|
|
|
|
|
|
|
|
|
|
| | | | | 关于RLC电路的重要性,我举个Basso先生采访记录
What is the hardest/trickiest bug you have ever fixed?
I remember a converter that was failing in production at power off. Usually, converters fails in specific conditions like start-up sequence, short-circuit or input overvoltage. With this design, the power supply was quietly destroyed when the operator was unplugging it from the test fixture. I received several boards and indeed, I confirmed that unplugging the converter was killing the power supply. Funny enough, all was intact in the converter, only the controller was affected. At a certain moment, I destroyed the main power MOSFET and replaced it to continue the tests. I was surprised to see that the power destruction no longer happened further to that change. What could be the relationship between the controller destruction and the power MOSFET? It took me a while to observe that a strong negative bias was occurring on the high-voltage pin of the controller. How could a negative voltage appear on a highly positive rail? After several days of investigation, I found that the transformer primary inductance was resonating with the bulk capacitor at power off. Depending on the MOSFET RDS value, the quality factor was more or less important, affecting the ringing below ground. By replacing the 20-A power MOSFET by a 10-A type, I simply contributed to damp the whole circuit, reducing the negative swing at turn off. A simple high-voltage diode in series with the pin fixed the problem.
|
|
|
|
|