← Back to: Battery Charging / Gauging / Protection / BMS
Intro & Scope: Why HV Aux Bias Is a Separate Topic
High-voltage battery systems in EV/HEV platforms often sit at 200–450 Vdc during charging, but the BMS “brain” (measurement AFE, MCU, logging, isolated comms) only accepts 5 V or 3.3 V. If this low-voltage brain is powered from the charger’s own auxiliary or from a non-isolated low-voltage rail, it will lose power as soon as the charger/bus sags, and critical charge-interruption events can no longer be logged. That is why the HV aux bias must be treated as its own topic.
In this page we keep the discussion strictly inside the charging branch of your BMS template: we only talk about an isolated, high-voltage-tolerant auxiliary supply whose job is to feed the BMS measurement and control domains and to stay alive long enough to write logs when the bus dips.
What we must write here
- Typical HV charging bus: 200–450 Vdc (sometimes 250–420 Vdc in real cars).
- Why reusing OBC aux rails is unsafe for logging: OBC turns off first → BMS “forgets”.
- Isolation level we expect: functional → basic → reinforced, automotive envelope only.
- Measurement-domain constraints: Y-cap, transformer CM path, and layout all change leakage readings.
- Why we need multiple outputs: 5 V for AFE, 3.3 V for MCU/logging, an extra bias for isolated comm.
- Why this lives under Charging: charging is exactly when the HV bus dips but logs must be kept.
What we do not write here
- No OBC / PFC / LLC main power stages.
- No pack contactor / pack-FET drivers.
- No HV insulation / leakage detection algorithm itself — only the power that feeds it.
Topology Choices: Isolated Flyback vs Push-Pull (for BMS Aux)
Once the scope is fixed — “isolated HV aux for the BMS, in the charging domain” — the next question is which topology we allow. For most BMS auxiliary rails in the 5–15 W band, an isolated flyback with secondary regulation is the default choice. Push-pull is reserved for cases where you already have a matched transformer, you need multiple tightly regulated outputs, or you want a cleaner EMI profile. PSR (primary-side regulated) solutions are discouraged because they introduce output drift that can corrupt measurement-domain accuracy.
What to explain
- Power window: most BMS aux rails stay within 5–20 W → flyback first.
- Isolation drives transformer choice: automotive isolation → often custom transformer.
- HV bus can dip: 250 → 420 V ranges; topology must remain stable and not restart.
- Why PSR is not liked: measurement-domain wants stable 5 V / 3.3 V → go secondary-regulated.
- When push-pull is valid: multi-output, tighter EMI, existing automotive controller, symmetric winding.
What we do not include
- LLC, half-bridge, PSFB → those belong to OBC/high-power DC/DC pages.
- Full synchronous bridges → only a short mention if paired with push-pull.
- Traction / body domain supplies → out of this charging scope.
With this chapter in place, later sections (Isolation & Safety, Secondary Regulation, Procurement) can simply reference: “use isolated flyback with secondary regulation unless a project-specific push-pull is already qualified.”
Isolation & Safety Envelope (Creepage / Clearance / Grade)
This chapter explains why low-cost consumer/offline SMPS controllers cannot be dropped in to replace an automotive HV aux bias in the BMS charging branch. The BMS sits on a 200–450 Vdc bus, feeds the measurement domain, and must keep logging even when the charging/OBC front-end drops out. That makes isolation grade, creepage, clearance and hi-pot testing non-negotiable and visible in the BOM.
We define the isolation stack here first so that later chapters (multi-output, procurement, cross-brand substitutions) can simply refer to this page: “comply with Chapter 3 isolation envelope.”
What must be defined
- Isolation levels: functional / basic / reinforced → this page: basic minimum, reinforced recommended.
- Creepage & clearance: calculated from highest pack/bus voltage and pollution degree.
- Hi-pot / dielectric test points: both on transformer and on the final assembly.
- Controller class: “DC bus only / automotive HV bus capable” — not generic AC offline.
- Regional differences: EU / US / China may push creepage higher → write it in the BOM remark.
Procurement remark (can be copied)
“Auxiliary HV isolated supply for BMS shall provide automotive-grade isolation at the bus voltage level, with creepage/clearance per project region (EU/US/CN). Consumer offline SMPS controllers shall not be used as replacements. If PCB creepage cannot be met, it shall be achieved by transformer construction.”
What we do not cover here
- No full-vehicle HV harness design.
- No vehicle-level insulation fault algorithm.
- No OBC/PFC power stage design.
For EU/US/China projects, specify the target region in the BOM remark so that PCB-level creepage/clearance can be backed up by transformer insulation when the board cannot meet the distance. This prevents purchasing from dropping in non-automotive offline controllers.
Secondary Regulation & Multi-Output Bias Rails
BMS auxiliary rails power the most sensitive part of the system: AFEs sampling high-voltage cells, MCUs logging charge events, and isolated communication. These circuits cannot tolerate 1–2% output drift from primary-side regulation (PSR). That is why this page requires secondary-regulated flyback with opto-isolated feedback and a post-regulated clean rail (typically 3.3 V) for measurements and logging.
Architecture to write
- Main output: 5 V, closed-loop, secondary-sensed, opto/TL431.
- Clean output: 3.3 V (or 1.8 V) made by an LDO from the 5 V to isolate AFE/Logger loads.
- Load events: AFE may pulse current during insulation/leakage measurement → LDO decouples MCU/log.
- Automotive environment: optocoupler CTR and reference drift over temperature → must be qualified.
- Purchasing rule: “if controller/opto/reference is substituted → re-validate AFE accuracy.”
PSR hazard to call out
“Primary-side regulated converters are not acceptable for BMS measurement-domain auxiliary rails because transformer tolerance, layout coupling and HV-bus noise all produce output drift. Any PSR-based replacement shall trigger measurement-domain re-validation.”
To keep the BMS measurement-domain stable, add to your purchasing notes: “secondary-regulated flyback / opto-isolated feedback / TL431 reference / post-regulated 3.3 V for AFE.” Any change to these parts → re-run AFE accuracy and leakage measurement validation.
Ride-Through / Hold-Up for Logging
During real charging events — unplug, contactor lifting, short HV bus valleys, or charger hiccups — the auxiliary input for the BMS also sags. If the BMS MCU and logging rail collapse at the same time, it becomes impossible to store “what voltage we had,” “which charge phase we were in,” and “what the temperatures were.” This chapter defines a logging-grade hold-up (tens to hundreds of ms) and not an OBC-grade bulk capacitor.
Drop / sag scenarios to describe
- Charging gun unplugged → HV bus collapses in tens of ms.
- Contactor open/close → 50–150 ms valley on the bus.
- Charger front-end current limit / hiccup → short, deep sag.
- Line/bus oscillation → aux follows the dip.
Logging requirement
BMS must write: timestamp + bus voltage + charge state/step + temperature + fault code.
MCU + AFE + isolated comm = Iload. Minimum time to finish = thold. Allowable voltage drop = ΔV = Vstart − Vend.
Capacitor sizing formula (BOM ready)
C = Iload × thold ÷ ΔV
Where: Iload = AFE + MCU + isolated comm, thold = time to complete log (50–200 ms typical), ΔV = allowed drop (e.g. 5.0→4.5 V or 3.3→3.0 V).
“Hold-up capacitor value shall be recomputed whenever controller, switching frequency or UVLO is changed.”
Where to place the hold-up
- Secondary-side cap: simplest, directly sustains 5 V / 3.3 V logging rails.
- Primary-side small bulk: prevents controller reset during bus dips.
- Prefer secondary-side for logging; add primary bulk only when controller UVLO is tight.
This hold-up is meant to finish the log, not to keep the whole charger alive. Write the formula in the BOM so that any later substitution of controller / UVLO / switching frequency will trigger a mandatory re-check.
Startup, Bias & Pre-Charge of HV Controllers
HV auxiliary controllers for BMS often power up by charging a VCC capacitor through a startup resistor from the HV bus. When the bus sags during contactor actions (e.g. 400 V → 280 V → 420 V), this startup path may no longer deliver enough energy, so the controller drops below its UVLO-off level and restarts again. This is why UVLO thresholds and startup current must not be changed casually in purchasing.
Typical startup path
- HV bus (200–450 Vdc) → startup resistor → VCC capacitor.
- Controller VCC rises to UVLO-on → controller starts switching.
- Auxiliary winding then powers VCC → startup resistor off.
Why it oscillates
- Bus dips during charging/contactors → startup power not enough.
- VCC falls below UVLO-off → controller stops → no aux winding.
- Controller restarts → BMS main MCU experiences brownout or reset.
UVLO window to preserve
Original design might be 12 V on / 8 V off (wide hysteresis).
Do not replace with chips that do 10 V on / 9.5 V off — this narrows the window and causes chatter.
BOM-ready sentences
“Controller UVLO thresholds shall match original design (startup/shutdown).”
“Do not replace with HV controllers that draw higher startup current.”
“Startup / soft-start time shall be coordinated with Chapter 5 ride-through.”
Keep the startup current, UVLO window, and soft-start time consistent with the original design. Otherwise the hold-up capacitor defined in the previous chapter may no longer cover the logging interval.
EMI / CM Noise Control for Measurement-Domain BMS
The HV auxiliary supply is a switching, isolated converter. Its primary switching edges, transformer leakage spikes, and secondary rectifier noise can all couple into the BMS measurement domain. This is different from a generic offline SMPS: here the AFE is actually measuring high-voltage behavior, so any common-mode (CM) current — including through a Y-cap — may look like a leakage path. That is why EMI components (Y-cap, snubber/RCD, switching frequency) cannot be removed or replaced casually.
Noise paths to document
- Primary switching node → transformer leakage spike.
- Transformer → secondary diode recovery → secondary return.
- Secondary return → AFE reference ground → measurement channel.
- Optional Y-cap → controlled CM path → seen by insulation check.
Y-cap trade-off
Y-cap makes EMI plots nicer but creates a deliberate leakage path.
In BMS insulation / leakage measurement this leakage will be detected.
“Y-cap values shall be coordinated with the BMS insulation-measurement page; do not increase Y-cap for EMI without consulting measurement-domain constraints.”
Do not shortcut EMI elements
- Do not delete / downsize primary snubber or RCD networks.
- Do not replace transformer with one of weaker shielding.
- Do not accept controllers with substantially different switching spectrum without re-test.
Grounding / layout note
Make the secondary of the auxiliary supply use single-point / star grounding so converter pulse currents do not circle through the AFE ground. This is especially important when the AFE is doing insulation/leakage checks on the same HV bus.
Whenever controllers, transformers, snubbers, or Y-caps are cross-brand substituted, re-run EMI around the measurement channels. This BMS page is in the charging domain but is bound by the insulation/leakage tolerance of the measurement domain.
Small-Batch Procurement & Cross-Brand Alternatives
Small-batch automotive orders (≤500 pcs, urgent delivery, dual-SKU acceptable) frequently trigger “looks similar” replacements: a consumer offline flyback in place of an automotive controller, a modified transformer in place of the original construction, or a cheaper PSR/single-output DC/DC in place of the secondary-regulated aux. This chapter lists three real-world substitution scenarios and provides BOM-ready English remarks to block unsafe replacements.
Scenario 1 — Controller out of stock
Purchasing drops in a generic offline flyback controller.
- Isolation grade may not match the BMS DC bus level.
- Startup current may be higher → Chapter 5 hold-up not enough.
- Switching frequency spectrum changes → Chapter 7 EMI must be re-run.
Scenario 2 — Transformer not available
Purchasing wants to change turns/layers/bobbin.
- Creepage/clearance may be reduced → violates Chapter 3 isolation envelope.
- Shielding/y-cap coupling changes → measurement leakage changes.
- Must specify: “transformer construction shall be equivalent.”
Scenario 3 — Cheaper aux DC/DC requested
Single-output or PSR module is proposed.
- No secondary regulation → AFE and logging rails will drift.
- Often industrial- or consumer-grade, not automotive.
- Contradicts Chapter 4 requirement: secondary-regulated flyback only.
BOM remarks to paste
1) Auxiliary bias must remain isolated and secondary-regulated; do not replace with primary-side regulated SMPS.
2) Isolation rating and transformer construction shall match the original automotive design.
3) Re-validate ride-through time and EMI on BMS measurement channels after any cross-brand substitution.
Seven-vendor mapping for HV auxiliary bias (charging domain)
These series are actually shipped, public parts from the seven vendors. Use them as series slots for an isolated, secondary-regulated HV auxiliary supply that feeds AFE / MCU / logging in a BMS. Do not downgrade to primary-side regulated (PSR) modules when the measurement domain requires stable rails. Each time you cross brands, run the validation list in the next chapter.
TI slot
Use when you want fully documented automotive isolated flyback.
- UCC28C43-Q1 — classic current-mode isolated flyback, opto-ready.
- UCC28740-Q1 — HV flyback with primary-side sensing, use only when AFE tolerance is re-validated.
- UCC25800-Q1 — isolated bias controller for automotive BMS / traction aux.
- Keep TL431 + optocoupler on secondary when feeding AFE / insulation monitor.
ST slot
High-voltage integrated switch, good for 5–15 W BMS aux.
- VIPer26K — 1050 V offline HV converter, check PCB creepage.
- VIPER17HD — smaller power, still HV, use TL431 for tighter secondary.
- BOM remark: “Check isolation / creepage / pollution degree for region.”
NXP slot
Best when the rest of the BMS already uses NXP MCU / AFE.
- TEA1755T — flyback/PFC combo, use flyback part to build aux.
- TEA1761 — green-mode flyback controller, re-test EMI bands.
- If NXP device is not available → cross to TI/onsemi with equivalent isolation, then re-test.
Renesas slot
Good documentation, but EMI must be re-validated.
- ISL6753 — current-mode PWM for isolated supplies.
- RAA223012 — 700 V AC/DC flyback, use on DC bus only after isolation check.
- BOM remark: “Run EMI on measurement-domain lines after Renesas substitution.”
onsemi slot
Frequently used in small-batch repairs / short lead-time builds.
- NCP1207A — HV current-mode, confirm DC-bus operation.
- NCP1562B — higher-power, active-clamp capable when BMS aux needs margin.
- BOM remark: “Re-test EMI, ride-through, isolation for every cross-brand replacement.”
Microchip slot (secondary)
Use mainly to create clean 5 V / 3.3 V from the isolated main output.
- MIC5283 — 120 V input LDO, ideal for HV-to-low local bias.
- MCP16331 — for small secondary rails (note VIN limit, use after isolation stage).
- Use Microchip for secondary clean rails; primary converter must still be automotive-grade.
Melexis slot (measurement)
Keep when your BMS accuracy depends on Melexis sensing.
- MLX91220 — isolated Hall current sensor for BMS.
- MLX91221 — higher bandwidth variant, used in diagnostics.
- BOM remark: “Keep Melexis parts; otherwise re-calibrate leakage / insulation measurement.”
After any vendor change among the seven above, re-test: EMI on measurement-domain lines (with / without Y-cap), ride-through/hold-up (see Chapter 5 formula C = I · t / ΔV), isolation/hi-pot (see Chapter 3), secondary-output accuracy (see Chapter 4), startup/UVLO chatter during HV bus dips (see Chapter 6).
Validation & test hooks (what to re-test after substitution)
This validation matrix is the answer to “why can’t we just buy a cheaper aux DC/DC?”. Every time you replace TI with onsemi, or swap an ST VIPer for a Renesas flyback, or modify the transformer construction, the BMS measurement domain can see a different EMI / leakage pattern and the hold-up may no longer match the BOM. Run the five items below and archive the waveforms.
1) HV bus dip test
Purpose: verify aux stays alive when the HV bus droops during charging / contactor events.
- Test profiles: 250 → 200 V, 400 → 300 V (or project-specific HV levels).
- Check if 5 V / 3.3 V drop below AFE / MCU minimums.
- Check if the HV controller restarts (UVLO chatter).
2) Ride-through / hold-up test
Purpose: verify the secondary-side capacitor can write the log before the bus disappears.
- Design formula: C = Iload × thold ÷ ΔV.
- Disconnect the input, measure actual hold-up time.
- If the new controller has higher startup / bias current → increase C and re-test.
3) EMI on measurement lines
Purpose: ensure the insulation / leakage monitoring page sees the correct leakage, not the Y-cap leakage.
- Measure with Y-cap installed and with Y-cap removed.
- Log CM noise that appears on AFE reference ground.
- Share results with the insulation-measurement page to avoid conflicting Y-cap values.
4) Isolation / hi-pot
- Run at transformer level and at system level.
- Applies whenever the bobbin, insulation tape, layer stack or supplier changes.
- Applies when AC/DC-style controllers are used on DC bus.
5) Thermal at rated load
- Use actual BMS enclosure / HV zone conditions.
- Log controller, primary switch, transformer temperatures.
- If the new controller has higher losses, increase clearance from measurement-domain parts.
Trigger list: controller PN change (e.g. UCC28C43-Q1 → NCP1207A), transformer construction change, regulation change (secondary → PSR), Y-cap value change, or small-batch / urgent order using a cross-brand substitute. In all cases, run the 5-item matrix above.
Request a Quote
FAQ for HV auxiliary bias in the charging domain
These 12 questions only cover this page’s scope: isolated HV aux supply, secondary regulation, ride-through for logging, cross-brand and sharing the rail with the insulation / leakage monitor. They do not cover OBC PFC/LLC, pack-FET drive or vehicle-level insulation algorithms.
Why can’t I reuse the OBC auxiliary power for the BMS when charging?
Can I replace a secondary-regulated isolated flyback with a cheaper primary-side regulated one?
How long should the ride-through be to finish BMS logging after a charge interruption?
C = Iload × tlog ÷ ΔV. For most charging interruptions the target is in the range of
tens to a few hundreds of milliseconds. If you change the controller, UVLO or startup current, you must re-measure the real hold-up time.
