Thermal Design Basics: Will This Part Overheat?
Heat is the most common cause of premature electronic failure, and most thermal problems are predictable at the schematic stage. Every power device has a maximum junction temperature and a thermal resistance to its surroundings. Multiply the power it dissipates by that resistance and you have the temperature rise. This article shows the calculation and what to do when the answer is too hot.
At a glance
- Tj
- Junction temperature, the temperature of the silicon itself. Datasheets specify its maximum.
- °C/W
- Thermal resistance: degrees of temperature rise for each watt dissipated.
- 0.3 W/m·K
- Approximate thermal conductivity of FR-4 laminate, a poor conductor.
- ~390 W/m·K
- Approximate thermal conductivity of copper, over a thousand times that of FR-4.
The problem
Linear regulators are the usual offender. They dissipate the full voltage drop multiplied by the load current as heat, and a small package that looks adequate on the schematic can exceed its limit at full load on a warm day.
Datasheet thermal resistance figures are measured on a specific test board. A real layout with less copper around the part will run hotter than the figure suggests, sometimes considerably.
Calculate the temperature at the worst case: highest ambient, highest load, least airflow. That is the condition that causes failures.
A worked example
A linear regulator drops 5 V to 3.3 V at 300 mA. It dissipates (5 - 3.3) x 0.3 = 0.51 W. If its junction-to-ambient thermal resistance on your board is 50 °C/W, the junction runs 0.51 x 50 = 25.5 °C above ambient.
At an ambient of 60 °C inside the enclosure, that is 85.5 °C, comfortably below a 125 °C limit. Raise the input to 12 V and the dissipation becomes 2.61 W, a rise of 130 °C: the part cannot survive, and a switching regulator is needed instead.
What good looks like
Calculate the power dissipated by each regulator, driver and power device at maximum load.
Multiply by the junction-to-ambient thermal resistance and add the highest ambient temperature inside the enclosure.
Where the result is too close to the limit, reduce the dissipation or improve the thermal path with copper area, thermal vias or a heatsink.
| Input voltage | Power dissipated | Temperature rise |
|---|---|---|
| 5 V | 0.51 W | 25.5 °C |
| 9 V | 1.71 W | 85.5 °C |
| 12 V | 2.61 W | 130.5 °C |
| 24 V | 6.21 W | 310.5 °C |
| W/m·K | |
|---|---|
| Air | 0.026 |
| FR-4 | 0.3 |
| Solder (SAC305) | 58 |
| Aluminium | 205 |
| Copper | 390 |
What it changes
Reliability
Parts running well below their limits last far longer.
No Surprises
Thermal problems are found by arithmetic rather than by a failed prototype.
Better Architecture
The calculation shows early when a linear regulator must become a switching one.
Efficient Layout
Copper is added where it carries heat, not everywhere.
Where to go next
Use the exposed pad on power packages. Solder it to a copper area tied through several thermal vias to an internal plane.
Measure temperatures on the prototype in its enclosure, under full load, and compare them with the calculation.
Problems this prevents
- A linear regulator dropping too much voltage
- Datasheet thermal figures assumed on a small layout
- Temperatures calculated at room ambient only
- An exposed pad left unconnected to copper
Discuss your project
Email support@aavico.com or use Get in Touch when the website finishes loading.