Facts every engineer should know about thermal management

By Warren Miller, contributing writer

Thermal management can be a bewildering topic even though we deal with it every day. What temperature do you want to set the air conditioner for in the car? Where should you direct the air flow? These are the basic questions that thermal management, in just about any application, deals with. When the temperature is set by the electronics on your printed circuit board and the air flow is the only fan that your budget can afford, things can get trickier. Let’s take a quick overview on the important thermal management topics every engineer should know.

Let’s start with a good definition of thermal management. It’s all about transferring heat from one place to another — managing the flow of heat, usually, from where you don't want it to where you do want it. In electronics systems, you almost always want to move heat from your device to someplace else, away from the device. Easy, right? You might remember reading that some of the most powerful computers of their day had more patents in thermal management than in electronics (like the Cray-1).

Basically, there are two ways to move heat in most electronic devices, and they both use convection, but they can do this either actively or passively. “Convection” simply means the movement of heat via a working fluid. This fluid, confusingly, is most often just air, but it can also be a substance such as water or even, also confusingly, a gas — perhaps a halocarbon like Freon. Now, usually the working fluid is heated at the place where you want to remove heat and then transferred to a location where you can dissipate the heat, often called a radiator. Think of the radiator in a car where the liquid is put in contact with a large number of metal fins that can transfer the heat into the air, cooling the liquid so it can be re-circulated back into the engine. If you need to use a pump or a fan, this is known as active convection. Passive convection takes advantage that a heated working liquid wants to rise, and this can move the heat away from the heat source naturally if there’s a convenient escape route for it. You can take a similar approach, using air as the working liquid, to cooling a high-power CPU. Some high-end gaming computers even use water as the working liquid and could, potentially, use it to heat the hot tub, too.

In the radiator of a car, the liquid is put in contact with a large number of metal fins that can transfer the heat into the air, cooling the liquid so it can be re-circulated back into the engine. Image source: Pixabay.

Passive convection can be a very efficient and cost-effective solution in many applications. If you don't need a pump or a fan, you can eliminate moving parts that can reduce a system’s operating lifetime significantly. The most common passive approach to thermal management is to use heat sinks. These metal “wings” attach directly to the heat source, usually a high-powered CPU or FPGA, to move the heat away from the device and into the surrounding air. If the heated air can flow out of the enclosure and cooler air can replace it, you end up with a very efficient method for getting heat out of your system. Placing high-powered components correctly can also help enhance cooling efficiency. Make sure air can flow to or from hot components that aren’t obstructed by taller components or connectors. Also, remember that air flow near an enclosure edge will be slower than in the middle, so it might be best to place high-power components in the middle of your board.

Now, this is interesting, but you may be asking why this is important. If you don't cool electronic devices, they can get very hot, hot enough to burn you or start a fire (or both). You definitely don't want this to happen on an airplane, right? Perhaps a bit less importantly, electronic devices that run too hot have shorter life spans.

So if you are asked to implement thermal management in your next project, no need to despair — you just need to ask some key questions. First and foremost, what is the target for the operating temperature of your system and what is the environment it will be operating in? You also need to know what kind of enclosure (if any) it will be in, and what escape routes are available to dissipate (transfer) heat. If you ask these questions, you’ll be well on your way to a good definition of your thermal management problem, which is always a good step in any engineering project.