← Back to: eFuse / Hot-Swap / OR-ing Protection
Definition & Boundaries
Electronic Circuit Breaker (ECB) = limit criteria (ILIM, Vdrop, Vout_UV, P-limit) + energy integration (I2t) + reset policy (Auto-Retry / Hiccup / Latch-Off) + event semantics (PG/FAULT, trip_reason, t_event).
Core Argument
- ECB = “limit criteria + energy budget + reset policy + event semantics”.
- eFuse ≈ baseline limit/thermal protections; may lack I2t curves or event logs.
- Hot-Swap = external MOSFET gate control + SOA focus (not an ECB).
- Ideal-Diode = OR-ing / reverse-feed blocking (not a tripping-policy device).
Required Fields
- VIN, ILIM, I2t (or P-limit), ttrip, policy mode, PG/FAULT semantics.
Illustrative Diagram
Protection Mechanism Overview: Thresholds, I²t, Foldback, P-limit
This chapter consolidates four protection families into one unified, testable state machine to simplify replacement and validation across brands and operating conditions.
Core Argument
- Thresholds: Current limit (ILIM), output under-voltage (Vout_UV), voltage drop (Vdrop).
- I2t: Integral of squared current over time ensures time-related protection.
- Foldback: Current is a function of output voltage (I = f(Vout)), improving stability for cold-start or “soft short”.
- P-limit: Power limiting (P = Vdrop × I), suitable for wide-voltage buses.
Formulas
- I2t trigger condition: ∫0→t i²(τ) dτ ≥ Iref² × ttrip
- P-limit trigger condition: Vdrop × I ≥ Pset
Unified State Machine
Operating regions: Normal → Limit → Trip → Policy (Auto-Retry / Hiccup / Latch-Off). Side thumbnails show the four limit curves.
Priority & Interaction (Short)
- Hard short: fastest of threshold/I2t/P-limit wins → Trip.
- Cold-start (large capacitance): prefer foldback to avoid nuisance I2t trips.
- Wide buses (48/54 V): prefer P-limit to constrain instantaneous thermal rise; I2t governs accumulated energy.
- Soft short: foldback + I2t dual criteria; use a small de-bounce window (e.g., 50–200 µs) to avoid oscillation.
Selection Window: Feasible Domain (VIN / I_NOM / I²t / Thermal)
Map electrical + thermal limits into a feasible domain so candidates pass A→A / A→B screening with the same yardstick.
Feasible domain = { VIN_range, I_NOM, I_LIM, I²t or P_limit, RθJA, Ambient (Ta), Package }.
Illustrative Diagram
Bus Binning (quick guidance)
Feasibility Checklist
- ☐ VIN bin & I_NOM fixed; package & RθJA match ambient Ta.
- ☐ Choose P_limit or I²t as primary guard per bus class.
- ☐ Compute Tj = Ta + Ploss · RθJA → derive Pset or I²t_budget cap.
- ☐ Validate cold-start / soft-short / hard-short without spurious trips.
Policy Domain: Auto-Retry / Hiccup / Latch-Off
Treat policies as timing-domain control. Measure PG/FAULT flips, open/restore windows, and cooling intervals to quantify stability and thermal cycling.
- Auto-Retry: fixed-period retries for transient faults; fast recovery, higher average heat on heavy loads.
- Hiccup: open → cool → retry; reduces average heat, but too-short cycles cause reset storms.
- Latch-Off: remains off until human/system reset; prevents repeated stress, no self-recovery at field.
Timing Windows & Thermal Implications
Selection Matrix (starter values)
Verification Notes
- Record trip_reason, retry_cnt, PG/FAULT timestamps at low/room/high T.
- Observe temperature saw-tooth; if average T rises, increase t_cool or switch to Latch-Off.
- Confirm N_max behavior and TSD interaction (no unintended lockout).
Events & Telemetry: PG/FAULT, Fault Codes, Timestamps
Define a minimal, brand-agnostic observable set (fields + semantics + timebase) so fleet logs are comparable and cloud-ready.
Visibility Acceptance (quick checklist)
- PG=1 (good), FAULT=1 (fault) after edge normalization; debounce documented.
- Required fields present: trip_reason, retry_cnt, latch_flag, t_event_ms, t_event_epoch.
- Units & types fixed; no renaming downstream; brand-raw kept only for audit.
Testing & Validation: Power-up / Short Injection / Thermal Cycling / Timebase Calibration
A reproducible minimal test set to qualify replacements across temperatures, loads, and wiring impedance.
Minimal Test Set
- Power-up with VIN ramp + large capacitance → record I_inrush, Vout ramp, and whether I2t/P-limit trips.
- Short injection (hard/soft) → record t_trip, trip_reason, retry_cnt.
- Policy timing → measure t_retry, t_cool, N_max; watch PG/FAULT flips.
- Thermal cycling → repeat at low/room/high; check I2t consistency, TSD threshold, and t_junc.
- Timebase calibration → set epoch_offset_ms; inject a known-time event; verify t_event_epoch in cloud.
Thermal & Reliability: Junction Temperature, Energy, Package
Quantify how temperature and energy budgets shape I²t / P-limit behavior so results are repeatable across ambient, package, and airflow.
Representative P/Ns by Bus Class
| Brand | Part Number | Voltage Domain | Package / Thermal Notes |
|---|---|---|---|
| Texas Instruments | TPS2595 | ≤ 18 V (eFuse) | Small QFN; check RθJA for continuous current at high Ta. |
| Texas Instruments | TPS2660 | 4.2–60 V (industrial eFuse) | Wide bus; P-limit selection dominates at 48/54 V. |
| Analog Devices / Maxim | MAX17613A | 4.5–60 V (eFuse) | Similar domain to TPS2660; verify I²t vs. P-limit interplay. |
| STMicroelectronics | STEF12 / STEF05 | 12 V / 5 V (eFuse) | Focus on cold-start foldback; small-outline thermal limits. |
| onsemi | NIS5021 | Automotive/Industrial 12 V | AEC options; check de-rating with ambient and PCB copper area. |
| Analog Devices (Linear) | LTC4282 / LTC4215 / LTC4227 | 48/54 V Hot-Swap | Controller + external MOSFET; verify board-level RθJA & SOA (Hot-Swap scope). |
| Texas Instruments | LM5069 / LM5066I | 9–80 V Hot-Swap (PMBus on LM5066I) | Use for 48/54 V racks; align policy timing and telemetry. |
| Infineon | PROFET+2 (e.g., BTS50010-1TAD) | Automotive 12 V (High-side) | Electronic protection behavior; verify thermal de-rating curves. |
Do & Don’t (Repeatability)
- Compute Pset ≤ (Tj,max − Ta)/RθJA − Pbase before selecting a P-limit.
- Split I²t into single-event and repeated-event budgets with a 50–200 µs debounce window.
- Document package, copper area, airflow; re-run thermal checks at low/room/high.
Cross-Brand Substitution (A→A / A→B)
Upgrade “boots once” swaps into a gated, auditable process: curve alignment, telemetry semantics, and policy timing under the same thermal conditions.
Representative Cross-Pairs (with P/Ns)
| Pair | Voltage Domain | What to Align |
|---|---|---|
| TI TPS2660 ↔ ADI/Maxim MAX17613A | 4.5–60 V eFuse | P-limit, I²t, PG/FAULT polarity, t_trip. |
| TI TPS2595 ↔ ST STEF05/12 | ≤ 18 V eFuse | I_LIM, short response, cold-start foldback. |
| TI LM5066I ↔ ADI LTC4282 | 48/54 V Hot-Swap w/ telemetry | I²t, policy timing, telemetry field mapping. |
| TI LM5069 ↔ ADI LTC4215 / LTC4227 | 48/54 V Hot-Swap | I_LIM, t_trip, I²t (same T & RθJA), MOSFET SOA (Hot-Swap scope). |
| Infineon PROFET+2 ↔ onsemi NIS5021-Q | Automotive 12 V High-side | Limit/thermal de-rating, diagnostic semantics. |
Acceptance Deliverables
- Gate table (A→A / A→B) and signed curve-alignment pack: I-V, I²t, P-limit, policy timing @ same T & RθJA.
- PG/FAULT semantic map + event fields aligned with the telemetry schema.
- Test evidence produced using the validation path (power-up, shorts, thermal cycling, timebase).
Layout & Probing Hooks (ECB essentials)
Minimal but critical hooks so Electronic Circuit Breaker (ECB) measurements are repeatable and PG/FAULT stays stable.
Probe Map (copy for silkscreen / test plan)
| Test Point | Signal | Notes |
|---|---|---|
| TP_VIN | Input voltage | Near connector; avoid shared return spikes. |
| TP_ISHUNT− / TP_ISHUNT+ | Kelvin shunt sense | 4-terminal shunt; symmetric routing to ADC pins. |
| TP_VOUT | Output voltage | After ECB; pair with load return. |
| TP_PG / TP_FAULT | Status lines | Open-drain + pull-up to VIO; RC 1–5 ms at system side. |
| TP_THERM | Temperature | Thermocouple/NTC near hotspot; log with events. |
Repeatability Checklist
- Kelvin sense from shunt to GND_REF; do not share heavy current loop.
- PG=1 good, FAULT=1 fault; HW debounce 50–200 µs; system RC 1–5 ms.
- Use coax spring ground for oscilloscope; document cable/impedance and ambient.
BOM Remarks & Sourcing Hooks
Procurement clauses that preserve telemetry, semantics, and policy equivalence for the ECB.
Tick-off Template (copy to RFQ / review form)
| Item | Supplier Evidence | Pass |
|---|---|---|
| Telemetry present (trip_reason, t_event_ms/epoch, retry_cnt, latch_flag) | CSV/JSON sample attached | [ ] |
| Policy equivalence (AR/Hiccup/Latch parameters) | timing plots + parameter sheet | [ ] |
| Curve alignment at same T & RθJA | I-V / I²t / P-limit plots | [ ] |
| PG/FAULT semantics and debounce documented | polarity map + RC values | [ ] |
| Part IDs, lot/wafer, package & thermal data | datasheet + lot info | [ ] |
Copy-paste Clauses
- ECB must support I²t or P-limit and report
trip_reason+t_event; parts without telemetry/logging are not acceptable. - Policy (Auto-Retry / Hiccup / Latch) must match the original design. If changed, firmware and cloud mapping must be updated first.
- PG/FAULT semantics must follow this page; cross-brand substitution requires curve-alignment records at the same temperature and RθJA.
Request a Quote
Troubleshooting Matrix & Fixes
Use a tight loop—Symptom → Trigger/Condition → Observable fields → Root cause → Fix—so logs are reproducible and comparable across brands. Fields follow this page’s schema:
i_lim_set, p_lim_set, i2t_budget, trip_reason, retry_cnt, latch_flag, t_event_ms, t_event_epoch, v_drop, t_junc, and PG/FAULT.
How to use this list
- Record the exact condition (VIN profile, load model, temperature) alongside the event.
- Always capture time + cause:
t_event_ms/t_event_epochandtrip_reason, plus at least one electrical measurement (v_dropor current). - Apply the matching Fix, then re-run the relevant step from the validation path.
False trip at cold-start (large Cout)
Trigger/Condition: VIN ramp into big bulk capacitance; inrush surge.
Observables: trip_reason=OC or PWR, PG chatter, t_event_ms, v_drop.
Root cause: Foldback window too tight or I²t too low for inrush.
Fix: Relax low-V foldback slope, lengthen VIN ramp, raise i2t_budget slightly, verify no short with soft-short test.
Hiccup storm (frequent auto-retries)
Trigger/Condition: Repeating load fault; temperature rising.
Observables: Rapid retry_cnt growth, alternating PG/FAULT, rising t_junc.
Root cause: t_cool too short or N_max too small; energy not dissipated.
Fix: Increase t_cool / t_retry; consider Latch-Off for persistent faults; verify stability at temperature corners.
Timestamp drift
Trigger/Condition: Logs misaligned with host time after power cycle.
Observables: t_event_epoch offset vs. external time, time_sync_src/quality (if present).
Root cause: Missing or stale epoch_offset_ms.
Fix: Re-calibrate offset at boot; log sync source/quality; re-emit a known-time event for verification.
PG/FAULT chatter on fast load steps
Trigger/Condition: Rapid load transients or wiring inductance.
Observables: PG flicker without trip, v_drop spikes.
Root cause: Insufficient debounce; ground return coupling.
Fix: Normalize polarity (PG=1/FAULT=1), add 1–5 ms RC at system side, series 33–100 Ω, star-ground small signal returns.
Thermal runaway at high ambient
Trigger/Condition: Elevated T_a, continuous high load.
Observables: High t_junc, early P-limit triggers, v_drop increase.
Root cause: Underestimated RθJA or power budget.
Fix: Reduce p_lim_set or improve cooling; re-compute with Tj=Ta+Ploss·RθJA; repeat thermal sweep.
Late short detection
Trigger/Condition: Hard short but long t_trip.
Observables: trip_reason=SC, mismatch between set and measured limit.
Root cause: Non-Kelvin shunt routing or insufficient bandwidth.
Fix: Kelvin four-terminal shunt; shorten PCB sense path; verify ADC/sample rate; check i_lim_set.
Overload misclassified as short
Trigger/Condition: Heavy but finite load during spin-up.
Observables: trip_reason=SC with modest v_drop.
Root cause: I²t threshold too low; foldback boundary too aggressive.
Fix: Raise i2t_budget or adjust foldback curve; add “soft-short” test case.
High v_drop but no trip
Trigger/Condition: Sustained drop across the path.
Observables: Large v_drop, stable PG, no trip_reason.
Root cause: P-limit disabled or threshold too high; wrong measurement node.
Fix: Enable/tune p_lim_set; confirm v_drop tap points and units; re-check RDS(on) vs temperature.
Excess retries with steady load
Trigger/Condition: Apparent normal operation yet periodic retries.
Observables: Slow retry_cnt growth; occasional PG flicker.
Root cause: Hidden soft-short or loop interaction at start-up.
Fix: Scope input/output rails; lengthen ramp; retune foldback; confirm load start-up profile.
Missing fields in logs
Trigger/Condition: Cloud records incomplete for certain events.
Observables: Absent trip_reason, retry_cnt, or t_event_*.
Root cause: Edge mapping renamed/filtered fields; unit mismatch.
Fix: Restore canonical names/units; re-ingest sample CSV/JSON; add schema validation in CI.
