Electronics

Thermal Resistance

Thermal-resistance model for heat flow — R_θJA, R_θJC, R_θCS, and heatsink selection.

Model

ΔT = P · R_θLike Ohm's law — temperature rise for given heat
R_θJAJunction to ambient — no heatsink
R_θJCJunction to case — datasheet value
R_θCSCase to sink (TIM + contact)
R_θSASink to ambient — heatsink rating
SeriesR_θJA = R_θJC + R_θCS + R_θSA

Typical values

ElementR_θ (°C/W)
TO-220 no heatsink60–80
TO-220 small heatsink10–20
TO-220 with fan1–5
TO-247 big extrusion0.5–2
Thermal pad (TIM, 1 mm)0.5–2 (1.5 °C·cm²/W)
Thermal grease, good< 0.1 °C·cm²/W
BGA package5–15 (to PCB)

Heatsink selection

  • Target T_J < 0.8 × T_J,max for reliability (e.g. < 100 °C if rated 125 °C).
  • Sizing: R_θSA_max = (T_J,max − T_A − P · R_θJC − P · R_θCS) / P.
  • Natural convection limits — above ~10 W per TO-220, add forced airflow or larger sink.
  • Mount with TIM — a dry joint is typically 5–10× worse than paste.

Worked example: a 15 W MOSFET

A TO-220 MOSFET dissipates 15 W in a 40 °C enclosure. Its junction-to-case resistance is 1.5 °C/W and the maximum junction temperature is 150 °C, so the 80% target is 120 °C. The allowed rise is 80 °C, giving a total of 80 / 15 = 5.33 °C/W. Subtracting the 1.5 °C/W junction-to-case and 0.5 °C/W pad leaves 3.33 °C/W for the heatsink. With a 3.0 °C/W heatsink, TJ = 40 + 15 × (1.5 + 0.5 + 3.0) = 40 + 75 = 115 °C, comfortably inside the limit. With no sink at about 60 °C/W, the rise would be 900 °C, which is why the part must be mounted on a sink at all.

How to measure it

  • Measure the case temperature, not the junction. Stick a thermocouple to the center of the top of the package or the metal tab, with a drop of thermal epoxy. Then TJ = TC + P × RθJC. Datasheets sometimes list a junction-to-top characterization value (ΨJT) for use with the top temperature of a plastic package.
  • Wait for steady state. A heatsink takes minutes to tens of minutes to stabilize; reading too early understates the temperature.
  • Use the diode method for ICs. A body diode or substrate diode has a forward voltage that falls about 2 mV per °C, so after calibration you can read the junction temperature electrically.
  • Run it in the real enclosure. Heatsink ratings assume free air; a closed case raises the local ambient and can double RθSA.

Notes

  • R_θJA depends heavily on PCB copper area — datasheet values assume specific standards (JEDEC 51).
  • Electrical analogy: temperatures are voltages, power is current, R_θ is resistance, thermal capacitance smooths transients.
Was this article helpful?