Get rid of audible noise in flyback SMPS

Hi,

Is there some way of get rid of audible noise in a quasi resonant flyback supply?

The noise appears at very low load conditions. It is not a high pitch sound, it's more like a sizzling sound. The controller is LM5023 which has a pulse skip feature. I looked at MOSFET's drain voltage signal. In fact, the controller makes a burst of pulses and then turns off for a while. I think this turn on and off condition is making the sound.

Could be a poor designed loop response capable of doing this?

What more could I look in order to keep the supply quiet over the entire load range?

Thank you,

Yuri

Hi,

This is one disadvantage of burst mode...

Usually burst mode is only with very low power. Can you confirm this?

Possible solutions:
* avoid burst mode by drawing a minimum current/ power
* avoid burst mode by modifying the regulation loop (gain).
* avoid burst mode: many switch mode controllers have selectable burst mode / continous mode
* avoid burst mode with higher value inductance

Find out (*) the device that generates the sound and replace it with another device.
* Inductance: moving windings, moving core. Either replace it or use some glue/silicone to avoid movement
* ceramic capacitors. They transform voltage into mechanical movement by their piezoelectric effect. Replace them with other (low noise) ceramics. Use wired (thm) capacitors, tantalum, electrolytics or foil capacitors.

(*) to locate the sound you could use a plastic pipe (small compressed air hose, fuel feed pipe..). One end next to your ear, the other moving across the pcb.

Klaus

There is also hiccup mode. Could this be the same as burst mode? Hiccup mode is known to produce audible noise, when irregular current pulses go through the inductor, at a frequency lower than normal. I imagine it happening when a strong current burst (or a few of them) cause the output to soar with light load. Then the system must take a rest for a few cycles, to let the output settle back down.

I know that distinctive angry hissing sound.....
Its when the feedback loop goes absolutely crazy outputting a nasty jittery intermittent output, especially with no load on the supply.

The most usual cause is not the feedback loop, but the PWM generator.

If there is a minimum duty cycle at which it just winks off, instead of gracefully falling right down to zero, you can have the situation where one (minimum) flyback pulse is still far too much energy.

The loop then shuts down, and starts back up some indeterminate time later producing one, or more often several output pulses, then shutting down again. This can be very fast and unstable, and sound like a broadband hiss.

A simple fix is to add a bleeder resistor to the output to add just enough load to make the problem go away.

A more complex solution might be to artificially reduce the duty cycle very slightly with a slow on, fast off, circuit. That may require an extra gate driver chip after the control chip. Its a brute force solution.

Last suggestion might be to try a different controller. That one is supposed to go down to zero duty cycle, and it probably does.

But what if it only reduces linearly down to say 5% duty cycle then winks off ?
It may not be capable of creating a 4% or 1% duty cycle.

The guy that wrote the specification is only saying it CAN go right to zero, not that it will have linear control right down to zero.

So check, you might find (for example) that you can reduce the duty cycle manually with a potentiometer down to say 500 nS, then it abruptly shuts off.

Thank you for the answers.

I found also this document, which is very useful for this subject: https://www.fairchildsemi.com/application-notes/AN/AN-4148.pdf

So first I varnished the transformer and nothing changed. Then I started to hear each part over the PCB using a paper tube... It was a little hard to distinguish the source of the noise, but I found out that was a ceramic capacitor making the hiss. I removed it form the PCB and the hiss went to a very low level, which is pretty acceptable. The capacitor was the one that ties both primary and secundary together. I replaced it with a film type, and it makes some hiss too (lower than the ceramic type makes).

Do you guys have a hint about a good replacement for this cap? I was using a ceramic disk type 4n7 2kV.

Yuri