How to Size a Heat Sink for Power Electronics: Step-by-Step Guide

Before selecting a heat sink, you need to understand the thermal path from your power semiconductor's junction to the ambient air. This path consists of several thermal resistances in series:

Junction-to-Case (Rth_jc): This is specified in the semiconductor datasheet. For a typical IGBT, it might be 0.1-0.5°C/W. This resistance is fixed by the device design.

Case-to-Sink (Rth_cs): This depends on the thermal interface material (TIM) you choose. Typical values:

• Thermal grease: 0.1-0.3°C/W
• Thermal pad: 0.3-1.0°C/W
• No TIM (metal-to-metal): 1.0-3.0°C/W

Sink-to-Ambient (Rth_sa): This is what you're trying to determine—the required heat sink thermal resistance.

The fundamental equation is: Tj = Ta + P × (Rth_jc + Rth_cs + Rth_sa)

Where:

• Tj = Junction temperature (°C)
• Ta = Ambient temperature (°C)
• P = Power dissipation (W)

Start by defining your constraints:

Maximum Junction Temperature (Tj_max): Check your semiconductor datasheet. Most power devices specify Tj_max of 150-175°C, but for reliability, derate by 20-30%. A design target of 100-125°C is common.

Maximum Ambient Temperature (Ta_max): Consider your worst-case operating environment:

• Indoor electronics: 40-50°C
• Outdoor/industrial: 50-60°C
• Under-hood automotive: 85-105°C

Power Dissipation (P): Calculate worst-case power loss in your semiconductor:

• Conduction losses: P_cond = I²×Rds_on (MOSFET) or Vce_sat×I (IGBT)
• Switching losses: P_sw = 0.5×V×I×(t_on + t_off)×f_sw
• Total: P_total = P_cond + P_sw

Example Calculation:

• Tj_max (derated) = 125°C
• Ta_max = 50°C
• P = 100W
• Rth_jc = 0.3°C/W
• Rth_cs = 0.2°C/W (with thermal grease)

Required Rth_sa = (Tj_max - Ta_max)/P - Rth_jc - Rth_cs Rth_sa = (125 - 50)/100 - 0.3 - 0.2 = 0.25°C/W

You need a heat sink with thermal resistance ≤0.25°C/W.

Now you need to find a heat sink that achieves your required Rth_sa. Heat sink performance depends on:

Heat Sink Types:

1. Extruded Aluminum: Most common, cost-effective for medium power. Typical Rth_sa: 0.5-5°C/W depending on size.

2. Bonded Fin: Higher performance for given footprint. Fins are bonded or brazed to base. Rth_sa: 0.1-1°C/W.

3. Skived Fin: Fine pitch fins cut from solid block. Excellent performance, higher cost. Rth_sa: 0.1-0.5°C/W.

4. Die-Cast: Complex shapes possible, good for high volume. Performance varies widely.

5. Liquid-Cooled Cold Plates: For very high power density. Rth_sa equivalent: 0.01-0.1°C/W.

Selection Process:

1. Calculate required heat sink surface area as a starting point: A ≈ P / (h × ΔT) Where h ≈ 10-25 W/m²K for natural convection, 50-150 W/m²K for forced air

2. Search heat sink catalogs for products matching your footprint and Rth_sa requirements

3. Consider mounting constraints—does the heat sink base match your component layout?

4. Account for altitude derating if applicable (air density decreases ~12% per 1000m)

If natural convection doesn't provide adequate cooling, you'll need forced air. Heat sink datasheets typically provide Rth_sa at multiple airflow velocities.

Airflow Velocity Estimation: For a fan blowing across a heat sink: V = CFM / (60 × A_flow)

Where:

• V = air velocity (ft/s or m/s)
• CFM = fan airflow rating
• A_flow = cross-sectional flow area (ft² or m²)

Important Considerations:

1. System Impedance: The actual airflow through your heat sink depends on system pressure drop. A "200 CFM" fan might only deliver 50 CFM in a constrained enclosure.

2. Fan Curve: Use the fan's pressure-flow curve to determine actual operating point based on system impedance.

3. Velocity Profile: Air velocity varies across the heat sink. Center channels typically see higher velocity than edges.

4. Temperature Rise: Air temperature increases as it passes through the heat sink. Account for this in your calculations.

Rule of Thumb: Forced air at 2-3 m/s typically reduces Rth_sa by 40-60% compared to natural convection.

Analytical calculations provide a starting point, but CFD simulation is essential for optimizing your design and catching problems early.

What CFD Reveals:

1. Hot Spots: Localized temperature peaks that analytical methods miss
2. Flow Distribution: How air actually flows through your heat sink (often different from assumptions)
3. Thermal Coupling: How multiple heat sources interact
4. Transient Behavior: Temperature response during power cycling

Simulation Best Practices:

1. Model Setup:

   • Include all significant heat sources, not just the main component
   • Use accurate material properties (temperature-dependent if needed)
   • Model TIM layers explicitly—they significantly impact results

2. Boundary Conditions:

   • Apply realistic inlet conditions (temperature, velocity profile)
   • Don't forget radiation, especially for natural convection
   • Consider solar loading for outdoor equipment

3. Mesh Quality:

   • Use inflation layers on all solid-fluid interfaces
   • Refine mesh in regions with high temperature gradients
   • Perform mesh independence study

4. Validation:

   • Always compare simulation to thermal measurements
   • Adjust contact resistances and boundary conditions to match data
   • Document assumptions and validation results

CFD gets you close, but nothing replaces physical testing to validate your thermal design.

Testing Methods:

1. Thermocouple Measurements:

   • Place thermocouples at critical locations (component cases, heat sink base, air inlet/outlet)
   • Use thermal paste to ensure good contact with surfaces
   • Allow system to reach steady state (typically 30-60 minutes)

2. Thermal Imaging:

   • Reveals temperature distribution across surfaces
   • Good for identifying hot spots and flow patterns
   • Requires knowledge of surface emissivity for accurate readings

3. Power Cycling Tests:

   • Simulate actual operating profiles
   • Verify thermal time constants match predictions
   • Identify any thermal ratcheting or progressive degradation

Test Conditions:

• Test at worst-case ambient (use environmental chamber if needed)
• Apply worst-case power dissipation
• If using fans, test at end-of-life airflow (fans degrade over time)
• Run for sufficient time to reach thermal equilibrium

Acceptance Criteria:

• All junction temperatures within limits with margin
• Temperature distribution matches simulation predictions (±10-15%)
• No unexpected hot spots or flow problems
• Thermal time constants appropriate for application

Learn from these frequent errors:

1. Using Datasheet Rth_sa Directly: Heat sink datasheets often specify Rth_sa under ideal conditions (uniform heat source, optimal orientation, specific airflow). Real-world performance may be 20-50% worse.

2. Ignoring Spreading Resistance: If your heat source is much smaller than the heat sink base, thermal spreading resistance can be significant. A small die on a large heat sink doesn't use the full fin area effectively.

3. Forgetting Interface Resistance: The TIM layer often contributes 0.1-0.5°C/W to the thermal path. This can be 10-50% of your thermal budget.

4. Assuming Linear Scaling: Doubling heat sink size doesn't halve thermal resistance. Returns diminish as size increases.

5. Neglecting Altitude Effects: At high altitude, reduced air density degrades both natural and forced convection. Derate heat sink performance accordingly:

• 3000m (10,000 ft): ~25% reduction
• 5000m (16,000 ft): ~40% reduction

6. Optimistic Airflow Assumptions: System airflow is almost always less than free-air fan ratings. Use system-level CFD or testing to determine actual airflow.

7. Ignoring Transient Requirements: Steady-state calculations don't capture temperature spikes during power transients. Verify thermal mass is adequate for your duty cycle.

Free Resource: Download our Thermal Design Checklist — an 8-page checklist covering all critical thermal design considerations for power electronics.