Transformer and coil audible noise suppression techniques 2

[homoly]

Hello,

I would like to get your opinion about the practical application of the different techniques to suppress the audible noise in coil and transformer design. I am interested mainly in coils and transformer types that can be found in different topologies of switching power supplies.

First I have found some theoretical solutions related to fixing method and housing. Can you give me advices which materials would be most suitable for anti vibration pads able to suppress the mechanical vibration of frequency of twice of mains created by magnetostriction of the core? In the case the enclosure will have to be designed what could be the most suitable materials?

I have found some additional techniques ( I call them core related ). First to reduce the magnetic flux in the core and use core material with more narrow hysteresis - this is pretty straight forward, but do somebody have some information how core material types and core shapes affect the audible noise concretely?

Then I have found out the varnish ( potting compound ) type should have effect on the amount of produced audible noise. Can anybody tell which parameter of the varnish could be observed, so I could be able to find out the softer varnishes able better to absorb mechanical vibrations?

Audible noise can be produced also by interaction of the wires in the winding via Amper force, but this is usually swamped by a noise of the core. Can anybody tell when this noise source can become a major issue ?

And last what I call a signal drive techniques. Firstly I have found that dithering the PWM signal around carrier frequency should reduce the noise. Can anybody tell me what could be the optimal dithering interval in % of the carrier frequency?

The second driver technique is to reduce the slope of the PWM pulses, but I suppose the reduction rate is subject for practical testing rather than theoretical analysis.

Thank you for the answers and hints.

Gabriel

 

[MacGyverS2000]

First, let me say I've looked at all of the questions you've asked so far on this board and it's quite a unique and interesting mix... what do you do for a job?

You have to be careful with the chosen potting compound. I've seen/heard of a number of designs where the potting compound did not allow enough flexure of the coils (typically talking about noise-reducing inductors or ferrite beads), causing a serious reduction of calculated specs. Something to be aware of.

I don't believe there is an optimal % of dithering without some form of qualification to the statement. The higher the dithering amount the lower the noise, at the expense of a wider transmitted bandwidth. You'll have to decide which trade-off is the right one.

Reducing the slope of the PWM pulses is simply another way of saying "reduction of slew rate". Reducing the slew rate reduces the transmitted bandwidth and reduces overshoot/undershoot.

As far as audible noise, well, that's personal preference. If I'm running a computer that gives off a loud hum all day long, I'll trade it in for another. If that same supply is used on a construction site, I doubt anyone would notice.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[ScottyUK]

I know there is probably a good reason but why are you switching in the audible band? Unless this is a _really_ big supply you should be able to switch way above the audible frequency range.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[homoly]

Thank you for your response and hints , macgyver2000. I am working as an UPS engineer, mainly focused on design validation testing and previously I was dealing with similar proffesional area as power supplies and inverters.

According noise from your PC I believe the major problem is the power supply noise. If this would became a problem ( lets say for some new colleague I would suggest changing the power supply to silent type or fixing the ordinary supply to chassis via some rubber pads under the washers of the screws fixing the supply to the chassis to avoid amplifying the noise by propagation of mechanical vibrations via the case of the computer. Putting the case of the computer on some anti vibration soft pad thus isolating it from solid floor should be also benefit. And if PC is not in warranty anymore, you could try to connect capacitor between base and emitter of the fan PWM transistor of the power supply in order to round up the PWM pulses, which will reduce the noise of the brushless DC fan.

ScottyUK: the low frequency humming ( 2 times the mains frequency ) is one type of audible noise caused by magnetostriction of the core, but in the case of wrongly designed or manufactured coil or transformer the audible noise can appear even if you are switching on few kiloHertz frequency. Some of these techniques are ment to be used in emergency cases when wrongly designed or manufactured coils are delivered and you have to suppress the audible noise to acceptable level. ( last time I was observing audible noise in line of inverter section switching on 7 kHz )

 

[melone]

Can you replace the "noisy" coils with multiple smaller coils, that resonate at different frequencies? I believe, I've seen that approach work successfully.

 

[homoly]

Thank you melon for the entry. I have inserted this thread to the forum to discuss about magnetic field generated audible noise in general, so this is also a good advice, when considering the redesign of the device.

Just to be more specific, in the case of feromagnetic core this is how I see it, similar to the solution of reducing the magnetic flux in the core ( i suppose the cores of the smaller coils are from the same material as for the big one ). So connecting smaller coils with less turns in series creates less audible noise on the partial coils and less overall noise, too.

 

[GonzaloEE]

I'd sink the entire coil in silicone. This may add some mechanical damping to vibrations.
I agree on that of working over 20Khz, to avoid baseband noise.

You may find some papers on Inet about advanced PWM techniques, in order to suppress audible intermodulation products and sub-harmonics, through some special PWM patterns.

PS: The PS of My PC is whistling all the time (high pitched noise), even in sleep mode :S, any ideas?

Good Luck!
Gonzalo

 

[homoly]

To GonzaloEE:

Thank you for the hint, could you advice some concrete link dealing with those advanced PWM techniques?

There are some special silencer kits for AT power supplies. The text is in slovak but clicking to the concrete item you will get the link to the manufacturer on the bottom of the page ( mainly sharkoon and primecooler ).

http://www.tntrade.sk/shop/index.php/viwsgrp/pc-tuning-odhlucnenie-zdroje

Additionally you can find some solution to fans and HDDs as well :

http://www.tntrade.sk/shop/index.php/viwsgrp/pc-tuning-odhlucnenie-hdd

http://www.tntrade.sk/shop/index.php/viwsgrp/pc-tuning-odhlucnenie-ventilatory

VERY IMPORTANT : If you decide to improve in Do-ti-yourself way with placing rubber washers under the screw ( fixation of PS enclosure to the chassis of the computer ), AVOID direct contact of the screw with the chassis ( the hole for the screw has to be bigger than the screw and the fan has to be mounted only on the rubber washers from both sides ). Otherwise the screw will represents the short circuit for the mechanical vibrations from the fan to the chassis and the efect of the rubber washer will be only marginal.

I will quote myself for one more option - "And if PC is not in warranty anymore, you could try to connect capacitor between base and emitter of the fan PWM transistor of the power supply in order to round up the PWM pulses, which will reduce the noise of the brushless DC fan." - I think this could be apllied also to power supply's fan, but this is for the fan noise not the high pitched noise.

As a last hope you can purchase special so called "silent" power supplies.

Good luck with testing, and if something helped let me know

 

[homoly]

Even if it is not the topic of this thread I have found quite useful article on fan generated audible noise:

http://www.braemacca.com/products/JMC+Fans_Cooling+Info_Acoustic+Noise.php

 

[homoly]

And I have found also the link to the article related to PWM smoothing of DC brushless fans:

ww1.microchip.com/downloads/en/AppNotes/00771b.pdf

 

[GonzaloEE]

Thanks homoly!

I'm doing some resarch on those papers about PWM 'black magic' I fiorst read on my student days. As far as I can remember, they talked about overlapping your signal with some PWM patterns of known frequencies. Of course, the total integrated area must match your desired duty cicle.

Thanks for your useful suggestions for my PS. I guess the solution is stop buying cheap PS from 'certain well-known asian country'

Regards,
Gonzalo