Electronics

Thermal Cheat Sheet

Quick reference for thermal-resistance, dissipation, and heatsink selection.

Core equation

Heat flowΔT = P · R_θ (analogous to V = IR)
Junction tempT_J = T_A + P · (R_θJC + R_θCS + R_θSA)
DeratingTarget T_J ≤ 0.8 × T_J,max

Approximate heat transfer coefficients (W/m²·K)

Mechanismh
Natural convection (air)5 – 25
Forced convection (air)10 – 500
Natural convection (water)50 – 1 000
Forced convection (water)300 – 12 000
Boiling water2 500 – 25 000
Radiation (painted matte)5 – 10 (at 300 K)

Heatsink rules of thumb

  • Orient fins vertically for natural convection.
  • Leave 25 mm clearance above/below fins.
  • Black-anodize fins for +10–15% radiative boost over bare aluminum.
  • Forced air (small fan) typically halves R_θSA.
  • Heat pipe inside bigger sinks spreads heat from the die footprint across the full fin area.
  • TIM: use just enough to wet the surface — excess increases resistance.

Ballpark TO-220 dissipation

SetupR_θJASafe P @ 25 °C ambient
No sink~60 °C/W~1.5 W
Small clip sink~20 °C/W~4 W
Medium extruded~8 °C/W~11 W
Large + fan~3 °C/W~30 W

Worked example: picking a heatsink

A TO-220 regulator dissipates 5 W in a 40 °C enclosure. Its maximum junction temperature is 125 °C, and the 80% derating target is 100 °C, so the allowed rise is 60 °C. The total thermal resistance can be at most 60 / 5 = 12 °C/W. Junction-to-case is 3 °C/W and the insulating pad adds 0.5 °C/W, so the heatsink must provide no more than 12 − 3 − 0.5 = 8.5 °C/W. A medium extruded heatsink (about 8 °C/W) just qualifies; a better choice is a smaller one with a fan or one rated near 5 °C/W. With no heatsink at all, 5 W × 60 °C/W = 300 °C would destroy the part.

Convection sanity check: heat leaving a surface is P = h × A × ΔT. A 10 cm by 10 cm plate (0.01 m²) with natural convection h = 10 W/m²K that is 50 °C above ambient sheds 10 × 0.01 × 50 = 5 W. This is why a bare small enclosure cannot dissipate tens of watts without airflow or a finned surface.

Common mistakes

  • Using ambient temperature outside the enclosure instead of the air temperature inside, which can be 10 to 20 degrees C higher around hot components.
  • Adding thermal resistances from different datasheets that assume different test conditions. RθJA from a datasheet depends on the test board and is not transferable to your layout.
  • Forgetting that the heatsink warms its neighbors. A sink at 80 °C is a heat source for nearby capacitors, whose life halves for each 10 °C.
  • Ignoring the thermal mass for pulsed loads: short pulses can be handled by a part with a large heat capacity even if the average power exceeds the steady-state limit. Use the transient thermal impedance curve in the datasheet.
Was this article helpful?