Potting some noisy transformers failed to reduce the sound. My question is where the problem might be.
The setup is a row of small (about 0.75x2x4 cm) transformers on a PCB as part of a power circuit. They are oriented so they lie flat and the core is parallel to the PCB. The unit is new.
I removed a plastic wrapper from around the coils, built a dam around each transformer, and filled with low-viscosity potting epoxy. The coils are completely encased by the epoxy but the outer frame is not.
I'm speculating that the coils were too tightly wound for the epoxy to get in, but I'm still surprised there was no noise reduction at all. It seems I misunderstand the mechanics of what is going on.
Edit: This is in a CCFL inverter circuit. I don't know the frequency, but I assume it is typical for this application. There is no sound when the CCFL are at full power, just when the lights are dimmed with PWM.
Best Answer
Grain oriented silicon steel laminations, which are used in building transformer cores, experience magnetostriction when they're subjected to magnetic fields.
That means their dimensions change - when they're subjected to magnetic fields - at a rate corresponding to the frequency of the magnetic field and a magnitude dependent on the intensity of the field.
Affixed rigidly to a structure like a PCB can make the PCB act like a sounding board and transmit the vibrations to the surrounding air and, eventually, to someone's ears.
The cure, ostensibly, is to isolate the transformer from the PCB with resilient "shock mounts" which will damp the oscillations by absorbing them.
-