[SOLVED] audio whistle from switching supply transformer

Hi. How many of you have encountered the audio whistle comming from transformer & filter inductor of switching power supplies that they have designed. What do you do to remove this audio noise. As an example I think I would use varnish in the windings for that.

I also think that if pwm chip is forced to give clock every cycle and in low duty cycle for less current demand and high duty cycle for high current demand then there is no need to worry about this whistle. It will vanish automatically. However if pwm chip output clock is random then this sound is sure to come especially at low current demand or when there is no load connected.

What you say about it all. what you do to remove it. do you use varnish and/or some tricky electronic circuit for that.

Dear Friend!
Hi
It is normal ! my mean is that voices!
In fact that voices are due to your core! the core of your inductor created these noises! you can optimize it with these two ways:
creation the gap space at your core .
and keep your core with wire ( my mean is that your core will shaking and its is due to the HF fluxes ) . and you should be ensure that your core can not be shaker! and you can use a big varnish for keeping your core!
Best Wishes
Goldsmith

goldsmith said:

creation the gap space at your core .

Click to expand...

does that mean to use gaped cores. but you know we use gaped cores in flyback topology only. for forward converters we don't use gaped cores for transformers (yes we use them in filter inductors). on the other hand if we use gaped cores in forward transformers then their purpose there is to increase the flux swing range without hitting saturation. anyway please clear how gap will reduce audio sounds.

goldsmith said:

and keep your core with wire ( my mean is that your core will shaking and its is due to the HF fluxes ) . and you should be ensure that your core can not be shaker! *and you can use a big varnish *for keeping your core

Click to expand...

is it not true that wound wire vibrates more than core itself. if it is the case then is not it true also that without using varnish, (vibrating) wire insulation will be degraded and may create short circuit within wound wire.

Hey Friend!
Hi
Remember that if the core be saturated , it can not work truly ( and its current waveform will destroy!!!!) ( the current of the inductor or transformer .) to prevent from core saturation , we have to do some things , like : gap space creation , or implement a big core ! and the reasonable way for this aim is gap space creation.
see the switching power supply text books and search this thing on them: "core saturation avoiding".
and about keeper wire , it should not be from conductance material.
Regards
Goldsmith

---------- Post added at 00:48 ---------- Previous post was at 00:46 ----------

and remember , that the DC currents , or mixed DC currents , can shift your core , to the saturation region.

brother goldsmith please think before you write.

brother goldsmith please think before you write.
what do you mean by that ? did you get my mean accurately ? Read some technical book in this area such as Switching power supply design of professor Abraham press man , or switching power supply cookbook of professor marty brown or some famous article in this area , and then say that thing!!! and be polite when you want to say any thing!
Regards
Goldsmith

---------- Post added at 16:54 ---------- Previous post was at 16:45 ----------

read switching power supply design of professor abraham press man about gap space design of a forward converter ( page 88 to 89)

---------- Post added at 17:27 ---------- Previous post was at 16:54 ----------

And what is your reason . when you said , that , forward converters , don't needed gap space?

Most circuits should not make noise under normal circumstances, even without varnished cores and windings. There shouldn't be any power transfer at audible frequencies, unless the switching frequency is <20KHz, or if the load is actually varying in the audible frequency range. Otherwise, there's probably some problem with the circuit which needs to be addressed. Core saturation, current limiting, closed loop instability are all possible explanations.

Dear mtwieg
Hi
When i built a class D audio amplifier at 900W power , my carrier frequency was 300KHZ .and my inductors ( of LPF) was noisy and i optimized that with things that i said at top.
and if the power is not high , the magnetic elements can not produce noises ( in fact their nose at low powers is pretty low).
Best Wishes
Goldsmith

---------- Post added at 09:29 ---------- Previous post was at 09:26 ----------

by the way , if your circuit has hiccup current limiter , that can provide oscillation at low frequencies .

mtwieg said:

Core saturation, current limiting, closed loop instability are all possible explanations.

Click to expand...

goldsmith said:

by the way , if your circuit has hiccup current limiter , that can provide oscillation at low frequencies .

Click to expand...

In reply to both of your comments about current limit I have also observed the audio noise phenomena. In current mode control it happens due to noisy current sense resistor and incorrect RC filter connected across it. In vero board designs we can not avoid it easily. Also I think in chips with leading edge blanking functionality this problem is addressed automatically.

goldsmith: don't irritate with my earlier comment. one leading designing company to whom I told this problem also think that this can be a new research to design converters with forcing the chip to produce clock every cycle to mosfet even how light load is connected to the power supply. thus eliminating audio noise that comes when clock to mosfet is irregular.

---------- Post added at 11:08 ---------- Previous post was at 09:34 ----------

Also remember that irregular clock (pulse skipping) can also be observed at heavy load if current sensing is improper as I told before.

goldsmith said:

Dear mtwieg
Hi
When i built a class D audio amplifier at 900W power , my carrier frequency was 300KHZ .and my inductors ( of LPF) was noisy and i optimized that with things that i said at top.

Click to expand...

Well that's not surprising if the supply is in an audio amp. Like I said, if the load (or line, or really any part of the converter) is being modulated in the audible range, then you'll likely get some noise. So of course that will be the case in an audio amp (though you would hope the magnetic noise is completely dwarfed by the speaker output).

[/COLOR]by the way , if your circuit has hiccup current limiter , that can provide oscillation at low frequencies .

Click to expand...

Yes, and it's also possible to get noise at light load if the controller has burst mode operation, or some other type of hysteresis in the feedback loop.

---------- Post added at 07:55 ---------- Previous post was at 07:49 ----------

ElectroPhysics said:

In reply to both of your comments about current limit I have also observed the audio noise phenomena. In current mode control it happens due to noisy current sense resistor and incorrect RC filter connected across it. In vero board designs we can not avoid it easily. Also I think in chips with leading edge blanking functionality this problem is addressed automatically.

Click to expand...

By leading edge blanking, do you mean that the controller has a set minimum on time? You should be careful of such methods, especially if your converter ever has to operate at light loads, or has a wide line voltage range. In practice I've seen that a proper combination of lowpass filtering on the current sense, and some slope compensation, prevents pulse skipping. The slope compensation in particular is useful because it increases the amplitude of the current sense ramp, while not increasing the switching interference which causes pulse skipping.

What controller IC are you using?

Now I'm in better position to tell the right reason. At light loads the current sensing circuitry measures the low current demand and lowers the duty cycle. In real scenarios this duty is so small that almost diminishing. Thus in next cycle a high primary current flows through mosfet and current sensing circuitry. This produces more than necessary power at output and thus in next cycles the cycle is skipped or duty cycle is lowered too much. One thing to be noted that this primary current when flowing at low output power is still at peak primary current level. Also this current is at random cycles. This high level of random current produces audible noise. The solution is that we can remove audible noise almost if we can keep primary current level to a lower level instead of at peak cutoff level at low output power. Then if possible we can also operate this SMPS in burst mode without danger of audio noise.

mtwieg said:

What controller IC are you using?

Click to expand...

The audio noise problem that I have observed practically in designs is common with ICs like UC384x series and SG352x series. However new chips like NCP1200 have addressed this audio noise problem with some genius internal design.

The audio noise problem that I have observed practically in designs is common with ICs like UC384x series and SG352x series. However new chips like NCP1200 have addressed this audio noise problem with some genius internal design.

Click to expand...

Stable switched mode operation at low load can be achieved with the said controllers, too. Most products on the market succeed in doing so.

Noise coupling at higher currents can produce subharmonic
oscillations in PWM controllers. You would observe this as
a bistable pulse width (never zero, but one of N repeating
widths as the current threshold event is kicked by last-cycle
residual noise..

Pulse skipping can also be bistable but (duh) the pulse jumps
between 2X (3X, 4X...) and zero current-delivery, and it
will be at minimum output current.

A 'scope and some thought would tell you which, or none.

If you went to higher fSW then you would have more
room away from audible, for these subharmonic processes
to do their thing without being noticed.

Now cores also do have natural audio frequencies in
the solid material's various dimensions and it's possible
that kicking the core can ring that up by magnetostriction.
But this ought to be a load-varying amplitude and not
an abruptly appearing tone, and have no evident pulse
width variation behind it.

Thanks guys for your kind help.

Best Regards
ElectroPhysics