A Resistance Temperature Detector (RTD) is a precision temperature sensor that calculates heat by measuring the predictable, highly linear change in electrical resistance of a pure metal—almost always platinum—as temperature changes. In power electronics and embedded systems, you cannot manage what you cannot measure. While cheap NTC thermistors are fine for basic ambient weather stations, they drift and fail to provide the lab-grade accuracy required to validate thermal models in high-power designs. When you need to know exactly how hot your power stage is running to prevent catastrophic silicon failure, the RTD (specifically the Pt100 or Pt1000 variant) is the gold standard.
Validating the Thermal Path: Rtheta Math Meets RTD Data
To understand why an RTD is critical, you have to look at the thermal path from the silicon junction to the ambient air. Every power component has a Junction-to-Ambient thermal resistance, denoted as RθJA (measured in °C/W). The fundamental thermal equation is:
TJ = TA + (PD × RθJA)
Where TJ is junction temperature, TA is ambient temperature, and PD is power dissipation. Let us run a concrete bench example. Suppose you are driving an IGBT in a motor controller that dissipates 50W. The datasheet lists an RθJA of 1.5 °C/W (with a specific heatsink), and your enclosure ambient (TA) is 40°C.
TJ = 40 + (50 × 1.5) = 115°C.
Because you cannot physically probe the silicon junction (TJ) without destroying the package, you measure the case temperature (TC) using a surface-mount Pt1000 RTD epoxied directly to the exposed thermal pad. By knowing the Junction-to-Case resistance (RθJC), you calculate backward: TJ = TC + (PD × RθJC). The RTD provides the exact TC data point needed to prove your thermal design works in the real world.
Heatsink Selection and Airflow: What the RTD Tells You
Let us say your RTD reads 125°C on the IGBT case during a 50W continuous load test, and your maximum safe TJ is 150°C. With an RθJC of 0.5 °C/W, your actual TJ is 150°C. You are right on the absolute limit. You need a better heatsink.
To find the required Sink-to-Ambient thermal resistance (RθSA), we use the formula: RθSA = (TC - TA) / PD. If we want to drop the case temp to a safe 90°C in a 40°C ambient: RθSA = (90 - 40) / 50 = 1.0 °C/W.
Looking at real-world extruded aluminum heatsinks, the Boyd (formerly Aavid) 7023B-MTG is a common TO-247/TO-220 stamped heatsink. In natural convection (0 LFM - Linear Feet per Minute airflow), its RθSA is roughly 1.8 °C/W. That is not good enough. However, if you add a 40mm DC fan pushing 200 LFM of airflow across the fins, the derating curve shows the RθSA drops to approximately 0.6 °C/W.
Interpreting the Derating Curve: Heatsink manufacturers provide a graph plotting Temperature Rise (°C) against Power Dissipation (W). The slope of this line is your RθSA. A steeper slope means worse thermal performance. By reading the curve at the 200 LFM line, you can visually confirm that at 50W, the temperature rise above ambient will only be about 30°C (yielding a 70°C case temp), safely buying you the thermal headroom your design needs.
| Sensor Type | Base Resistance | Linearity | Best Use Case |
|---|---|---|---|
| Pt100 RTD | 100 Ω @ 0°C | Excellent | Industrial 3-wire/4-wire probes |
| Pt1000 RTD | 1000 Ω @ 0°C | Excellent | PCB-mount, 2-wire microcontroller ADC |
| NTC Thermistor (10k) | 10,000 Ω @ 25°C | Poor (Exponential) | Basic ambient sensing, inrush limiting |
| Type K Thermocouple | N/A (Generates mV) | Good | Extreme temps (>300°C), exhaust gas |
Failure Signatures: How Hot is Too Hot and Thermal Stress
How hot is too hot for a power semiconductor? While a modern silicon MOSFET or IGBT might have an absolute maximum TJ rating of 150°C or 175°C on the first page of the datasheet, running it there is engineering malpractice. For long-term reliability, you should design your thermal path to keep TJ below 105°C.
When your RTD data shows you are consistently pushing past 125°C, you will begin to see specific failure signatures of thermal stress:
- Thermal Runaway: The RDS(on) (on-state resistance) of a MOSFET has a positive temperature coefficient. As the die gets hotter, resistance increases. Higher resistance means more I2R power loss, which generates more heat, further increasing resistance until the silicon melts or the package vents.
- Solder Joint Fatigue: Every material has a Coefficient of Thermal Expansion (CTE). The silicon die, the copper leadframe, and the FR4 PCB all expand at different rates when heated. Repeated thermal cycling (heating up under load, cooling down at idle) causes microscopic shear stresses. Over time, cracks form in the die-attach solder layer, increasing RθJC and eventually causing an open circuit.
- Gate Oxide Degradation: High junction temperatures accelerate Time-Dependent Dielectric Breakdown (TDDB). The thin silicon dioxide layer insulating the gate will eventually fail, leading to a catastrophic gate-to-source short.
According to Texas Instruments thermal design guidelines, every 10°C increase in operating temperature roughly halves the expected lifespan of the component due to these accelerated chemical and mechanical degradation mechanisms.
Resistance Temperature Detector FAQ
What is the difference between a Pt100 and Pt1000 resistance temperature detector?
A Pt100 has a resistance of 100 ohms at 0°C, while a Pt1000 has 1000 ohms at 0°C. Both use platinum and share the same temperature coefficient curve. However, for PCB-mounted electronics and microcontroller integration, the Pt1000 is vastly superior. Because its base resistance is 10 times higher, the resistance of the copper trace leads (which might add 1-2 ohms) introduces a negligible error. A Pt100 requires expensive 3-wire or 4-wire Kelvin sensing to cancel out lead resistance, whereas a Pt1000 can often be read accurately with a simple 2-wire voltage divider.
How do I wire a resistance temperature detector to an Arduino or ESP32?
Do not wire an RTD directly to the ESP32's internal 12-bit ADC; the internal ADC is notoriously non-linear and noisy, which will ruin the precision advantage of the RTD. Instead, use a dedicated RTD-to-Digital converter IC like the MAX31865. The MAX31865 provides the precision constant-current excitation needed to read the RTD without causing self-heating, handles the linearization math in hardware, and outputs the exact temperature via SPI to your Arduino or ESP32. If you must use an analog approach, use an external 16-bit ADC like the ADS1115 with a precision 0.1% reference resistor.
Why use an RTD instead of an NTC thermistor for power electronics?
NTC (Negative Temperature Coefficient) thermistors are cheap and highly sensitive, but they suffer from severe non-linearity, meaning you need complex Steinhart-Hart equations in your firmware just to get a readable temperature. More importantly, NTCs suffer from long-term drift and self-heating errors. In safety-critical thermal shutdowns—like a 5kW solar inverter or an EV motor controller—you need a sensor that will read exactly the same at 120°C today as it will five years from now. RTDs offer that long-term stability and linearity, ensuring your thermal protection triggers exactly when designed, preventing catastrophic field failures.
