Acoustic noise in power supplies

[HighTechPower]

Hi. Welcome back.
I'm designing power electronics products & projects professionally. The only bottleneck I'm facing is that in almost 90% of designs it's ferrite core transformers generates unwanted loud acoustic noise. I want to know that if I'm the only person facing this issue or others too have this issue and most important how they rectified this problem. Plz also share with me as much links as possible as I'm gonna do a new research on that again. More than a decade ago I written a non-professional article about this issue and is attached herewith. Acoustic noise is ruining my power electronics business so need your suggestions seriously.

 

[KlausST]

Hi,

nowadays switch mode power supplies usually work with a ultrasonic switching frequency. Thus - even if they cause noise - it should be impossible to hear.

So it´s not the switching noise that causes theses audible frequencies, but something else. Maybe mains, maybe subharmonics, maybe the regulation loop.
--> Thus you need to find out the root cause of the frequencies. Maybe with a spectrum analyzer.
Maybe the load causes these frequencies, maybe a low quality IC, an unstable regulation loop, a bad PCB layout...

But what makes these frequencies audible.
You say it´s the transformer.
But also caramics capacitors cause audible noise due to their piezo like behaviour.
--> Thus you need to find the part that causes the noise. Try potting the transformers, or just some glue on the coil. Try different manufacturers. Avoid operation close to saturation.
Use capacitors with different ceramics, different voltage rating, different manufacturer, different mounting method, use millig slots...

Klaus

 

[FvM]

As you have already found out yourself, audible noise of power supplies is mainly related to discontinuous operation, subharmonic oscillations or unstable control circuit. In special cases it's caused by pulsating load, e.g. a GSM transmitter. The former can be avoided by using control circuits that don't fall in discontinuous operation at low load. In the remaining cases, impregnated or even molded transformers may be used.

 

[Easy peasy]

as above - continuously operating pwm above 20kHz, and or limit the modulation at audible frequencies using clever control and energy storage schemes - there exists special glue for magnetics cores + potting in thin then thicker varnish.

 

[HighTechPower]

Another culprit that I missed to mention may be ground bouncing which is causing erratic behavior.

 

[Easy peasy]

how many kW ?

 

[HighTechPower]

how many kW ?

I designed mostly 25W to 500W both DC-DC & AC-DC in different topologies.

 

[KlausST]

Hi

I missed to mention may be ground bouncing

this is why I mentioned the PCB layout.

Klaus

 

[byjuncal]

Hi Klauss and everyone, (this is my first message here),

Indeed, it is really trickie to design a completely quiet SMPS if the operation conditions are very differents.

I have some experience facing audible noises problems. Many people, directly point to the magnetics as the principal source for the audible noise. However, capacitors are another frequenly culprits, and not only ceramics but film too.

In my experience, the more common reason for the audible noise is the burst or discontinuos switching at lights loads. There are ways to deal with it. Control loop stability is another common reason that in fact happen usually at light loads but can generate audible noise at many different conditions. A bad layout at the end can results in a instable control too.
In my case, generally, the controller used or the control strategy solve practically all the problems.
If you want to describe more in detail some of the issues you are facing, maybe we can give you more details of how to deal with them.

Byjuncal

 

[HighTechPower]

In my case, generally, the controller used or the control strategy solve practically all the problems.
If you want to describe more in detail some of the issues you are facing, maybe we can give you more details of how to deal with them.

Byjuncal

Well bro first of all I let you know that in some of the designs the acoustic noise was present at random or at all load conditions instead of just at light load. One thing is sure that random switching of MOSFET with on & of band pattern in audible range was the root cause.
I had an old idea of my to provide a forced minimum pulse OR with actual gate pulse. The end result that though now there was no cylce skipping so no random pulse and acoustic noise reduced but it was still there as other operating conditions including PCB layout, control loop etc were still the same.

--- Updated Feb 25, 2022 ---

In other words now duty ratio at a particular load was vibrating back & forth about it's mean position when observed on oscilloscope and was the root cause of acoustic noise in this scenario though now there was no pulse skipping.

--- Updated Feb 25, 2022 ---

Here OR mean analog version of OR gate.