Working as an electrical engineer, I have often been faced with thermal design problems. They are usually in the form of: “Given a maximum system temperature X, ensure that the maximum temperature of all devices in the design does not exceed Y.”  Temperature X is usually a customer spec, while temperature Y is almost always driven by MTTF constraints on the semiconductors used in the design.  This sounds simple enough until you realize that:

• The system temperature is often not clearly defined.  Is it the ambient air temperature?  The temperature of the printed circuit board the part is mounted on?  The temperature of a baseplate (usually a sizable piece of aluminum) that won’t even exist in the finished design?
• The MTTF spec is usually based on things you can’t measure directly (at least not easily or accurately), like the junction temperature of a transistor that is on a die you can’t even see, inside a package you don’t have a thermal model for.  In addition, MTTF numbers are often wildly inaccurate, don’t account for duty cycle, etc.  Yikes.
• Chances are, the guy who worked on the part before you hasn’t checked the thermal readings in years, so it’s actually running way over the limits, and now you have to fix it.  A small change in the design can drastically affect the numbers

At the end of the day, numbers are scribbled on envelopes or entered into spreadsheets, guesses estimates are made, and everyone resolves to develop a better thermal model next time, which of course never happens.

I stumbled upon this book at the Stanford University Bookstore a few months ago.  Given my experience (frustration) with thermal design, I couldn’t help but pick it up and start reading.

It’s a fairly quick read and thoroughly entertaining.  Kordyban’s style is very informal.  The chapters are in the form of several short stories about fictional characters at a made-up company called TeleLeap.  These characters have to solve a series of design problems, which are used as examples to explain several concepts of thermal design.  There are no lengthy derivations and the technical discussions are pretty understandable, even to someone who never took thermodynamics in college (like me).

In particular, I found Kordyban’s discussions of the errors that can creep into thermocouple measurements, the difficulty of measuring junction temperature directly, and the problem with pin-fin heatsinks very interesting and educational.  I won’t say that I am an expert in thermal design having read this book, but I do have just a little bit more insight into what’s going on (and what to avoid).

Unfortunately, the book is out of print, so you’ll have to find a used copy and pay some big bucks – unless you get lucky like I did and find one that’s still sitting on the shelf.  No, you can’t have mine!

Hot Air Rises and Heat Sinks: Everything You Know About Cooling Electronics Is Wrong