← Back to Low Dropout Regulators (LDOs)
1) Introduction: Supervision vs Protection (for LDOs)
Protection = stop damage; Supervision = see it early + tell MCU.
In LDO-based rails we often enable protection first (current limit, short-circuit, OTP, active discharge, reverse guards). That saves the device. But in automotive, camera, sensor, or lighting modules you also need to observe and report the rail state. That is supervision.
This page focuses only on UV / window comparators, temperature sensors, fault IRQs, and optional I²C/PMBus readout. It does not describe hard power cut-off, eFuse, or hot-swap functions. For that, use the sibling page “Protection Set for LDOs”.
In other words: the previous page answers “can the LDO survive?” while this page answers “can the system know a rail is drifting or overheating before damage?”.
2) UV / Window Comparators for LDO Rails
Most LDOs expose only a Power-Good (PG) pin with a relatively wide threshold. That is fine for simple “rail is up” checks, but supervision often needs narrower and sometimes two-sided detection. That is why we add UV-only comparators or window comparators.
UV-only (single-sided) supervision
UV-only supervision asserts a fault when VOUT < VUV_th. It is the best choice for single-bias LDO rails, sensor-bias lines, or automotive lighting nodes where “too low” is the main concern.
Window (two-sided) supervision
A window comparator asserts when VOUT is too low or too high: VOUT < VUV_th or VOUT > VOV_th. This is preferred for camera rails, ADC references, precision sensors, and multi-rail LDO groupings because it can also catch “wrongly calibrated / drifting high” rails.
Engineer’s BOM fields (must record)
- V_UV_th – undervoltage threshold for the supervised LDO rail.
- V_UV_hys – hysteresis to prevent chattering with DC/DC ripple.
- V_OV_th (if window) – upper/window threshold to detect overvoltage or calibration errors.
- Output type – open-drain (wired-OR to MCU) or push-pull.
- Response time (ms/us) – must match the startup profile of the upstream DC/DC so it does not trigger false faults.
Some LDOs expose only PG and do not integrate a configurable window. This has to be confirmed during cross-brand sourcing, especially when you switch between TI, ST, NXP, Renesas, onsemi, Microchip, and Melexis.
3) Temperature Sensors on/near LDO
Not every LDO temperature sense can be read out. Many devices only have an internal self-protect sensor that feeds the OTP or thermal foldback circuit and stays invisible to the MCU. For supervision and telemetry we need a readable / reportable temperature source.
This section therefore separates self-protect sensors (internal-only) from supervision sensors (exporting a flag, IRQ, or register value).
Self-protect sensor (internal-only)
Built into almost every LDO. It trips at a fixed OTP point — for example 150°C off / 130°C resume. It is not meant for system diagnostics, and many datasheets do not expose this temperature to the outside world. You only see the effect: the LDO shuts down or current is reduced.
Supervision sensor (telemetry-capable)
Designed to tell the MCU “I am getting hot” before a real shutdown. It usually offers:
- T_warn – warning temperature where a flag or IRQ is raised.
- T_shutdown – for reference, the real OTP point (may still be internal).
- Output method – dedicated flag pin, fault IRQ, or readable I²C/PMBus register.
- Clear / hysteresis – temperature delta to clear the warning.
For automotive and lamp-chamber rails the requirement is often: “tell me it is hot so I can derate,” not “shut down right away.” That is why supervision-grade temperature sensors matter.
4) Fault IRQs vs. PG Pins
PG is “when the rail is OK, I tell you OK.” Fault IRQ is “when something is wrong, I call you right now.”
PG is perfect for simple LDOs and single-rail products; it masks minor disturbances. Fault IRQ is needed when you have to react to undervoltage, overvoltage/window, overtemperature warning, or rail-miss events — typical in automotive, industrial, and camera LDOs.
| Item | PG (Power Good) | Fault IRQ |
|---|---|---|
| Trigger | rail reaches PG threshold | UV / OV / OT / window / rail-miss |
| Direction | positive (OK → signal) | negative (fault → signal) |
| Speed | usually slower / filtered | fast, non-masked if unconfigured |
| Best for | simple LDOs, “is it up?” checks | automotive, industrial, camera, multi-rail |
| Risk / Note | may hide small faults | MCU may be interrupted continuously |
| Sourcing question | PG level, pinout | can IRQ be masked / shared / open-drain? |
Sourcing reminder: always check whether the fault IRQ can be masked or reused. Otherwise, during slow DC/DC ramps or noisy rails the MCU will keep receiving interrupts.
5) Optional I²C / PMBus Telemetry
“Optional I²C/PMBus” means the LDO or its companion supervisor can export supervision data over a serial bus. This is usually available on higher-tier automotive / industrial LDOs or PMICs, not on basic regulators.
Typical data to read:
- Vout status — rail OK / undervoltage / overvoltage (window).
- Temperature — overtemp warning, sometimes raw temperature.
- Fault log / status bits — which rail, what fault, is it latched.
I²C is a lightweight choice when you only need to read a few flags and your MCU already has I²C. PMBus is for power-oriented boards with multiple rails and future event logging.
Cost-down hint: if you only need voltage/window + temperature flags, you don’t have to pay for PMBus parts — choose an LDO with hardware flags or simple I²C instead.
6) Validation & Logging Hooks
Supervision is only useful if you can prove its thresholds and log the events. This section outlines how to validate UV/window, temperature warning, and fault IRQ stability — then write the measured values into the BOM so you can cross-check TI ↔ ST ↔ NXP ↔ Renesas ↔ onsemi ↔ Microchip ↔ Melexis later.
UV / window validation
Use a programmable supply. Start from nominal and sweep downward to capture the UV trip. If a window comparator is present, sweep upward to capture the OV / high window trip. Record:
- V_UV_trip_meas
- V_UV_hys_meas
- V_OV_trip_meas (if window)
- flag/IRQ latency
Temperature validation
Use hot air or a chamber. Raise the temperature until the T_warn flag or IRQ asserts. If you have I²C/PMBus, read back the status register to confirm the same event. Record:
- T_warn_meas
- T_clear_meas / hysteresis
- output method: flag / IRQ / register
IRQ stability validation
Create a small drop-out or an OT event and observe whether the IRQ:
- triggers once (pulse/level)
- keeps firing during slow ramps
- can be masked/cleared via I²C/PMBus
Then write: irq_mode=level/pulse, irq_maskable=yes/no, irq_debounce=… into the BOM remarks.
7) Seven-Vendor Supervision / Telemetry Pointers
All seven focus vendors on this site can supply parts that watch an LDO rail (UV / window / temp / fault) and report it to the MCU. This section is only a pointer — it shows “this brand has this line” and gives typical part numbers. It does not repeat protection, start-up or PMIC pages.
TI
LDOs with PG + PMBus power monitors / sequencers → build supervisable rails.
- UCD90120A — 12-rail PMBus power supply sequencer
- UCD9090A — multi-rail monitor/sequencer
- TPS386000 / TPS3808 — voltage supervisors for LDO outputs
Use when you need PMBus logging or many rails behind simple TI LDOs.
STMicroelectronics
ST LDO + supervisor/reset IC = narrow UV + fault indication.
- STM706 / STM705 / STM708 — supply supervisors
- STM6719 — dual/3-rail supervisor
- STM1810 series — low-power reset
Best when you only need hardware flags (no PMBus).
NXP
Automotive PMIC/SBC often expose UV/window/fault for camera, ADAS, gateway.
- PF5023 (PF50xx family) — multi-rail automotive PMIC
- VR5510 — safety/automotive PMIC with monitoring
- FS56xx / FS84/FS85 families — voltage monitors for car MCUs
Good match for “UV/window separated” LDO supervision.
Renesas
Camera/vision PMIC + classic voltage supervisors → watch small LDO rails.
- ISL78083 — 4-channel automotive camera PMIC with monitoring
- ISL88003 / ISL88031 — voltage supervisors
- RAA/RL78 ecosystem PMICs with PG/reset
Use for surround-view / camera / DMS modules.
onsemi
Automotive / lighting LDOs with PG, easy to pair with external supervisor.
- NCV8164 / NCV8165 — AEC-Q100 LDO, PG
- NCP170 / NCP718 — low-Iq LDOs for sensor rails
- NCP730 — LDO with power-good indicator
Fits car lighting, camera bias, RF auxiliary rails.
Microchip
I²C/SMBus power/energy monitors — log what the LDO rail is doing.
- PAC1934 — 4-ch power monitor, I²C
- PAC1943 / PAC1944 — SMBus power monitors
- MCP1316 / MCP1319 — reset/supervisor for simple LDO rails
Best for industrial + automotive mixed designs.
Melexis
Sensor-oriented, SMBus/I²C-readable, good when you need “tell me it’s hot”.
- MLX90614 — SMBus IR temperature sensor
- MLX90632 — compact thermal sensor
Use when the LDO is biasing a sensor and you must report thermal status to ECU.
All pointers above stay inside the seven brands, so small-batch buyers can keep procurement simple and still get UV/window, temperature and fault reporting for LDO rails.
8) Get Supervision-Ready LDO Parts
Small-batch automotive / camera / industrial LDO projects often need visibility (UV, window, temp, IRQ, I²C/PMBus), not just regulation. Tell us what your rail must report and we will map it to TI, ST, NXP, Renesas, onsemi, Microchip, Melexis.
If you only need voltage/window + temperature flags, you don’t have to pay for PMBus parts — we can down-spec to simpler LDO + supervisor options.
Paste your BOM here or attach it in your WordPress form. We will return seven-vendor options (TI / ST / NXP / Renesas / onsemi / Microchip / Melexis) within ~48h.
Supervision options you can tell us: UV only, UV/OV window, temperature warning, fault IRQ, I²C, PMBus, maskable IRQ.
Frequently Asked Questions
What is the difference between PG and a fault IRQ on an LDO?
PG reports “rail is in a good window” and often has wide thresholds. A fault IRQ reports “something went wrong” such as UV, OV, or OT and is usually faster. Use PG for basic bring-up, IRQ for supervision.
Do I need a window comparator if I already have PG?
PG normally only flags “OK”. A window comparator can flag both “too low” and “too high”, which is important for sensor, ADC reference, or camera rails. If you must detect overvoltage too, add the window.
How to test undervoltage supervision on an LDO rail?
Use a programmable supply, start from nominal and slowly decrease the voltage until the UV flag or IRQ asserts. Log the exact trip point, hysteresis, and latency and write them into the BOM remarks.
Can I read LDO temperature over I²C/PMBus?
Only if the LDO or its companion supervisor exposes a readable temperature register or OT flag. Many LDOs have an internal OTP sensor that is not readable. Check the datasheet for “telemetry” or PMBus pages.
Are supervision functions available on all automotive LDOs?
No. Many AEC-Q100 LDOs only give PG or enable. For real UV/window/temp reporting you may need a higher-grade LDO, a small PMIC, or an external supervisor from the same brand.
How to avoid false IRQs when the upstream DC/DC starts slowly?
Mask or delay the IRQ during start-up, or use a supervisor that supports blanking. Slow ramps often cross the UV threshold multiple times. A maskable open-drain IRQ is the safest option.
Can a simple reset/supervisor IC replace LDO built-in PG for tighter UV thresholds?
Yes. External supervisors from TI, ST, Renesas, and Microchip often have tighter UV thresholds and better hysteresis than a built-in PG. Place them after the LDO output and log the measured trip point.
When should I choose PMBus instead of plain I²C for LDO supervision?
Choose PMBus when you have several rails, need standardized status words, or plan to log power events. For 1–3 rails and simple flags, plain I²C is cheaper and easier to integrate.
What should I write into the BOM after supervision tests?
Record measured UV/OV trip, hysteresis, T_warn, IRQ mode (pulse/level), bus type (I²C/PMBus), and whether IRQ is maskable. This makes cross-brand replacement inside the seven vendors much easier.
Can multiple LDO fault IRQs be wired-OR to one MCU pin?
Yes, if the IRQs are open-drain/open-collector and you add a pull-up. But you still need to read individual status over I²C/PMBus or via GPIO to know which rail reported the fault.
Why does my LDO PG not trigger on small droops or noise?
PG circuits usually include filtering and wide thresholds, so short droops or small noise are ignored. For tighter supervision you need a window comparator or an external supervisor IC.
Do I need temperature warning if the LDO already has OTP?
OTP only protects the LDO. A readable temperature warning lets the MCU derate lamps, cameras, or radios before shutdown. In automotive/lamp-chamber designs, keep both.
How do I supervise a camera / lamp-chamber rail that heats up slowly?
Use a supervision-grade temperature sensor or a PMIC that exports T_warn, and sample it over I²C/PMBus. Slow heating needs logging more than fast shutdown, so store events in your firmware.
Which of the seven vendors support multi-rail monitoring around LDOs?
TI, NXP, and Renesas have the richest multi-rail / PMBus options. ST, onsemi, and Microchip cover simpler PG/supervisor cases. Melexis is great when you need temperature data with car sensors.
Can I use these supervision signals for field/service diagnostics later?
Yes. If you log UV/OV/OT/IRQ events and keep them in nonvolatile memory or upstream over PMBus, service teams can see which rail misbehaved and which LDO to replace.
