Thermal and Heatsink Calculator: Junction Temperature, Plates and Pulsed Power
From dissipated power to junction temperature: thermal chain, heatsink selection, chassis as a heatsink, enclosure effects and pulsed power.
Thermal design is a chain of resistances in series, and like every chain it fails at the link nobody checks. The manufacturer specifies the junction-to-case path, the heatsink comes from the catalog, and between them sits the insulator — which can contribute as much as the heatsink itself. Then there are three things the datasheet cannot know: how the assembly is oriented, whether it is enclosed in a case, and whether the power is continuous or pulsed.
This tool brings all of them together. It handles multiple components on the same heatsink, calculates a homemade plate or folded chassis from its actual dimensions, adds the temperature rise from internal to external air when an enclosure is present, and uses the datasheet thermal network for pulsed operation.
Components
One component per row: the power it must dissipate, the junction-to-case thermal resistance of its package, the insulator used for mounting, and the number of identical devices. Rows left at zero are ignored.
| component | P (W) | Rth j-c (K/W) | Tj max (°C) | insulator | qty |
|---|---|---|---|---|---|
Why multiple components on one heatsink are not just a simple sum
The heatsink sees the sum of all dissipated power and settles at one temperature, but each component rises further above it through its own insulator and junction-to-case path. The device dissipating more power, or mounted with a poorer thermal interface, runs hotter: two identical devices on the same heatsink reach the same temperature, while a regulator and a power transistor do not.
The practical rule follows directly: it is better to group low-dissipation devices on a common heatsink and reserve a dedicated heatsink for the device dissipating the most power. Doing the opposite — one very hot device heating the heatsink shared by everything else — can impose the thermal limit on components that would otherwise be fine.
Heatsink
Plate or chassis dimensions, used only in calculated mode. Length is the vertical dimension when the part is upright.
Black anodizing is not cosmetic
Under natural convection, radiation can contribute as much as convection, and sometimes more: in the calculations on this page the two heat-transfer coefficients are often similar. Radiation, however, is proportional to surface emissivity, and polished aluminum has an emissivity of about 0.09 compared with more than 0.8 for a black-anodized or painted surface.
In practice, the same piece left bare loses more than half of its radiative heat-transfer capability, and its thermal resistance can worsen by several tens of percent. With forced air this matters much less because convection dominates. Visible color is irrelevant here: in the infrared, white paint behaves much like black paint.
Ambient and enclosure
If the heatsink is inside a closed enclosure, the air cooling it is not room air but the warmer air inside. The enclosure itself creates the additional temperature rise through its external surface, and that rise must be included.
Temperatures
The temperature chain from ambient air to the hottest junction. Each step is a thermal resistance, so the dominant contribution is immediately visible.
Required heatsink
Junction temperature of each component versus heatsink thermal resistance, with each device limit and your current operating point.
If you build the heatsink yourself
A bent aluminum plate, the enclosure base, or the rear panel can work well as a heatsink, provided it is thick enough to conduct heat to the edges. Below a certain thickness, the area far from the mounting point stays cool and contributes very little.
Plate thermal resistance versus size for three thicknesses. Where the curves separate, the plate has become too thin to use all of its area effectively.
Pulsed power
When power is not continuous, the junction does not have enough time to reach its steady-state temperature. What matters is thermal impedance, which starts near zero and rises toward the steady-state thermal resistance. Manufacturers publish it as a curve and, increasingly, as a table of four resistance-time-constant pairs, which is directly usable for calculation.
Foster network: four resistance and time-constant pairs, as published in the datasheet.
Thermal impedance versus pulse duration for a single pulse and several duty cycles. At long times the curve approaches the steady-state thermal resistance; beyond that point the behavior is effectively continuous-power operation.
Mounting
What can ruin an otherwise correct thermal design
- The thermal compound must be a film, not a layer: it fills microscopic surface irregularities between two surfaces that are already touching. If a bead squeezes out around the package when you tighten it, there is too much, and its thermal conductivity is roughly ten times lower than that of the metal it has displaced.
- A non-flat mounting surface can matter more than choosing the right insulator. A heatsink cut with a saw can leave burrs around the mounting holes; remove them before assembly.
- The mounting pressure matters. A loose screw can double contact thermal resistance; overtightening can deform a plastic TO-220 package and compromise the die attachment.
- Keep the fins vertical, with the channels open at both top and bottom. A heatsink laid flat can lose roughly a third of its performance, and one with its fins facing a panel only two centimeters away no longer behaves like a finned heatsink: it behaves more like a plate.
- If the enclosure is closed and the heatsink is inside, a better heatsink does not solve the problem: the enclosure is the bottleneck. Move the heatsink outside, or make it part of the rear panel.