So a lot of folks trying to manage heat in their high powered, tightly packed electronic devices get overwhelmed by complexity, since their devices can often look like this:
But even if your device is a high density LED, a handheld device, or a hot-running CPU, there are only three steps your thermal analysis can EVER include. Once you master those three, thermal management becomes less like the complex picture above and more like this:
Step 1: Get the heat out of the chip
For our purposes, “the chip” could be whatever dense, heat generating device you’re trying to cool, like an LED or whatever.
Step 2: Get the heat out of the area
Unless your device is all computer chips exposed to air, you always need to move the heat to an area where it can be expelled. Which brings us to step 3:
Step 3: Get the heat out of the device
And that’s it. Even in the most complex devices, that’s all there can be to thermal management. Relaxing, isn’t it?
(And if you were wondering, those 3 clear, simple images were made in SolidWorks Composer, right from the CAD models, including the color and arrows. If you want the same for your easy-to-read manuals and instructions, let us know!)
How do those steps look in action? Creating a Flow Simulation CFD model of a chip on a circuit board like this:
We can get a pretty picture of what the natural convection flow looks like, which is mainly as we expect:
But that pretty picture doesn’t really tell us which of our 3 fundamental steps are working or failing. To do that, I’m going to “Probe” along that vertical gray sketch line that all you readers thought I forgot to hide in the previous two pictures.
Using that sketch line (that I left in the previous pictures ON PURPOSE) to measure temperatures along its length, we get a graph like this:
That’s Celsius temperatures along the length of that vertical sketch line. Not that useful? How about if I put some labels on it:
So, pop quiz hotshots: from the above graph, where is the thermal management problem in this device? Which of our defined steps, 1, 2, or 3? Just from that 2 minute CFD analysis, the answer is staring us in the face.
Let’s work our way there. Notice that the bottom of the board, leftmost of the graph, is at ambient. (Because I set it that way in the analysis.) But notice that the air above the sink (far rightmost) never reaches ambient, because we’ve got this chimney effect of hot air rising up. That’s a nice sanity check, but that’s not our thermal management problem.
Those readers used to working with thermal conductivities will have realized which portion of the graph is the problem, the flat parts or the steep parts. Look at the flat part which represents the aluminum heat sink. What does that mean?
It means that there is relatively little temperature drop along that line, to move all that wattage (in this case around 30 W) to ambient. Little temperature drop to move a lot of heat? That’s the units of thermal resistance, and you want that to be very low. You want it to take very little temperature difference to move a lot heat, which is why folks use aluminum as a conductor. (There are materials in use right now that are 1000x better conductors out there than Al or Copper- we’ll talk about those in a moment.)
So where is the problem? Not where the line is flat, but where it is steep:
Moving heat from the case of the CPU to the heat sink is a very steep line, (a very high thermal resistance). It takes a lot of temperature difference to move that wattage, leading to our high Junction temperature. (As does the other steep line, moving heat from the CPU junction down into the circuit board, in case we care about that.) This is a classic Step 1 thermal management problem: Getting the heat out of the chip. Solve that and we’re most of the way home. See how easy it becomes if you break it up into those 3 steps?
But how do we solve that problem? What are the current technologies out there for more efficiently getting the heat out of small, tightly packed areas and creating better, flatter graphs and lower CPU junction temperatures for your devices?
If you want to learn about that and see the rest of this story play out, check out the recorded webinar below! It compares the effectiveness of the still-exotic thermoelectric coolers and heat pipes to the reliable stand-bys of heat sinks and fans, and even ask if you can’t just use conduction for everything. It also explores how different envelope conditions affect your choice, using SolidWorks Flow Simulation CFD software to illustrate the sometimes surprising results.