A recent hand-built prototype turned up on my desk with "low voltage rails" tagged on it. It uses a 1.8V and 3.3V buck regulator formed by a pair of TPS6204X ICs. The rails were at 1.6V and 3.0V, respectively...odd, especially considering these are both fixed voltage devices. A quick FLIR image of the board showed some rapid heating of the two switcher ICs and local bypass tantalum capacitors. On further investigation, I found both SMD inductors were each turned 90 degrees on their pads, effectively sending the SW node straight into the bypass cap. What I found curious and quite impressive is that the devices held the rail voltage at a safe level in spite of this failure.
What are some unexpected but "happy" failures you've experienced?