LNK306 Audible Noise Problem

try to reduce your flux density.

check your inductor. I pasting a paragtaph from the datasheet regarding the selection of the inductor

Inductor L1

Choose any standard off-the-shelf inductor that meets the

design requirements. A “drum” or “dog bone” “I” core

inductor is recommended with a single ferrite element due to

its low cost and very low audible noise properties. The typical

inductance value and RMS current rating can be obtained from

the

LinkSwitch-TN design spreadsheet available within the

PI Expert

design suite from Power Integrations. Choose L1

greater than or equal to the typical calculated inductance with

RMS current rating greater than or equal to calculated RMS

inductor current.

Thank you for the reply.

The inductor we are using is one of the Coilcraft inductors suggested in AN-37, so I assume its construction is okay.  PIXIs Designer suggested 470 uH when I ran it - we are using 680 uH.  Current ratings for this inductor appear okay (Irms and Isat).

When you say "try to reduce your flux density" do you mean use a physically larger inductor, use a larger value inductor, or either?

I thought you were using your own assembled inductance. That is why I suggested to reduce BP.

One way to test if the inductor is part of the problem is to put two inductors in parallel. The total inductance will be one half but the current trough each inductor will be one half too. This is just to see if the noise goes away.

Another possible scenario is that the controller is bunching pulses when it is switching on and off.

In other words it switches on for 3 consecutive clock pulses and it turns Off for 3 consecutive clock pulses.

When you have this situation, you will have audible frequencies in the DRAIN waveform spectrum.

Instead it should be skip only one pulse at time

Hope this info is useful to you

I tried placing two inductors in parallel and noise appeared to lessen.  I can decrease the inductance a little bit more.

I have noticed that the inductor sometimes charges up again in the middle of discharging, while the current through the inductor is still around 0.35 A - the operating mode is MDCM since the output current (0.2 A) is less than half the current limit (about 0.5 A minimum for the LNK306).  Do you see any problems with this?

Please post a picture of the situation you are describing.  I can gete a better understanding  with the picture

Here is a picture - I placed a 1 Ohm resistor in series with the inductor and am measuring the voltage across it.

The waveform in pink is the 1 Ohm current sense resistor voltage.

Something is wrong with your feedback.

The controller should not group 2 pulses like in your picture

I was wondering, could I get any advice regarding the feedback of the circuit I am using?  The schematic is illustrated in the attached picture.  I modeled this after the constant current buck-boost topology illustrated in AN-37.  Also I have two feedback pictures for this particular circuit.  One is the voltage across a 1 Ohm resistor I placed in series with the inductor, so it basically is the inductor current.  I also have a picture of the feedback as well, and it is close to the feedback voltage of 1.65 V when the inductor starts charging up again.  The pink is the voltage across the feedback cap, and the blue waveform is an ungrounded connection to one side of the feedback cap.  The hair on the blue waveform indicates that the power transistor has turned on.

Any ideas regarding this?  Any input is appreciated.

I too have had audible noise coming from our circuit. Ours is a low wattage output (can be lightly loaded at times) and when the controller skip cycles, it can do so at a frequency down in the audible range. This caused our ceramic caps to sing like a night at the opera. We had to switch over to either tantalums or electrolytics for our output filters. Not sure what you are using but this worked for us.

For a buck boost converter, you should take the feedback from the +Vout node and couple it through a diode and an RC network to the feedback pin of the LinkSwitch. See figure 1 B in the below link

http://www.powerint.com/sites/default/files/product-docs/an37.pdf

I am using the constant current buck-boost topoloy, as illustrated in Figure 6 of AN-37.  The setup I have is intentional - this circuit delivers constant current to batteries and functions as a battery charger.

You can try to rearagnge the feedback as I suggeted and then see if the audio noise disaperas