I have a PCB with multiple power ICs mounted on it. The board is thermally connected to an aluminum plate acting as a heatsink on the backside. I’m trying to understand how much of the total heat is dissipated from the top side of the PCB into the air versus how much is conducted through the board into the aluminum plate. I’m looking for a way to estimate or measure this for the whole PCB, not just individual components.
1 Like
To estimate how much heat is being conducted into the heatsink versus dissipated from the top of the PCB, you can thermally isolate the aluminum plate so that only the top side of the PCB is exposed to the environment. In this setup, the heat that transfers into the heatsink won’t be able to escape, allowing you to measure how much energy is being absorbed by it alone.
If you know the mass and specific heat capacity of the aluminum plate, and you measure the temperature rise over a known period (using a thermocouple, for example), you can calculate the amount of power being conducted into it. For better accuracy, you might want to use a more massive heatsink so the temperature change is slower and easier to measure. Just make sure the heatsink is not dissipating heat during this test—ideally, it should only absorb heat from the PCB
2 Likes
Just to add a useful baseline for the thermal analysis: the total heat generated by the PCB is equal to the input electrical power minus any electrical power output, whatever isn’t delivered to downstream loads ends up as heat.
While this doesn’t directly tell you how much heat escapes through the top vs. the heatsink, it gives you the total amount of heat that needs to be dissipated. That can help when cross-checking the results of any measurement or estimation method you use for heat path distribution.
2 Likes
Building on what Nikh mentioned, directly measuring how much heat goes to the top vs. the heatsink can be quite challenging. But what you can do is experimentally observe how the system behaves thermally.
For example, try measuring the temperature of the heatsink or the PCB while varying the input power. Over time, this can help you establish a relationship between power dissipation and temperature rise.
If you can’t vary the actual circuit power easily, another approach is to bolt a known metal-cased resistor to the heatsink, dissipate a controlled amount of power through it, and observe the resulting temperature rise. That can give you insight into how efficiently the heatsink conducts and stores heat, helping you estimate the relative heat flow. Hope this helps.
2 Likes
It’s worth noting that standard FR-4 PCB material has relatively low thermal conductivity. This often leads to significant temperature variations across the board, especially with power components. While internal copper planes can help spread heat somewhat, in many designs you’ll still see considerable temperature gradients — especially in power-dense areas. This non-uniformity can affect measurements and complicate efforts to pinpoint exactly how much heat is flowing to the heatsink versus being radiated or convected from the top side.
2 Likes
Here’s another idea to consider, you could try mounting the PCB inside a small insulated enclosure (like a styrofoam cooler), with the aluminum heatsink exposed on the outside and the board itself enclosed inside.
If you seal the container and power the board, any heat not conducted into the heatsink will stay inside and raise the internal air temperature. By monitoring the air temperature rise over time and knowing the volume and specific heat of the air, you can estimate how much heat is being dissipated into the enclosure from the top side of the PCB.
It’s a bit of a DIY setup, but it could help estimate the convective contribution compared to what’s going through the heatsink.
2 Likes
Building on Fen’s approach: once you’ve monitored the temperature rise in the sealed chamber caused by the powered PCB, you can use a known power resistor to calibrate the setup.
Place the resistor inside the same container and power it from a DC supply. Adjust the voltage so that the temperature rise over time matches the same profile you saw with the PCB. This profile typically follows an exponential curve like:
T(t) ≈ A × (1 - e^(-t/τ))
where A is the final temperature rise, τ is the thermal time constant, t is time, and e is Euler’s number.
The power you need to dissipate in the resistor (P = V²/R) to reproduce the same thermal response gives you a good estimate of the heat the PCB was dumping into the container — i.e., the amount not conducted away by the heatsink. This method helps you put a number to the top-side thermal dissipation. Hope this helps!
2 Likes
Wow thanks I have to do a better job at coming here to check the replies.