At the most basic level, what does a heat sensor do? It converts thermal energy into an electrical signal—usually a varying resistance (like an NTC thermistor) or a proportional voltage (like a TMP36 or internal silicon bandgap diode). But in embedded systems and power electronics, simply reading a temperature value on an I2C bus or ADC pin is useless unless you understand the thermal physics happening behind that number. A heat sensor doesn't fix thermal problems; it merely reports them. To actually solve overheating, you must master the thermal path from the silicon junction to the ambient air.
This guide moves past basic sensor polling and breaks down the thermal resistance math ($R_{\theta}$), derating curves, and practical heatsink selection required to keep your microcontrollers, LDOs, and power MOSFETs alive on the bench and in the field.
The Thermal Path: Junction-to-Ambient Math
When a component dissipates power as heat, that heat must travel through several physical boundaries before it reaches the surrounding air. We model this using thermal resistance, measured in °C/W (degrees Celsius per Watt). The core equation governing this path is:
TJ = TA + PD × (RθJC + RθCS + RθSA)
- TJ: Junction temperature (the actual silicon die temperature).
- TA: Ambient temperature (the air inside your enclosure, not just the room).
- PD: Power dissipated in Watts.
- RθJC: Thermal resistance, Junction-to-Case.
- RθCS: Thermal resistance, Case-to-Sink (includes thermal paste or pad).
- RθSA: Thermal resistance, Sink-to-Ambient (the heatsink and airflow).
Worked Example: SOT-223 LDO Overheating
Suppose you are using an AMS1117-3.3 LDO in a SOT-223 package to drop a 12V battery rail down to 3.3V for an ESP32 drawing 500mA.
First, calculate power dissipation: PD = (12V - 3.3V) × 0.5A = 4.35W.
If you rely solely on the datasheet's Junction-to-Ambient ($R_{\theta JA}$) rating for a SOT-223 with minimal PCB copper pour (typically ~50°C/W), the math is brutal:
TJ = 25°C + (4.35W × 50°C/W) = 242.5°C.
Derating Curves and Failure Signatures
How hot is too hot for this part? Every semiconductor datasheet includes an "Absolute Maximum Ratings" table, usually listing a maximum junction temperature ($T_{J(max)}$) of 125°C or 150°C. However, designing to the absolute maximum is a rookie mistake. According to the Arrhenius equation, the failure rate of electronic components roughly doubles for every 10°C increase in operating temperature. For high-reliability embedded designs, the practical hard stop is keeping TJ below 100°C, and ideally below 85°C.
As temperature climbs, components exhibit specific failure signatures before they catastrophically die:
| Failure Signature | Affected Components | Physical Mechanism |
|---|---|---|
| Timing Skew | MCUs, FPGAs, SRAM | Carrier mobility drops at high temps, slowing transistor switching speeds and causing setup/hold time violations. |
| Electromigration | Power ICs, MOSFETs | High current density combined with heat physically moves metal atoms, thinning traces until they vaporize (open circuit). |
| Parametric Drift | Electrolytic Capacitors | Electrolyte boils off through the rubber seal, increasing Equivalent Series Resistance (ESR) and destroying power rail stability. |
| Thermal Runaway | BJTs, Power Diodes | Forward voltage drop decreases as temp rises, drawing more current, generating more heat in an uncontrollable positive feedback loop. |
For a deeper look at how package geometry affects these limits, Texas Instruments provides excellent application notes on Semiconductor and IC Package Thermal Metrics, detailing how PCB copper thickness drastically alters $R_{\theta JA}$.
Heatsink Selection and Airflow Tactics
Let's fix our 4.35W SOT-223 LDO problem. We need to select a heatsink based on wattage, not guesswork.
Target: Keep TJ ≤ 85°C in an enclosure where ambient air (TA) reaches 45°C.
Knowns: PD = 4.35W. RθJC = 3°C/W (from datasheet). RθCS = 1.5°C/W (using a standard 0.5mm silicone thermal pad).
Rearranging the thermal equation to solve for the required Sink-to-Ambient resistance:
RθSA = [(TJ - TA) / PD] - RθJC - RθCS
RθSA = [(85 - 45) / 4.35] - 3 - 1.5
RθSA = 9.19 - 4.5 = 4.69°C/W
You need a heatsink with a thermal resistance of 4.69°C/W or lower. A standard stamped-metal clip-on sink won't cut it; you need an extruded aluminum profile. The Wakefield-Vette 680-150AB (a TO-220/SOT-223 compatible extruded sink) offers an $R_{\theta SA}$ of roughly 4.5°C/W in natural convection.
What Airflow and Enclosure Changes Buy You
If your physical constraints prevent using a massive 4.5°C/W heatsink, you must alter the enclosure dynamics. Natural convection relies on the chimney effect—hot air rising. If your enclosure is sealed, the "ambient" temperature inside quickly rises to match the junction temperature, rendering the heatsink useless.
- Venting: Adding lower intake and upper exhaust louvers allows natural convection to flush the enclosure. This can drop internal TA by 10°C to 15°C.
- Forced Air: Adding a small 40mm brushless fan (e.g., 10-15 CFM) across a smaller heatsink (like the Aavid 533202B02500G, rated ~12°C/W passive) will drop its effective $R_{\theta SA}$ to roughly 4.0°C/W, easily meeting our 4.69°C/W target.
- Thermal Potting: In sealed, outdoor IoT nodes where airflow is impossible, potting the entire PCB in thermally conductive epoxy (like MG Chemicals 832TC) turns the entire enclosure into a giant heatsink.
For more on calculating convection coefficients in enclosed spaces, All About Circuits offers a solid primer on bridging the gap between datasheet ideals and real-world enclosures.
Frequently Asked Questions: Heat Sensor Functions
What does a heat sensor do when an ESP32 or MCU hits its thermal limit?
Modern microcontrollers like the ESP32-S3 or STM32 series feature internal silicon bandgap temperature sensors tied directly to the system management unit. When the internal die temperature crosses a critical threshold (often around 125°C), the sensor triggers a hardware-level PROCMON (Processor Monitor) interrupt. This forces the MCU to clock down, disable high-draw peripherals like WiFi/Bluetooth radios, or initiate a hard thermal shutdown to prevent silicon melting. The sensor acts as the final fail-safe when passive thermal management fails.
What does an external heat sensor do that an internal silicon sensor cannot?
An internal MCU sensor only measures the junction temperature of the logic die itself. It cannot tell you how hot the power MOSFETs, LDOs, or battery cells on the other side of the PCB are. External sensors (like NTC thermistors or digital I2C sensors like the TMP102) are placed at specific thermal choke points—strapped to motor windings, pressed against lithium cell casings, or suspended in the enclosure's dead-air zones. They provide a spatial thermal map that a single internal diode simply cannot achieve.
What does a heat sensor do differently on a BMS compared to a logic board?
On a logic board, a heat sensor typically triggers performance throttling. On a Battery Management System (BMS) for LiFePO4 or Li-ion packs, the heat sensor is a critical safety interlock. Lithium cells undergo irreversible exothermic decomposition (thermal runaway) if charged below 0°C or pushed past 60°C. The BMS reads the NTC thermistors taped directly to the cell bodies and will physically open the charge/discharge MOSFETs if the cells breach the safe envelope, prioritizing fire prevention over system uptime.
