SM transformer audible noise and structure natural frequency

[Jakoviz]

Hello,
it's my first posting on the site. My problem is not wholly just electronics, but hopefully someone can provide good advice.

I need to examine audible noise coming from a switch mode transformer component. In theory, the noise is created mainly because of the flux excitation affecting the ferrite core, making it vibrate.

I've been given to understand that a high-freq. transformer, such as this one, normally only creates ultrasonic freq. sound, which naturally doesn't cause noise disturbance. However, there are something called 'transient' and 'burst operation', which cause the noise freq. to come down to human hearing range. These two concepts I don't really understand.

I was first thinking of determining the natural frequency of the transformer structure and measuring the sound amplifying effect of resonance. I am wondering if it is specifically the resonance amplifying the sound that leads to disturbing levels of noise.

Any suggestions how to determine the natural frequency of the transformer's structure or even individual parts of the structure (core, bobbing, windings)? Any ideas how to make the resonance less disturbing? Are there feasible ways to raise the natural frequency higher, close to or over 20kHz? Or how to keep the frequency in the ultrasonic range in this 'burst operation'?

 

[Diver300]

I can only offer a tiny bit of advice about switch-mode buck converters.

If the load current on an inverter is small, it has to run at an extremely small duty cycle. There is a lower limit on the on-time of the switch, partly caused by the capacitance of the components. Also, if the inductor current isn't allowed to get large enough during the on-time, then all the inductor energy will go to charging up stay capacitances and there will be no current in the diode, so there is no advantage in having a switch mode converter over a linear one.

The result of this is that the switch operates once, and then there is a long pause with the switch turned off and no current in the inductor, while the output capacitor supports the load. This can often be heard as a series of clicks. It is impossible to get the clicks close enough together to make all the sound above 20 kHz at low load, because the sound comes from the repetition rate of the pulses, not from the character of the pulses.

I have found that running the inductor at lower currents than its rating helps.

I guess that in continuous mode, there won't be any sound below 20 kHz. In that case, at low load you could:-

1) Turn the inverter on and off so that the output voltage varies more. During the on times, it will be running continuously so inaudible, and when off it is probably silent. It does mean that you will get worse voltage variation.

2) Use a smaller switch mode, or a linear regulator for low load conditions.

 

[smanches]

It's the same reason cell phones are heard over your speakers. They're frequencies are obvious out of hearing range. But the frequency at which they multiplex the transmission is right in hearing range. It's the multiplexing you hear not the transmission frequency itself.

Just thought I'd share that.