/diy/ - Do It Yourself

>>1434226
first, to get a target thermal resistance to ambient, you need to know the power you're dissipating and how much temperature rise you can tolerate. divide °C by W to get °C/W target value
most reputably manufactured heat sinks will include a thermal resistance value directly in the specifications. to get some ideas of size vs. thermal resistance ratings, browse catalogs, but note that shape has an influence too. for example, this one https://www.addicore.com/to-220-heatsink-p/ad285.htm specifies 2W dissipation at 30°C rise. that means its thermal resistance is 15°C/W. this other one https://www.jameco.com/z/507302B00000G-TO-220-Heatsink-With-1-Hole_326596.html shows the thermal resistance as a graph, which looks like about 28°C/W in still, free air. they also include thermal resistance measurements with varying amounts of air flow over the assembly, which is a nice touch
also note that whatever you use to attach the device to the heat sink also resists thermal flux and will have its own °C/W rating. a silicone TO-220 shim might have a thermal resistance of 2.5°C/W. a 0.75mm TO-220 insulator with heat sink grease might have a thermal resistance of 0.1°C/W. direct attachment to the heat sink with heat sink grease causes air gaps which are significant but possibly unpredictable
the component body also has thermal resistance to consider between the body and the chip. see https://www.fairchildsemi.com/application-notes/AN/AN-4166.pdf for the general idea and some characteristic graphs for one manufacturer, and the specific device datasheets for details and exact specs
now, if you have an assembly and know the thermal resistance of all the steps along the path, you can measure the power dissipation (P = I*E) and figure approximate junction temperature, or (tricky) measure heat sink temperature to figure power dissipation