How the Temperature and Resistance Formula Dictates Heatsink Sizing

The Dual Nature of Resistance: Electrical vs. Thermal

When designing power electronics, the term "resistance" pulls double duty. You are simultaneously fighting electrical resistance (which generates heat) and thermal resistance (which restricts heat flow). Understanding the temperature and resistance formula in both domains is the difference between a reliable power supply and a melted silicon die.

In the electrical domain, the resistance of a conductor changes with temperature according to the linear approximation formula:

R(T) = R_0 [1 + α(T - T_0)]

Here, R_0 is the baseline resistance at reference temperature T_0 (usually 20°C or 25°C), and α (alpha) is the temperature coefficient of resistance. For copper PCB traces, α is roughly 0.00393/°C. But for the silicon channel in a power MOSFET, the R_DS(on) has a positive temperature coefficient that typically increases by 0.4% to 0.7% per degree Celsius. By the time a MOSFET junction hits 125°C, its R_DS(on) is often 1.5 to 2.0 times higher than the 25°C datasheet value. If you size your thermal management based solely on the 25°C electrical resistance, your component will silently overheat due to increased I²R losses at operating temperature.

To manage this, we shift to the thermal domain, where we use thermal resistance (R_θ), measured in °C/W. This is the thermal equivalent of Ohm’s Law: ΔT = P × R_θ. The temperature difference across a material equals the power dissipated multiplied by its thermal resistance.

Thermal Path Math: Junction-to-Ambient Calculations

Heat flows from the silicon junction (J) through the case (C), across a thermal interface material to the heatsink (S), and finally into the ambient air (A). The governing thermal path formula is:

T_J = T_A + P_D × (R_θJC + R_θCS + R_θSA)

Let’s walk through a real-world sizing example. We are driving a 12A continuous load using an Infineon IRFZ44N MOSFET in a TO-220 package. The ambient temperature inside our enclosure (T_A) is 45°C. We want to keep the maximum junction temperature (T_J) under 110°C for long-term reliability, well below the 175°C absolute maximum.

First, calculate power dissipation (P_D). The IRFZ44N has a typical R_DS(on) of 17.5mΩ at 25°C. Factoring in the silicon temperature coefficient at our target 110°C, the hot resistance is roughly 30mΩ. P_D = I² × R = 12² × 0.030 = 4.32W. (Note how much higher this is than the 2.52W you would calculate using the 25°C datasheet value).

Plugging into the thermal formula:
110 = 45 + 4.32 × (1.0 + 0.5 + R_θSA)
65 = 4.32 × (1.5 + R_θSA)
15.04 = 1.5 + R_θSA
R_θSA = 13.54 °C/W

We need a heatsink with a thermal resistance of 13.54 °C/W or lower. Because our wattage basis is relatively low (4.32W), a small extruded aluminum clip-on sink is sufficient. For higher wattage designs, this math scales linearly, demanding progressively massive heat sinks.

Heatsink Selection, Derating, and Airflow Dynamics

With a target R_θSA of 13.54 °C/W, we can select a real component. The Wakefield-Vette 210-1 is a classic TO-220 extruded aluminum heatsink. In natural convection (still air), its thermal resistance is approximately 12.5 °C/W, which safely clears our 13.54 °C/W requirement.

However, heatsinks do not perform in a vacuum. Datasheets provide derating curves that show how maximum allowable power drops as ambient temperature rises. A typical power semiconductor is rated for 100% power dissipation at 25°C case temperature, and derates linearly to 0W at 175°C. If your enclosure traps heat, raising the local ambient to 75°C, the part is already derated by roughly 33% before it even turns on. According to DigiKey's thermal management guidelines, failing to account for enclosure ambient temperature is the most common cause of field failures in power supplies.

What airflow and enclosure changes buy you: Adding a 40mm fan (like a Noctua NF-A4x10) blowing at 1.0 m/s across the fins cuts the sink-to-ambient resistance in half. Conversely, placing a passive heatsink inside a sealed IP65 enclosure without thermal coupling to the chassis walls effectively turns the enclosure into an oven. If you must use a sealed enclosure, bolt the TO-220 directly to the aluminum chassis wall using a thermal pad, treating the entire enclosure mass as your heatsink.

Failure Signatures: How Hot is Too Hot?

Silicon junctions can physically survive up to 150°C–175°C, but how hot is too hot for this part from a reliability standpoint? The industry relies on the Arrhenius equation, which models reaction rates (like chemical degradation inside a chip) against temperature. The practical rule of thumb in power electronics is that every 10°C increase in operating temperature above 85°C halves the expected lifespan of the component. Designing for a 100°C junction temperature yields a vastly more reliable product than pushing it to 130°C.

When thermal limits are exceeded or poorly managed, components exhibit distinct failure signatures:

• Solder Joint Fatigue: Repeated thermal cycling (power on/off) causes the silicon die, copper leadframe, and tin-lead or SAC305 solder to expand and contract at different rates due to Coefficient of Thermal Expansion (CTE) mismatch. This leads to micro-cracking at the solder joints, eventually causing an open circuit or high-resistance thermal bottleneck.
• Electromigration: At high temperatures and high current densities, metal atoms in the silicon metallization layers physically migrate. Over thousands of hours, this creates voids (opens) or hillocks (shorts) inside the IC. As noted in Texas Instruments application note SNVA489, keeping the junction below 105°C drastically mitigates electromigration risks in standard silicon processes.
• Thermal Runaway: In Bipolar Junction Transistors (BJTs), the base-emitter voltage (V_BE) drops as temperature rises, causing the transistor to draw more current, which generates more heat. In MOSFETs, the positive temperature coefficient of R_DS(on) increases conduction losses. If the heatsink cannot dissipate this escalating wattage, the part destroys itself in a catastrophic short-circuit failure, often cracking the epoxy package.

Frequently Asked Questions

How does the temperature and resistance formula apply to copper traces on a PCB?

The same linear formula R(T) = R_0 [1 + α(T - T_0)] applies to your PCB copper. If you design a 1oz copper trace to carry 5A, and it heats up from 25°C to 75°C, its resistance increases by roughly 20%. This means the voltage drop across the trace and the I²R heat generation will also increase by 20%. For high-current paths (like battery management system shunts or motor controller phase traces), you must calculate trace width using the hot resistance, not the room-temperature baseline, to prevent localized PCB hot spots.

Why does my MOSFET get hotter when I add a larger gate resistor?

This seems counterintuitive, but it relates to switching losses rather than conduction losses. A larger gate resistor slows down the dV/dt and di/dt during turn-on and turn-off transitions. During this transition period, the MOSFET is in the linear region, simultaneously sustaining high voltage and high current. The longer the transition takes, the more switching energy is dissipated as heat per cycle. If you are driving a PWM signal at 50kHz, a gate resistor that is too large will cause switching losses to dwarf conduction losses, overwhelming your heatsink.

Can I use the temperature and resistance formula to calculate NTC thermistor values?

No, the linear α formula only works for pure metals like copper, aluminum, and tungsten over limited temperature ranges. NTC (Negative Temperature Coefficient) thermistors are made of sintered metal oxides and exhibit a highly non-linear, exponential drop in resistance as temperature rises. To calculate the temperature and resistance relationship for an NTC thermistor, you must use the Steinhart-Hart equation: 1/T = A + B(ln R) + C(ln R)³, where A, B, and C are specific coefficients provided by the thermistor manufacturer (like Vishay or Murata). For simpler approximations over a narrow range, the Beta (β) parameter equation is used, but it will yield significant errors if extrapolated beyond a 20°C window.