At the bench, knowing the resistance temperature detector definition is only half the battle. A Resistance Temperature Detector (RTD) is a precision sensor that exploits the predictable, highly linear change in electrical resistance of pure metals—typically platinum—as temperature fluctuates. But in power electronics, an RTD isn't just a thermometer; it is the critical feedback node in a closed-loop thermal management system. If you are designing inverter bridges, motor drives, or high-current DC-DC converters, relying on silicon junction temperature estimates isn't enough. You need physical, localized thermal data to keep your power stages from cooking themselves.
This guide bridges the gap between sensor theory and thermal reality. We will map RTD specifications to thermal path math, interpret derating curves, and size a real-world heatsink using actual wattage and airflow metrics.
RTD Sensor Selection and Thermal Specifications
The standard resistance temperature detector definition centers on the platinum PT100 or PT1000 element, which exhibits a temperature coefficient of resistance (TCR or alpha) of 0.00385 Ω/Ω/°C. At 0°C, a Pt100 reads exactly 100 Ω, while a Pt1000 reads 1000 Ω. The higher base resistance of the Pt1000 makes it vastly superior for embedded PCB applications because it minimizes lead-wire resistance errors and reduces self-heating from the excitation current.
When instrumenting a power module, you must choose the right sensor for the thermal environment. The table below breaks down the real-world trade-offs for high-power thermal monitoring.
| Sensor Type | Nominal Resistance (at 0°C) | Useful Temp Range | Linearity & Accuracy | Self-Heating Risk in High-Gain Circuits |
|---|---|---|---|---|
| Pt100 RTD (Wirewound) | 100 Ω | -200°C to +600°C | Excellent (±0.1°C), requires 3/4-wire Kelvin sensing | Moderate (1mA excitation = 0.1mW) |
| Pt1000 RTD (Thin-Film SMD) | 1000 Ω | -50°C to +150°C | Very Good (±0.3°C), 2-wire sensing often sufficient | Low (0.1mA excitation = 0.01mW) |
| NTC Thermistor (10k B25/85) | 10,000 Ω (at 25°C) | -40°C to +125°C | Poor (Highly non-linear, Steinhart-Hart math required) | High (Resistance drops at high temp, increasing current draw) |
| Type K Thermocouple | N/A (Generates μV signal) | -200°C to +1250°C | Fair (±2.2°C), requires cold-junction compensation | None (Passive generation), but susceptible to EMI noise |
For closed-loop thermal throttling on a modern IGBT or SiC MOSFET module, the surface-mount Pt1000 (such as the Heraeus M222 series) is the undisputed winner. It solders directly to the PCB pad adjacent to the power die, surviving reflow profiles while providing fast thermal transient tracking.
Thermal Path Math and Derating Curves
How hot is too hot? While silicon junctions (TJ) in modern power MOSFETs are rated for an absolute maximum of 150°C or 175°C, running them at those limits is a fast track to field failures. As a rule of thumb, every 10°C increase above 105°C halves the expected lifespan of the component due to accelerated electromigration and bond-wire fatigue. Your thermal design target should cap TJ at 105°C under worst-case continuous load.
To guarantee this, we use thermal path math. The total junction-to-ambient thermal resistance (RθJA) is the sum of the resistances in the heat flow path:
RθJA = RθJC + RθCS + RθSA
- RθJC (Junction-to-Case): Dictated by the silicon die and package. For a TO-247 SiC MOSFET, this is typically around 1.0°C/W.
- RθCS (Case-to-Sink): Dictated by your Thermal Interface Material (TIM). A 0.25mm thick Bergquist Sil-Pad 2000 yields roughly 0.3°C/W.
- RθSA (Sink-to-Ambient): Dictated by your heatsink and airflow. This is the only variable you control.
Worked Example: You are dissipating 30W continuous in a TO-247 package. Ambient enclosure temperature (TA) is 40°C. Target TJ is 125°C (allowing a safety margin below the 150°C absolute max).
ΔT = 125°C - 40°C = 85°C.
Required Total RθJA = 85°C / 30W = 2.83°C/W.
Required RθSA = 2.83 - (1.0 + 0.3) = 1.53°C/W.
You need a heatsink with a thermal resistance of 1.53°C/W or lower. For deeper mechanical mounting insights, the Texas Instruments SLPA005B Thermal Pad Application Note provides excellent torque-to-thermal-resistance data for TO-247 packages.
Heatsink Selection and Airflow Enclosure Changes
To hit our 1.53°C/W target, we cannot rely on passive convection for a standard extruded profile unless it is massive. We must introduce forced air. Let's look at the Aavid Thermalloy 7022B-MTG, a standard extruded aluminum heatsink for TO-220/TO-247 packages.
In natural convection (0 LFM - Linear Feet per Minute), the 7022B-MTG has an RθSA of roughly 4.5°C/W. That is far too high for our 30W load. However, heatsink manufacturers publish derating curves based on airflow velocity. According to Boyd Corporation (Aavid) Thermal Design Guides, pushing air across the fins drastically alters the boundary layer physics.
At 200 LFM (a gentle breeze), the RθSA drops to about 2.0°C/W. Still not quite enough.
At 400 LFM, the RθSA drops to approximately 1.4°C/W. This clears our 1.53°C/W requirement.
What airflow changes buy you:
Converting LFM to CFM (Cubic Feet per Minute) for enclosure design requires multiplying the LFM by the cross-sectional area of the airflow path. If your enclosure duct is 2 inches by 2 inches (0.278 sq ft), achieving 400 LFM requires a fan pushing roughly 111 CFM. Instead of a massive blower, you can optimize the enclosure by narrowing the duct cross-section directly over the fins. Using a Noctua NF-A4x20 PWM (40mm fan pushing ~20 CFM) in a tightly shrouded 1-inch by 1-inch duct yields an impressive 2880 LFM, dropping the heatsink resistance well below 1.0°C/W and giving you massive thermal headroom.
Failure Signatures and RTD Placement Strategy
Knowing where to place your RTD is just as critical as the math. If you epoxy a Pt100 to the plastic body of a TO-247 package, you are measuring the plastic temperature, not the silicon. The thermal gradient across the mold compound can easily be 15°C to 20°C higher at the junction. Place your SMD Pt1000 on the top-layer copper pour, directly adjacent to the source/drain pins, or embed a bare-wire Pt100 into a milled pocket in the aluminum heatsink base, 2mm below the mounting surface.
When thermal management fails, the hardware leaves specific forensic signatures before it catastrophically shorts:
- TIM Pump-Out: Over hundreds of thermal cycles, the coefficient of thermal expansion (CTE) mismatch between the silicon die and the aluminum heatsink causes the TIM to physically 'pump' out from the center of the package. Your RTD will show a slow, creeping increase in baseline case temperature over months of operation, even with identical ambient conditions.
- Solder Joint Creep: If the PCB copper pour gets too hot, the SAC305 solder attaching the power pins undergoes intermetallic growth and micro-cracking. This manifests as sudden, erratic spikes in your RTD readings during high di/dt switching events, caused by transient resistive heating at the cracked joint.
- Die Attach Delamination: The solder layer between the silicon die and the copper leadframe cracks. The thermal impedance (RθJC) spikes. You will see the junction temperature (calculated via the MOSFET's internal diode or an embedded sensor) diverge wildly from the case temperature measured by your external RTD.
By combining a rigorous understanding of the resistance temperature detector definition with hard Rθ math, you move from guessing if your heatsink is big enough to engineering a verifiable, closed-loop thermal system. Calibrate your RTD, respect the derating curves, and shroud your airflow.
