How Does a Resistance Temperature Detector Work in Thermal Management?

The Physics of RTDs: How a Resistance Temperature Detector Works

A Resistance Temperature Detector (RTD) operates on a fundamental metallurgical principle: the electrical resistance of pure metals increases in a highly predictable, nearly linear fashion as temperature rises. In power electronics thermal management, we almost exclusively use Platinum RTDs (Pt100 or Pt1000) because platinum offers superior chemical stability and a repeatable temperature coefficient of resistance (TCR), typically standardized at an alpha (α) of 0.00385 Ω/Ω/°C.

For a standard Pt100 sensor, the nominal resistance is exactly 100Ω at 0°C. Using the simplified Callendar-Van Dusen equation for temperatures above 0°C, the resistance at temperature T is:

R(T) = R₀(1 + αT)

At a 100°C junction temperature, a Pt100 RTD will measure precisely 138.5Ω. This linearity is what makes RTDs the gold standard for closed-loop thermal management in motor drives, inverters, and high-power LED arrays.

Sensor Selection for Power Electronics

Translating RTD Readings into Thermal Path Math (Rθ)

Knowing how does a resistance temperature detector work is only half the battle; the real engineering value comes from mapping that RTD data to the silicon junction. In power electronics, we rarely place the RTD directly on the silicon die. Instead, we bond it to the exposed copper tab of a TO-247 or TO-220 package to measure Case Temperature (T_C).

To find the actual Junction Temperature (T_J), we rely on the Junction-to-Case thermal resistance (R_θJC) provided in the manufacturer's datasheet. The governing thermal path equation is:

T_J = T_C + (P_D × R_θJC)

Worked Thermal Path Example

Let’s assume we are driving an Infineon IKW40N120H3 IGBT in a solar inverter.

• Dissipation (P_D): 50W (measured via V_CE(sat) × I_C)
• Datasheet R_θJC: 0.75 °C/W
• RTD Measured T_C: 85°C

T_J = 85°C + (50W × 0.75 °C/W) = 122.5°C.

This tells us the silicon is running hot but within safe limits. If the RTD was misapplied and read ambient air instead of the case, our thermal loop would falsely assume T_J was safe, leading to catastrophic failure.

Heatsink Selection and Derating Curves

How hot is too hot? While modern silicon can physically survive up to 150°C or 175°C, reliability engineering dictates a maximum design T_J of 125°C. Every 10°C increase above this baseline halves the expected operational lifespan of the component due to accelerated electromigration and dielectric breakdown.

If your RTD indicates T_C is pushing your T_J past 125°C, you must lower the Sink-to-Ambient thermal resistance (R_θSA). We calculate the maximum allowable R_θSA using the total thermal path:

R_θSA ≤ [(T_J(max) - T_A) / P_D] - R_θJC - R_θCS

Using our previous 50W IGBT example, assuming a worst-case ambient enclosure temperature (T_A) of 40°C, and using a Bergquist Sil-Pad 2000S200 thermal interface material (R_θCS ≈ 0.20 °C/W):

R_θSA ≤ [(125 - 40) / 50] - 0.75 - 0.20
R_θSA ≤ 1.7 - 0.95 = 0.75 °C/W

Real-World Heatsink and Airflow Selection

To achieve an R_θSA of 0.75 °C/W or better, a passive stamping won't cut it. You need an extruded aluminum profile. A Fischer Elektronik SK 409 150 SA profile provides roughly 0.45 °C/W in natural convection, giving you a comfortable 40% thermal margin.

What airflow and enclosure changes buy you: If your enclosure is sealed and ambient rises to 60°C, your required R_θSA drops to 0.35 °C/W. By adding a Sunon MF40101VX 40mm MagLev fan pushing 16 CFM directly across the SK 409 fins, you force the convective heat transfer coefficient higher, dropping the effective R_θSA to approximately 0.25 °C/W, rescuing the design without enlarging the enclosure.

Interpreting the Power Derating Curve

When reading a MOSFET or IGBT datasheet, the Safe Operating Area (SOA) and Power Derating curves are critical. A standard derating curve shows 100% power dissipation capability at T_C = 25°C, dropping linearly to 0% at T_C = 150°C.

If your RTD reads a case temperature of 85°C, the device is operating at 65°C above the baseline. (150 - 85) / (150 - 25) = 52%. Your 300W rated IGBT can now only safely dissipate 156W. If your control loop ignores this RTD derating data and pushes 200W through the device, the silicon will enter thermal runaway.

Failure Signatures: What Thermal Stress Looks Like on the Bench

When thermal management fails, the autopsy on the bench reveals specific physical signatures. Recognizing these helps you tune your RTD placement and alarm thresholds.

• Bond Wire Lift-Off: The aluminum wires connecting the silicon die to the package leads expand at a different rate than the silicon. Cyclic thermal stress (read by the RTD as wide T_C swings) causes fatigue at the wedge bonds, eventually snapping the wire and resulting in an open-circuit failure.
• Solder Joint Creep and Kirkendall Voids: The die-attach solder degrades under sustained high temperatures. Voids form, increasing R_θJC. You will notice this on the bench when the RTD reads a normal T_C, but the device still triggers its internal thermal shutdown—the hidden R_θJC has degraded, meaning T_J is much higher than the math predicts.
• RTD Delamination: If the Kapton tape or thermally conductive epoxy securing the RTD to the TO-247 tab degrades, the sensor lifts off. It begins reading ambient air temperature instead of case temperature. The thermal management loop assumes the system is cool, drives the semiconductor harder, and melts the package.

RTD Thermal Management FAQ

How does a 4-wire resistance temperature detector work to eliminate lead errors?

In a 2-wire RTD setup, the resistance of the copper lead wires adds directly to the Pt100 measurement. If your leads are 2Ω each, your ADC reads 104Ω at 0°C, introducing a massive 10°C error. A 4-wire (Kelvin) RTD uses two wires to inject the excitation current and two separate wires to measure the voltage drop across the element. Because the ADC's sense inputs have high impedance, virtually zero current flows through the sense wires, meaning no voltage drop occurs across them. This isolates the pure Pt100 resistance, yielding sub-0.1°C accuracy required for precision thermal throttling.

How does a resistance temperature detector work compared to a thermistor for fast thermal transients?

While an RTD offers superior accuracy and linearity for steady-state heatsink management, it has higher thermal mass (especially wirewound Pt100 probes). If you are protecting a MOSFET from microsecond short-circuit thermal spikes, an RTD is too slow to react. In these transient scenarios, a glass-encapsulated NTC thermistor (like the Vishay NTCLE100E3103) is preferred. Its tiny bead geometry yields a thermal time constant (Tau) of under 1 second, allowing the comparator circuit to trip the gate driver before the RTD even registers the temperature delta.

How does a surface-mount resistance temperature detector work on a PCB copper pour?

Surface-mount RTDs (like the TE Connectivity PTS1206) are soldered directly onto a PCB. To use them for thermal management of a nearby power IC, you must route a thermal via array from the IC's exposed pad to the copper pour beneath the RTD. The RTD measures the temperature of the PCB copper, not the silicon directly. Because FR4 fiberglass is a severe thermal insulator, the RTD relies entirely on the copper planes to conduct the heat. You must empirically calibrate the offset between the RTD's copper reading and the actual junction temperature in a thermal chamber before deploying the algorithm in production.