is for builds where the ideal automotive isolated CAN-FD device is not available locally, lead-time is unstable, or you must mix vendors. We stay inside the charging branch: the goal is to keep the stack link alive near a charger dv/dt zone, not to optimize cost or to rewrite pack-level protection.
Procurement tiers (top → fallback → de-rated)
-
Preferred route: buy an automotive isolated CAN-FD device with ≥50 kV/µs CMTI.
Examples (reference only, keep brand neutrality):- TI: ISO1042BQDWVRQ1, ISO1044BDWVRQ1
- NXP: TJA1044GT/3Z + external high-CMTI digital isolator
- Renesas: ISL71710M + CAN FD transceiver front-end
- onsemi: NCV7344 / NCV7357-D + TVS at connector
- Microchip: ATA6560-GAQW / ATA6561 + isolation in front
- ST: SPSB0815-TR / SPSB0813-TR in charger-side modules
- Melexis: use for node-ID/presence sideband, not main high-rate link
-
Fallback route: use a high-CMTI digital isolator (≥50 kV/µs) + a locally-available automotive CAN-FD transceiver from TI/ST/NXP/Renesas/onsemi/Microchip.
This is the most realistic small-batch path: isolator decides CMTI, transceiver decides bus ecosystem. - De-rated route (lab / pilot only): industrial-grade CAN-FD + very strong front-end (connector TVS + RC/CM choke + close-in surge) → must de-rate charging current and must increase comm retry window.
Cross-brand replacement logic
When you cannot get the exact IC from the original brand, do not match only on protocol name (e.g. “CAN FD”). Match on these four axes:
- CMTI: target ≥50 kV/µs for charger dv/dt zone. If lower, de-rate: If the part is not ≥50 kV/µs CMTI, de-rate the charging current and increase comm retry window.
- Automotive grade: AEC-Q100 / -Q1 / automotive variant preferred. If you must use industrial temp, mark the build as “pilot / lab”.
- Surge / transient capability: can it survive at the connector where charger MOSFETs inject noise?
- Isolation position: some vendors isolate first and transceive later; some do transceiver-then-isolate. Mixing is ok if CMTI and protection are explicitly stated in BOM.
BOM remark templates
Copy any of the following into your BOM or ERP so purchasing does not down-spec the link:
1) Charging dv/dt zone use: isolated comm must survive high common-mode transients.
2) High-CMTI (≥50 kV/µs) isolator required; do not substitute with low-cost digital isolators.
3) Do not substitute with non-automotive CAN-FD transceivers.
4) Place TVS at connector; add RC / CM choke before isolation stage.
5) If automotive grade (-Q1 / AEC-Q100) not available, mark build as pilot / low-volume only.
Distribution / sourcing notes for low volume
- Prefer distributors that can provide cut-tape / partial reel, so you do not over-buy isolation parts.
- For automotive-labeled ICs, check date code and packaging to avoid mixing industrial-temp stock.
- If you must mix TI/ST/NXP/Renesas/onsemi/Microchip/Melexis, document the weakest CMTI segment and force de-rated charging there.
Risk rules:
• Industrial-temp part instead of auto → allowed only for demo/pilot → reduce charging current.
• Low-CMTI part instead of high-CMTI → increase comm retry window and reporting timeout.
• Non-isolated part instead of isolated → lab/prototype only, not for real charger dv/dt zone.
Out of scope: pricing, specific distributor names, non-compliant/grey channels, and pack-level protection policies — those belong to business or system pages, not to this charging-branch comm page.
Request a Quote
Frequently Asked Questions — Charging-Branch Isolated Stack Comm
The following questions stay strictly inside the charging branch. They help you keep the link alive near charger MOSFETs, align it with balancing, and stop purchasing from downgrading CMTI or automotive grade.
Why do we need high-CMTI isolation for a charger-side stack bus?
Charger power-path and balancing FETs create dv/dt in the tens of kV/µs. A low-CMTI isolator will show sporadic frame errors. Using ≥50 kV/µs devices keeps the isolated stack communication available while the charger runs CC/CV/taper logic.
Can an isolated CAN-FD link replace an isoSPI-like stack bus in a 4–6S pack?
Yes, if you preserve the measurement-domain timing. Map the AFE’s sample window, then send the frame over isolated CAN-FD. If the AFE exports strict SPI frames, add a high-CMTI digital isolator in front of the CAN-FD transceiver.
How do I detect a missing or unresponsive stack node without stopping charging?
Run power-on discovery, then periodic heartbeat. If one node goes missing, first degrade charging (limit current/voltage, pause balancing on that cell). Only after repeated failures report to the global BMS — do not open pack-FET from here.
What transient/surge level should the interface survive near the charger power-path?
Use a 3-layer approach: TVS/ESD at connector, surge/CM filter near the transceiver, and a high-CMTI isolator behind it. Design for the worst dv/dt that your charger MOSFETs and balancing circuits can generate.
How to avoid cell-balancing FET activity from corrupting stack communication frames?
Align communication to “quiet” windows: balancing → short settle → AFE sample → comm slot. Mark any frames captured inside the balancing window as soft-fault and send them with timestamp to the BMS.
Can I share the isolated supply for both the comm link and the AFE measurement domain?
Possible, but not ideal in a noisy charging zone. Sharing can couple transients back into the AFE. Prefer separate isolated supplies or at least separate filtering stages so measurement stays clean.
What polling / refresh rate is recommended in CV / taper stage?
You can slow down polling once voltage is already high. Give balancing more room and let the comm link avoid FET noise. Keep discovery/heartbeat alive so the charger always knows who is online.
How should link faults be reported — local charger only, or global BMS?
First to the local charging controller so it can degrade safely; then to the global BMS for visibility and logging. This keeps the decision in the charging domain and avoids unnecessary pack trips.
Can I mix devices from TI/ST/NXP if isolation ratings differ?
Yes, but document the lowest CMTI segment and de-rate around it: lower charging current, longer retry windows, better TVS. Mixed-vendor chains must still place surge protection at the connector.
How to write BOM remarks so purchasing won’t downgrade isolators?
Add: “Charging dv/dt zone”; “High-CMTI (≥50 kV/µs) isolator required”; “Do not substitute with non-automotive CAN-FD”; “Place TVS at connector”. This blocks PC/industrial CAN from being dropped in.
When is a ring / dual-path topology preferred over a daisy chain?
Use ring / dual-path when node count is high, cables are long, or you need strong chain-integrity checks. It lets you identify which segment failed without bringing the whole charging branch down.
Does isolated CAN-FD need CM chokes or TVS near charger MOSFETs?
Yes. At least one TVS/ESD stage at the connector and one CM/choke stage before the isolator. This is what actually keeps the high-CMTI device from being hit by the raw dv/dt of the charger switches.
