Wavy signals - solution for a test problem

Test engineer: “I can’t get the units to pass the parametric test at 50Hz!”

Me: “How many of them?”

Test: “Sometimes maybe about a quarter.”

Me: “’Sometimes?’ What does that mean?”

Test: “Even the failures are inconsistent. When we test the failures again, they sometimes pass. And the good UUTs sometimes fail a second round. They only fail at 50Hz.”

Me: “And with known good units?”

Test: “They have the same inconsistent failures.”

I went down to the test stations at the assembly line. The tester injected a signal, measured the THD (harmonic distortion), compared the value, and returned a pass/fail result. Everything worked as expected, except at 50Hz. The test setup seemed to be functionally correct all around.

There was not enough visibility for me, so I grabbed an oscilloscope to watch the output being measured. Nice clean signals, even at 50Hz. UUT pass. But wait! Halt the test programme there and keep the input at 50Hz. Every 10 seconds or so, the signal started getting very… wavy.

I found that the 50Hz input was clashing with the 50Hz mains supply to the UUT. The two frequencies were not synchronised and were probably about 0.1 Hz different. That accounted for the waviness every 10 seconds. My solution was to use a 55Hz input instead, which still met test requirements and was not affected as much by the 50Hz mains power.