Comparing Types of Resistance Temperature Detector for Power Electronics

Why RTD Selection Dictates Thermal Path Accuracy

When you are pushing 50W through a TO-247 MOSFET or an IGBT module, guessing your junction temperature ($T_J$) based on a thermocouple taped to the case is a fast track to silicon failure. To properly validate a thermal design, you need to measure the case temperature ($T_C$) and ambient temperature ($T_A$) with high precision, allowing you to calculate the actual thermal resistance of your system. This is where understanding the different types of resistance temperature detector (RTD) sensors becomes critical.

An RTD calculates temperature by measuring the change in electrical resistance of a pure metal (usually platinum) as it heats up. Unlike thermocouples, which generate a micro-voltage at a junction and are prone to cold-junction compensation errors, RTDs offer absolute resistance values that map directly to the ITS-90 temperature scale (NIST ITS-90). For power electronics thermal management, the right RTD type gives you the sub-degree accuracy required to read derating curves and size heatsinks without over-engineering the cooling system.

Comparing Types of Resistance Temperature Detector

Not all RTDs are built for the same thermal environment. The choice between element materials, construction methods, and wiring topologies dictates your sensor's response time, self-heating error, and noise immunity.

Calculating Thermal Resistance (Rθ) and Derating

To use an RTD for thermal validation, we rely on the fundamental thermal path equation from junction to ambient:

Rθ_JA = Rθ_JC + Rθ_CS + Rθ_SA

Where:
Rθ_JC = Junction-to-Case thermal resistance (from the silicon datasheet).
Rθ_CS = Case-to-Sink thermal resistance (your thermal interface material).
Rθ_SA = Sink-to-Ambient thermal resistance (your heatsink and airflow).

We cannot measure $T_J$ (junction temperature) directly without destructive physical decapsulation. Instead, we bond a PT1000 thin-film RTD directly to the MOSFET case to measure $T_C$, and suspend a second PT100 in the enclosure to measure $T_A$. Using the known power dissipation ($P_D$), we calculate the actual $T_J$:

T_J = T_C + (P_D × Rθ_JC)

Interpreting the Derating Curve

Once your RTD confirms your $T_C$, you must cross-reference the datasheet's Safe Operating Area (SOA) and current derating curve. A typical MOSFET derating curve shows 100% continuous drain current ($I_D$) at $T_C$ = 25°C, dropping linearly to 0% at $T_C$ = 150°C. If your RTD reads a case temperature of 90°C during a 30A load test, the curve dictates you are only allowed to pull ~60% of the rated 25°C current. If your design requires 30A, and 30A represents 75% of the datasheet rating, your thermal design has failed the derating check, and you must lower $T_C$.

Heatsink Selection & Airflow: What the Data Tells You

Let’s run a concrete heatsink selection based on RTD feedback. Suppose you are designing a motor drive using an Infineon IKW40N120 IGBT (TO-247 package).

• Power Dissipation ($P_D$): 35W (measured via VCE(sat) × Ic)
• Target $T_J$: 115°C (derated from 175°C max)
• Enclosure Ambient ($T_A$): 45°C (measured via suspended PT100)
• Rθ_JC: 0.6 °C/W (datasheet)
• Rθ_CS: 0.4 °C/W (using Bergquist Sil-Pad 400VOS thermal pad)

Step 1: Calculate Required Rθ_JA
Rθ_JA = (T_J - T_A) / P_D = (115 - 45) / 35 = 2.0 °C/W

Step 2: Calculate Required Heatsink Rθ_SA
Rθ_SA = Rθ_JA - Rθ_JC - Rθ_CS = 2.0 - 0.6 - 0.4 = 1.0 °C/W

Step 3: Select the Heatsink and Airflow
We need a heatsink that achieves ≤ 1.0 °C/W. Looking at the Boyd/Aavid thermal catalogs, the Wakefield-Vette 680-125AB extruded aluminum TO-247 heatsink has an Rθ_SA of ~3.2 °C/W in natural convection. This is insufficient. However, the derating curve for this heatsink profile shows that at 400 LFM (Linear Feet per Minute, or ~2 m/s) forced airflow, the Rθ_SA drops to 0.9 °C/W.

The Fix: We mount the 680-125AB and add a 40mm x 40mm axial fan (e.g., Sunon MF40101VX) blowing directly across the fins. We verify the fix by re-reading the case-mounted PT1000. If the RTD now reads $T_C$ = 71°C, our calculated $T_J$ is 71 + (35 × 0.6) = 92°C. We are well below our 115°C target. Furthermore, simply adding a 15cm² exhaust vent to the top of the enclosure drops the internal $T_A$ by roughly 4°C, buying you an additional ~0.1 °C/W of system margin.

Failure Signatures of Thermal Stress

When thermal management fails, the silicon doesn't always melt immediately. Often, the failure signatures manifest in the physical materials, which your RTD data can help diagnose if you track trends over time:

• Die Attach Delamination: If your RTD shows $T_C$ remaining stable at 60°C, but the silicon is suddenly tripping its internal thermal shutdown, the Rθ_JC has increased. The solder or sinter layer between the silicon die and the copper leadframe has cracked or voided due to thermal cycling. Heat is trapped in the junction while the case stays cool.
• Thermal Pad Creep and Pump-Out: Over hundreds of heat/cool cycles, the TO-247 package expands and contracts at a different rate than the aluminum heatsink. This "pump-out" effect squeezes liquid thermal grease out of the interface, increasing Rθ_CS. Using a phase-change material (like Honeywell PTM7950) or a solid elastomer pad prevents this.
• RTD Self-Heating Error: If your PT1000 reads 2°C higher than a reference thermocouple in still air, check your excitation current. Pushing 1mA through a 1000Ω RTD dissipates 1mW ($I^2R$). In a vacuum or still air, this self-heating can cause a +0.5°C to +1.0°C false reading. Drop your ADC excitation current to 0.1mA (0.01mW dissipation) to eliminate this error.

The Decision Tree: Pick Your Sensor

Stop guessing which sensor to buy. Use this decision matrix to terminate your selection process based on your specific thermal measurement scenario.