The problem could be PBC lay-our , or could be too much flux density , or something else.
The first thing I would try to do is to reduce the flux density (B) in the inductor. Because the peak flux density is defined as Bp= (V*Ton)/(N*A) you may try an inductor with larger cross sectional area (A in the formula) or increase the number of turns (N). Second, review your PCB layout and avoid cross areas of high power with feedback components.

Hello,

I did similar design (5V and 360mA). Unfortunately audible noise is not acceptable. I've tested both coils from AN-37 plus also 1mH from Coilcraft and TOKO without big progress. Moreover I ordered other manufacturers (PANASONIC, SUMIDA ... some Chinese ... 680uH and 1mH .... size from 9.5mm to 11.5m diameter) and they failed.
Design is based on transoptor.
I've excluded PCB configuration (2 layouts were tested) :(
It looks that conception of IC is not OK. We planned to order 100K in 2 years but now I'm looking for buckcup plan like NCP1014.

Regards
Krzysztof

Thanks for Posting your question into forum. Could you please send us the Schematic?Also the procedure of Audibility Noise Test. Are you hearing audible noise at full load or light load?Is it an Isolated design or Non-Isolated design.

Regards,
PI-KSAN

Hello,

Here I attach scheme - this is new 6V PSU but components and problem is the same.
We designed small PSU for home automation plug: wireless communication + relay + some logic (uP). PCB board shape is close to circle with diameter around 29mm (1.6mm, 4 layers). PSU board and mainboard are placed in plastic housing. We didn't do detail noise measurements using microphones etc. because noise (frequency around 2-3kHz for 680uH coil) can be heard from couple meters which can be accepted for noisy rooms but at night this kills.
Even if You remove PCB You have the same situation.
In "low power mode" aplication uses around 50mA, when relay is on: 140mA-150mA, when communicates even 320-340mA. Because of that we also observe modulation of noise frequency.
Answering on Your question: for no load or very light loads (up to 30-40mA) we almost don't hear anything. Noise appears for bigger load.

For now we're going to test old coils on new layout and final plastic housing. Also SMD coil is considered to be tested. For now we don't have more ideas how to solve that issue with this IC.

We're also interested in cooperation with PI to improve that design (I have a lot of measurement data but I cannot send it via open forum).
Moreover we're able to deliver right away 1 PCB board (rev.02 with some probelm with EMI - filter coil to close to main coil) or in 2 weeks newer version prepared for certication if You would like to evaluate it on Your side.

As mentioned - we have big customer with large order QTY starts from april so we feel time pressure.

Regards
Krzysztof

Thanks for Posting your question into forum. The Inductors that you were using are varnished? Generally when the inductors or transformers are not varnished , you will hear a lot of audible noise. Make sure the inductors that are using Varnished. I see that you are seeing a 2Khz-2kHz Audible Noise, Could you please send us the Primary Switching Waveforms i.e., Primary VDS and Drain Switch Current. If the switching frequency pattern is around 15KhZ you can expect audible Noise. for the Load conditons that you are using you shouldn't see lot of Audible Noise. On What basis you selected the Inductance Value? Also Could you please send us the Inductor/Transformer Design SpreadSheet. Please download our PI Expert and Enter your Specifications, Use the Part that you are currently designed for and Run the Spreadsheet on our PI Expert. Please make sure your Inductance value is matching as Per Spreadsheet. We are Happy and More than welcome to help you out.

Regards,
PI-KSAN

Hi,

We've tested Coilcraft RFB0810-681L (also many others) alone and fully varnished using special transformer varnish - no effects. We also covered coil using proper coating (no resin) - no effect.
Coil assembled on 30mm wires also gives audible noise but smaller - coil legs moves vibrations to PCB.
Selection bases on AN-37 and EMI measurement results. In application note it's written to do not use coils <680uH.
1mH is a trade off between EMI, noise and heat dissipation.
I will try to test PI Expert.
On hand I have some screen shots of coil current.

Regards
Krzysztof

What are you using for your reference zener D3? I've got almost identical problem and I've become very suspicious of the reference. See my other post about a question about DI-124. A lot of zeners will not be in spec with a resistor value like the ones given in AN-37 (not enough Iz to get them fully turned on). Even if they do get enough current I wonder if they don't switch softly and cause the noise when they do. I'm going to try with one of these as soon as I get the parts:

http://www.mouser.com/ProductDetail/Diodes-Incorporated/AZ431AN-ATRE1/?qs=sGAEpiMZZMtOXy69nW9rM4Q2kDhi2o9qbK2IRnddm0FvfNilZ59p3w%3d%3d

http://www.mouser.com/ProductDetail/Vishay-Semiconductors/MMSZ4686-E3-08/?qs=sGAEpiMZZMtQ8nqTKtFS%2fGqv07jO8JlbW1g%2fqSMrWow%3d

Hi,

Well I was using standard 2% diode from NXP TDZ5V6J. Diode is additionally biased to maintain proper voltage (1.2K ... tolerance analaysis will be done sone to prove that value). I would use TL431 or other precise regulator but I really don't have a space for it.

Regards
Krzysztof

Hi

I made calculations with PI expert. It suggest to use coil above 82uH (absolute minimum) - default proposal is 470uH.
On the other hand in AN-37 it's written to do not go below 680uH because of some transient situations like start-up.
Moreover I made some tests with 470uH from WE but its also noisy.

The best results I have with SMD coils like 744 771 468 1 from WE but I have no conformation if they are 400VDC tested.

Regards
Krzysztof

Well looking at Table A1 of AN-37, it seems that larger L drives the frequency down, perhaps
towards the 16 kHz that someone (I think a PI employee) said would always produce
audible noise. Does PI expert give numbers for FSavg or FStype for the different
inductor proposals?

Hi,

There are 2 source of noise:
- winding (can be solved by varnishing)
- magnetostriction (can be solved by lowering max B excursion)

Noise can be worse if You hit mechanical resonance of coil or system or both.

Indeed, increasing inductor causes decreasing FSavg proportionally (L is in denominator). But if You look into my screen shoots You will see that FSavg is beetwen 2KHz and 5KHz - very nasty range for human. I don't know how I should get 16KHz - how this value was calculated or estimated?

I mentioned about NCP1014. We've tested this IC. It works in CCM - silent is perfect (current excursion from around 190mA up to 400mA). I used the same "noisy" coils like for LNK306D. The disadvantage of that IC is that You will not get as much current as from LNK306D so I would prefere stay with actual IC.

I ordered special coils, fully varnished from local supplier - I will inform on this forum if it's better.

Regards
Krzysztof

Hi Thanks for Posting reply back into forum. First Please run PI Expert to make sure your inductance is matching as per the PI Exper SpreadSheet, I looked at your current waveforms Full Load. It doesn't look like normal typical drain current, the switching frequency is in audible frequency range. Please run the PI XLX Transformer Spreadsheet, it's an Very Optimised Design Software where we kept lot of effort in it. Please use the inductance as per the spreadsheet and run the audible noise test , Capture the Switching Current Waveforms.

Regards,
PI-KSAN

Hi,

As I wrote: absolute min calculated inductance (usig PI expert) is 82uH. Taking some margin I would choose 220uH (PI expert proposed 470uH). On the other hand in AN-37 it's written to do not use coils below 680uH. Despite this I have tested two coils 470uH (one TH and one SMD). Both are making noise (SMD is much quiet).

Do You have any research data that why 680uH was choosen as a min limit?

Current waveform - this is coil waveform. Documentation about this IC is very limited from PI side. There is no detailed information (for example in AN) about MDCM. When IC knows to go into CCM or rather MDCM? What waveform I should expect at full load?

Ps. I will try to test PI XLX Transformer Spreadsheet.

Regards
Krzysztof

I tried with a better reference (diodes inc. AZ431-A) that has a sharp knee and lists switch mode supplies as an intended application. Unfortunately, this had no effect on the noise. It did still produce a correct output voltage.
Note that the Rz value range specified in AN-37 will result in currents that are too low for some zeners to turn on fully, so its another way of avoiding that problem (other than just using a smaller resistor than specified in AN-37).

Thanks for posting your comment and suggestion. Could you please send us your schematic and at what load your are hearing audible noise?

Regards,
PI-KSAN

I have the same problem than other colleagues regarding audible noise.

We are using LNK305 with direct feedback to regulate a 5V@150mA output. We basically use what indicate with a SMD coil 680uH Irms0.28A, attached file.

At light loads, 10mA, we measure around 2kHz swiching pulses with the expected jitter, but as this frequency is in the worst band for audible noise this produce an unnaceptable ringing. I wonder if the spreadsheet data FSmin of 62kHz is only for CCM, because in high current loads, 100mA, the frequency only increase to 22kHz - 35kHz, never at 60kHz, but with a more audible white noise because of the jitter.

How could we improve this results. Is it only a matter of a good inductor choice?

I would like to know if Krzysztof could solve this issue, and how?

Hi Jose,

In my oppinion the only thing that You can do is oversize coil. I was not able to solve that issue in my application. I don't have enough spacce to place bigger coil. I found out that SMD shielded coils behave better than leaded. Moreover coil manufacturers like WE are not interested with developing special coils so I quit this solution.

I moved to NCP1014 from OnSemi. Application is almost identical to LNK306 except extra small elcap. Now PSU is totally quiet.

Currenty we're testing this soultion but right now I can give some pluses and minuses.

LNK306:

- very robust (none of almost 200 samples didn't fail; tested in real life and lab) [+]

- good EMI, large safety margin [+]

- good overload behavior [+]

- audible noise [-]

NCP1014:

- two different operating frequencies (65kHz and 100kHz), You can play with coil size, Pout etc. [+]

- totally quiet (continuous current mode) [+]

- lower output current (we managed to get 0.5A for 100kHz and 0.32A for 65kHz) [-]

- worse EMI especially for 100kHz (tested on prototype, we're waiting for final PCBs) [-]

- overload capabilities and robustnees are under testing [?]

Regards

Krzysztof

... they are very appreciated. I will test with oversized coils with larger inductance of 1mH, since the performance of this LNK305 in MDCM is completly in the audio range. It seems that the Isat must be double the expected peak current in order to minimize the noise....

Hello Mr. Krzysztof,

My name is Michał Stelmach and I'm Field Application Engineer in Gamma company (one of PI official distributors).

I found You issue described in this forum. You can always ask for technical help directly, using my email address (michal.stelmach@gamma.pl). I'm also in contact with PI engineering support, so I'm quite sure we can help You in case of any new design.

Best regards,

Michał

Thanks Michal for the Help.

Hello,

I'm having the same problem too regarding audible noise. I'm using LNK305 with direct feedback to regulate a 5V@150mA output. In other design I'm using LNK616 for a power supply 5V@1.2A. In both cases, despite I followed the recommended design, I'm facing audible noise issues.

It's a shame that after look carefully the posts at the forum, I saw many people suffering the same problem and I haven't seen anyone posting back telling the problem was solved. Not even after PI guys help.

Besides, I'm facing low efficiency issues that are still unsolved.

I'm planning to launch a product for consumer market next year, with a volume of at least 10k pcs/month just for starting. It's a pitty this LNK ICs doesn't seem to behave like it's expected and I haven't had much support to solve the problems. I'm seriously thinking of redesign the power supply with other IC from other manufacturer like Krzysztof did.

If there is someone that has succeded in design a low power supply with good efficiency (more than 60%) and no audible noise problem, please, get in touch with me!

I'm still open to perform tests if it's required by support team, otherwise, sadly I'll definetly quit Power Integration and perform new designs from other manufacturers.

Kind regards

I see that you design the powersuppy, what's your topology , Audible noise will be reduced by varnishing the Transformers. We made detailed Application note on how to reduce audible noise , it's there in the web. Howver, Attached the link below :

https://ac-dc.power.com/sites/default/files/product-docs/an24.pdf

If that doesn't help, let us know.

I see that this application note could help reducing audible noise in cases that transformers are used, such as LNK616 power supplies.

However in designs transformerless like LNK305 with direct feedback, this solution couldn't be applied. So, what's your solution in such cases?

Marcelo ,

At what Load and line input , you are seeing the Audible Noise Issues.  Please look at the Switching Activity of the part , if there is an Pulse Bunching in the Switching waveforms, the audible noise could be caused by the lay out which introducing noise into the FB pin of our IC . If the Switching activity is normal , however you still hear audible noise, reduce the Inductor Flux  and also make sure the inductors that used are Varnished off the shelf inductors. What kind of Inductor core is used in your design?

I'm hearing noises from 10% to 100% load on 120/240VAC input.

To make it clear, I'm using the inductor as specified. The PN is RL622-681K-RC, the core material is ferrite.

In a practical way, how I supposed to reduce inductor flux? 

Could you please send us the Switching waveform of your design at 120 VAC , 100% Load. ? Also could you please send us your Inductor design PI Expert Spreadsheet?