Power Integrity Helper ICs / Hot-Swap & eFuse with Telemetry

Hot-Swap & eFuse with Telemetry

Inrush control, current limiting, and fast disconnect with I²C/PMBus telemetry and programmable SOA protections.

Telemetry-enabled hot-swap and eFuse path: source → inrush and ILIM → SOA-aware FET → fast disconnect → I2C/PMBus telemetry — Figure: Telemetry-enabled hot-swap / eFuse — inrush control and current limit feed SOA-aware FET drive, events are reported over I²C/PMBus.

1. Introduction & page loop

Telemetry-enabled hot-swap and eFuse ICs protect an input rail during plug-in, cable connect, or short-circuit events — and at the same time expose what actually happened over a digital bus. This makes the inlet “visible” to the rest of the power system, not just protected.

Typical use: 5–12 V board inputs, 12/24 V industrial and automotive branches, and 36–60 V telecom or storage backplanes where real inrush, line length, and load behavior vary from slot to slot. With telemetry you can read the real current/voltage, correlate faults with other rails, and tighten limits without changing the PCB.

This page is part of the Power Integrity Helper ICs series and only covers hot-swap / eFuse devices that expose status and configuration over I²C or PMBus.

For ride-through or supercap-backed rails, see the “Hold-Up & Backup Rail Managers” subpage. For isolation / reverse protection topics, see “Power Bus Isolators & Protection Helpers”.

2. Why telemetry on a hot-swap / eFuse?

A classic hot-swap protects the rail but stays blind: when it trips, you cannot tell if it was a real short, an inrush that was too fast, or simply a weak supply. Adding telemetry turns that “blind trip” into an observable event — the system can read back voltage, current, temperature and the exact fault code.

1) Inrush varies in the field

Connector, cable, or output capacitor changes can push real inrush above the design value. Telemetry lets you see that peak and re-program the ramp or current limit — no resistor swap needed.

2) SOA margin is tight at 36–60 V

With long harnesses or 48 V racks, MOSFET SOA can be the limiting factor. Reading VDS and IDS (or power) over the bus tells you if the device is de-rating or about to trip.

3) Align with PMBus managers

If the platform already polls PMBus, a telemetry eFuse is the cheapest way to bring the inlet into the same fault log and time base.

4) Remote re-tune

After you see the real startup profile, you can tighten ILIM, slow down the gate, or change retry counts over I²C/PMBus — ideal for fielded units and small-batch builds.

If the goal is to keep the rail alive during an interruption, treat it as a hold-up / backup design, not as a telemetry hot-swap design.

3. Device Types & Topologies

This subpage only covers telemetry-enabled hot-swap controllers and eFuses — devices that can report V/I/T and fault causes over a digital bus and accept configuration (ILIM, ramp, retry, masks). Basic eFuses without I²C/PMBus go to the generic hot-swap/eFuse page.

Telemetry hot-swap and eFuse device types: controller plus external FET, integrated I2C eFuse, automotive diagnostic high-side switch, monitor with gate control — Figure: Four telemetry-capable protection shapes — controller + external FET, integrated I²C/PMBus eFuse, automotive/diagnostic switch, and monitor + gate-control combo.

Controller + external FET

For 12/24/36/48 V, high current, and boards that need to pick their own MOSFETs or parallel devices. Telemetry reports actual inrush and SOA usage; thresholds can be re-programmed.

Typical: telecom shelves, storage backplanes, industrial slots.

Integrated I²C / PMBus eFuse

Single-package high-side switch with current limit, thermal shutdown, and status/fault registers. Best for 5–12 V boards that want protection plus visibility.

Typical: servers, edge gateways, compact controller boards.

Automotive / diagnostic switch

AEC-Q parts that detect harness faults, shorts-to-battery, overtemp, and can report them over I²C or SPI. Still a telemetry-enabled protection device.

Typical: lighting, cameras, domain controllers.

Monitor + gate-control combo

For retrofits where the power train already exists but needs remote current/voltage reading and the ability to adjust limits or retry behavior from software.

Typical: fielded units, mixed-vendor shelves.

Out of scope: basic eFuses without digital interface, ride-through / supercap-backed solutions (see “Hold-Up & Backup Rail Managers”), and pure isolation/backfeed topics (see “Power Bus Isolators & Protection Helpers”).

4. Telemetry Channels & Register Model

What makes these parts different is not just protection, but what they tell the system and what the system can change remotely. A telemetry-capable hot-swap/eFuse should at minimum let you read voltage, current, temperature, and a latched fault code — and let you write current limit, gate profile, and alert masks.

Telemetry register map for hot-swap and eFuse: read VIN/VOUT, current, temperature, status; write ILIM, gate slew, retry, alert masks; align PMBus status words — Figure: Typical telemetry register map — reads for V/I/T and status, writes for current limit, inrush/gate profile, retry and alert masks, plus PMBus alignment.

Telemetry reads (must-have)

Input / bus voltage (VIN / VBUS)
Load / sense current (IOUT / VSENSE)
Output / protected rail voltage
Die / board temperature
Fault & status word (OC, UV, OV, OT, retry)

If a device cannot provide these, treat it as a basic eFuse, not as a telemetry-enabled protection IC.

Writable / configurable

Current limit / ILIM / IFAULT level
Gate slew / dV/dt / inrush profile
Retry count / auto-restart / latch-off
Alert / interrupt masks
(Optional) power / energy accumulator config

Most vendors require you to read back ILIM after writing — do not assume the new value was accepted.

Event & PMBus alignment

Single ALERT / SMBALERT# line for quick fault indication
Latched fault code for post-mortem read
Map to PMBus: STATUS_WORD / READ_VIN / READ_IOUT
Keep polling cadence aligned with power manager
Optional: push fault to central logger

For centralized sequencing and black-box style logging, pair this with the “PMBus/SMBus Power System Managers” subpage.

Document in BOM remarks: VIN range, maximum load current, ILIM register value, digital interface (I²C or PMBus), device address, and qualification class (AEC-Q if needed). This makes cross-brand replacement much easier later.

If your goal is to keep events during power loss, use a power health logger / black-box recorder in addition to the telemetry eFuse.

5. Inrush / Current Limit / SOA Programming

Telemetry turns protection into a closed loop: first you power the board and see the real inrush and trip conditions, then you re-program the gate ramp, current limit, retry policy, or SOA window until it matches the backplane capability. Do not freeze ILIM or dV/dt from the datasheet alone.

Inrush and SOA programming for telemetry-enabled hot-swap and eFuse: measure real inrush, adjust gate ramp, set current limit, and enforce SOA-based disconnect — Figure: Program on real data — measure actual startup inrush, then write gate slew, ILIM, retry, and SOA/power limit over I²C/PMBus.

1) Measure real inrush

Power the board in its real slot/backplane, then read VIN, VOUT, IOUT and the fault/status words. Cable length, source stiffness, and output capacitors often make inrush higher than the lab estimate.

Rule: log-before-tuning.

2) Program ramp / dV/dt

Set gate slew or pick a startup profile (normal, heavy-cap, pre-charge). Slow the ramp for weak supplies or long cables, faster ramp for local supplies. Confirm the new ramp via scope and bus readback.

Keep VOUT rise and current peak in balance.

3) Current limit, foldback, retry

Write ILIM to the device, then read it back to confirm. Enable foldback for long shorts, hot compartments, or loads like lamps/cameras. Choose auto-retry for fielded units; latch-off for lab/ATE.

Document ILIM, retry count, and foldback in BOM remarks.

4) SOA / power limiting

For 36–60 V and long harnesses, use power- or time-based protection: the IC watches VDS × IDS and derates or trips when the MOSFET SOA is exceeded. Log the status right before disconnect for QA.

If you cannot prove SOA, do not relax ILIM.

Final step: capture the oscilloscope waveform together with I²C/PMBus status and store both in the NPI/DFX sheet. That way procurement can safely swap to an equivalent telemetry-enabled hot-swap/eFuse later.

If the requirement is “keep rail alive during supply interruption”, switch to the “Hold-Up & Backup Rail Managers” subpage.

6. System Integration with PMBus / Managers

A telemetry-enabled hot-swap or eFuse becomes a first-class node in the PMBus power tree: it protects locally and fast, but it also reports faults, operating points, and configuration status to the system power manager. Trips must never depend on a PMBus round-trip; PMBus is for configuration, logging, and fleet tuning.

PMBus integration flow for telemetry-enabled hot-swap and eFuse: local fast trip, shared SMBALERT line, manager polls status words and reconfigures ILIM and retry policy — Figure: Local-fast, remote-slow — the eFuse trips instantly, then the PMBus manager collects status words, time-stamps the event, and optionally reconfigures thresholds for the whole system.

1) Local-fast, remote-slow

The device limits current or disconnects on its own, within microseconds to milliseconds. The manager only learns about it over SMBALERT#/I²C and logs or updates policies.

2) Address & shared alert

When several telemetry eFuses share one bus, plan addresses first and then decide whether ALERT is ORed or individual. Avoid collisions with PMBus power managers and rail monitors.

3) Polling & cadence

Align eFuse polling (for READ_VIN / READ_IOUT / STATUS_WORD) with the rest of the PMBus tree, typically 50–200 ms. Do not poll a single eFuse at 1 ms if other rails are slower.

4) Configuration flows

On cold start, the manager pushes ILIM, retry, and alert masks to all eFuses. During service or field tuning, the manager can relax or tighten limits on a single slot and log the change.

Map device data to standard PMBus commands whenever possible (STATUS_WORD, STATUS_IOUT, READ_VIN, READ_IOUT). Use MFR-specific registers only for advanced SOA or diagnostics to keep tools simple.

For centralized fault trees and black-box style history across power losses, pair this page with “PMBus/SMBus Power System Managers” and “Power Health Loggers & Black-Box Recorders”.

7. Selection Guide by Voltage / Current / Interface

Select in this order: 1) input voltage bucket → 2) expected load current → 3) telemetry interface (I²C, PMBus, SPI). This ensures the device can both protect the rail and report events to the power manager.

Selection matrix for telemetry hot-swap and eFuse by voltage, current, and interface: 3.3–5 V board, 5–12 V server, 12–24 V industrial, 36–60 V telecom — Figure: Pick device by voltage first, then current, then the bus (I²C / PMBus / SPI). Higher voltages and higher currents typically require controller + external FET.

3.3–5 V board-level

Small boards, MCU/SOM, peripherals. Prefer integrated telemetry eFuse / smart high-side with I²C.

NXP NX30P6093 (5 V, I²C, 8 A)
ST STi²Fuse (low-voltage variants)
Microchip PAC1934 + local switch

If no telemetry is required, move to the generic eFuse page.

5–12 V server / networking

Backplane / fan / control-card inputs. Need current limit, fault codes, and PMBus or I²C.

TI LM25066 / LM25066A (2.9–17 V, PMBus)
ST STi²Fuse (telemetry smart fuse)
onsemi eFuse + I²C monitor combo

For >15 A, switch to controller + external FET.

12–24 V industrial / automotive

Domain controllers, lighting, cameras, PLC I/O. Diagnostics and line/harness fault detection required.

ST VNF1048F, smart HS with diagnostics
NXP MC33XS2410 (4-ch, SPI)
Renesas hot-swap + ISL2802x monitor

Interface can be I²C or SPI; SPI is common in automotive HS.

36–60 V / 48 V telecom

Shelves, storage, baseband. Need PMBus, SOA-aware gate control, event logging.

TI LM5066 / LM5066I / LM5066H (5.5–90 V, PMBus)
Renesas hot-swap (48 V) + PMBus monitor
onsemi protected HS for 48 V (check SOA)

Always program and verify SOA at this voltage.

Current buckets

≤ 5 A

Integrated smart eFuse / HS; I²C only; small boards.

5 – 15 A

Most server/industrial cards; PMBus preferred.

15 – 40 A

Controller + external FET; log-before-tuning; SOA check.

> 40 A / paralleled

High-power shelves; mandatory PMBus + external FET + SOA.

Interface rule of thumb: I²C/SMBus for small loads and automotive HS; PMBus for multi-rail server/telecom systems; SPI for multi-channel automotive/industrial switches. If the device only signals a fault pin, classify it under the basic protection page, not here.

8. Brand Mapping (TI / ST / NXP / Renesas / onsemi / Microchip / Melexis)

Use this map to stay inside the seven preferred vendors when building a telemetry-enabled hot-swap / eFuse BOM. Each card lists real part numbers; if a part does not expose I²C/PMBus itself, pair it with the noted monitor to keep the page intent.

Figure: Seven-brand mapping for telemetry-enabled hot-swap / eFuse. Stay within these families to simplify cross-brand replacement and sourcing.

Texas Instruments (TI)

PMBus-based hot-swap and high-voltage controllers; best choice for 48 V, SOA-aware systems.

LM25066 / LM25066A / LM25066I (2.9–17 V, PMBus)
LM5066 / LM5066I / LM5066H (5.5–90 V, PMBus)
LM5064 (48 V controller, telemetry-capable line)
Use with: TI power system managers for centralized logs

Back to: Power Integrity Helper ICs

STMicroelectronics (ST)

Smart eFuse and automotive high-side with diagnostics; good for 12–24 V domains.

STi²Fuse (smart eFuse, I²C)
VNF1048F (automotive HS, diagnostics)
STEF05 / STEF12 (non-bus eFuse → use generic page)
Pair with: ST power monitors for full telemetry

Back to: Power Integrity Helper ICs

NXP

Multi-channel smart HS with SPI diagnostics; ideal for automotive and industrial slots.

MC33XS2410 (4-ch, SPI, diagnostics)
MC33XS2100 / similar XS-series HS
NX30P6093 (5 V, I²C, 8 A load switch)
Note: many NXP HS parts are SPI, not PMBus

Back to: Power Integrity Helper ICs

Renesas

Hot-swap + digital power monitor two-chip approach; good for 24–48 V and PMBus-centric systems.

Renesas / Intersil hot-swap controllers (48 V)
ISL28022 / ISL28023 / ISL28030 (I²C/PMBus monitors)
ISL68200 / other PMBus power ICs for managers
Combine controller + monitor for full telemetry

Back to: Power Integrity Helper ICs

onsemi

Protected/high-side switches and eFuse families for 5–24 V; add external current/voltage monitor to bring it into this telemetry page.

NIS5021 / NIS5135 (protected eFuse)
Automotive smart distribution (NCV series)
Pair with: onsemi current sense / PMBus monitor
Non-bus parts → generic eFuse / hot-swap page

Back to: Power Integrity Helper ICs

Microchip

Use Microchip for the telemetry/measurement side, then gate or protect with an external HS/eFuse.

PAC1934 (4-ch power monitor, I²C)
MCP39F521 (energy/power monitor)
MIC2545A / similar HS switch for the power path
Combine for “telemetry-enabled eFuse” behavior

Back to: Power Integrity Helper ICs

Melexis

Use Melexis when the protected load is a sensor/actuator that must report health or run in an automotive compartment. Treat it as a “diagnostic supply protection” case.

MLX81113 / MLX81150 (LIN-based actuator/supply)
MLX92292 (sensor with diagnostics)
Use external HS/eFuse to add hard protection
Log sensor rail events through the system manager

Back to: Power Integrity Helper ICs

Always capture: VIN bucket, max load current, chosen interface (I²C/PMBus/SPI), programmed ILIM, and preferred brand in the BOM remark. This is what lets purchasing switch between TI ↔ ST ↔ NXP ↔ Renesas ↔ onsemi ↔ Microchip ↔ Melexis without breaking the telemetry-based protection flow.

9. Validation & Production Test Hooks

Validation for a telemetry-enabled hot-swap / eFuse is a two-channel task: capture the analog waveform (VIN, VOUT, IGATE, ILOAD) and at the same time capture the digital view (I²C/PMBus reads, status words, fault codes). Bring-up must always do “log-before-tuning” to avoid locking ILIM from a lab-only condition.

Validation hooks for telemetry hot-swap: scope VIN/VOUT/IGATE, I2C or PMBus readback, ALERT test pad, and fault-inject point for production — Figure: Scope the power path and read the bus at the same time. Production must have I²C/PMBus header, ALERT pin and a fault-inject jumper.

Layer 1 — Bench / bring-up

Power the board on its real backplane, scope VIN/VOUT/IGATE, and in the same run read READ_VIN / READ_IOUT / STATUS over I²C/PMBus. Save waveform + register capture as the reference profile.

Layer 2 — System / PMBus

Put several telemetry eFuses on one bus, enable shared SMBALERT#, and let the PMBus manager poll at 50–200 ms. Check that no alert is lost and that trip does not wait for the bus.

Layer 3 — Thermal / corner

Repeat hot-plug, short, and inrush in a hot box, with upstream DC/DC ripple and long harness. Read device temperature and confirm fault counters do not spam from switching spikes.

Production test hooks

1) PWR_IN / PWR_OUT pads

Provide pre-FET and post-FET pads so ICT/ATE can power and measure the protected path.

2) I²C / PMBus header

SDA, SCL, GND, ALERT, optional ADDR. Used to read device ID, STATUS_WORD and to write ILIM once.

3) SMBALERT# / FAULT pin

Production pulls this line low and checks the controller/tester receives the interrupt.

4) Fault-inject / load-step point

Small jumper/resistor to force an overload/short. Immediately read STATUS to confirm proper fault code.

Record in BOM / NPI sheet:

Telemetry hot-swap / eFuse PN
VIN bucket (3.3–5 / 5–12 / 12–24 / 36–60 V)
Programmed ILIM / retry policy
I²C / PMBus address and bus type
Test hooks: I²C header + ALERT + fault-inject pad

If you need event retention across power loss, add the “Power Health Loggers & Black-Box Recorders” subpage components.

Frequently Asked Questions

FAQ for telemetry-enabled hot-swap: why add I2C, how to test SMBALERT, how to change ILIM in the field, how often to poll — Figure: FAQ focuses only on telemetry-enabled hot-swap / eFuse. Basic eFuse, backup rails and black-box logging are covered on their own subpages.

Why add I²C/PMBus if the hot-swap already trips?

Telemetry tells you why it tripped (inrush, OC, SOA, hot). It also lets you tighten ILIM, change retry count, and align the device with the system power manager.

Can it tell me the real inrush on the backplane?

Yes. Power once in the real slot, read VIN/IOUT/status, then re-program the gate ramp or ILIM. This “log-before-tuning” step is mandatory for long cables and soft sources.

What is the minimum telemetry set I should expect?

VIN / VOUT, load current, temperature and a latched fault/status word. Without these, treat the device as a basic eFuse, not as a telemetry-enabled hot-swap.

How do I test SMBALERT# / FAULT during production?

Use the ALERT test pad to pull the line low, then confirm the tester or host reads the status register. Optionally inject a short via the fault pad and read the fault code.

Does a telemetry eFuse need PMBus to trip fast?

No. Trip, current limit and SOA actions are always local and fast. PMBus/I²C is only for configuration, polling, and logging after the event.

How do I avoid address collisions on one bus?

Reserve an address range for hot-swap/eFuse, program each unit once during production, and read back the new address. Keep ALERT lines either ORed or separate per slot.

Can I change ILIM in the field without rework?

Yes. Write ILIM over I²C/PMBus, then read it back to confirm. Store the new value in your service log and update BOM remarks to support later replacements.

What if the device trips before it can report?

Use devices with latched status or push the fault to the PMBus manager over ALERT. The manager time-stamps it and keeps the record even if the slot power drops.

How often should I poll a telemetry hot-swap?

Match the rest of the PMBus tree, typically 50–200 ms. Polling a single eFuse at 1 ms on a shared bus can disturb other rails and managers.

Can I log SOA-related events over the bus?

Many PMBus-based hot-swap controllers expose power/energy or power-limit status. If not, log the overcurrent/foldback code in the system power manager instead.

How do I verify my inrush profile really changed?

Re-power the slot, scope VOUT and current, and read back the ramp/gate-control registers. Keep before/after plots with their I²C/PMBus logs in your validation folder.

Is an automotive diagnostic HS the same thing?

It is functionally close: protects and reports. But it often speaks SPI and targets harness faults. Keep it on this page only if it reports health/overload.

What should purchasing see in the spec?

Part number, VIN bucket, programmed ILIM, interface and address, and required test hooks. Without these, cross-brand replacement is risky.