Lessons learned during validation testing part 2 (do's and don'ts)

I feel very fortunate to have gotten to jump right into board design right out of school, which I can't believe was so long ago. The first 6 years of my career was a trial by fire that taught me a lot of great lessons on design practice. To this day I still make recommendations on designs based on what I learned about designing in plenty of operational margin and saving cost wherever you can. Yes those things don't really go together, but the point is I never take for granted the value inherent in both philosophies. It really helps me explain the cost / risk equation when working with other customers on their designs. The second part of my lessons learned blogs is focused on a lesson I learned about temperature testing. (part 1 posted here - Lessons learned during validation testing part 1 (do's and don'ts) )

Under a time crunch to get a product released, I had planned out a tight schedule to perform temperature testing over the weekend on a piece of pole mounted telecom equipment. This thing had to work in Alaska and Arizona inside a sealed box over an extreme temperature environment. I had thermal probes attached to critical components as well as in strategic places around the enclosure. I had a data logger set up to measure and record these temperatures. During the test the equipment would be power cycled to ensure it would startup and shut down properly at all temps, and also run for different periods of time while data traffic was passed and the temperatures were monitored. The longest test involved 16 hours of continuous operation at 65 C at full capacity. Here is a very basic block diagram of the system and an excerpt from the test report I create oh so many years ago

I went into the office periodically over the weekend just to make sure everything was still in sync and working. Come Monday I was happy to see everything still operating as expected. Then a colleague happened to walk by the temperature chamber while I was working and asked how it went. I told him and his response was "did you turn off the chamber?" I thought for a second somewhat confused and told him no. He told me to turn off the chamber power and see what happens. The chamber was still running set to 65 C so I turned off the power with everything still sealed up. Much to my disappointment I very quickly started seeing my readings rise. That 18.8 C worst case rise rather quickly became a 30 C rise. Why? Airflow. Even in a confined space, airflow dramatically lowers the highest temperatures as it constantly moves the hottest air to other locations. I knew that, which is why we had a fan tray inside the sealed product enclosure... What hadn't occurred to me was that the temperature chamber itself was "blowing" hot air inside it and that air movement was altering my test results.

From that point on, my new test procedure included placing my equipment under test inside a cardboard box in the chamber to eliminate airflow across the surface of the board / enclosure. To this day I still ask questions about the test setup when I start talking to someone about thermal testing. There is a lot of truth to the sentiment that the best way to learn something is to do it wrong once.