Solution Finder Get Tech Support

Audible noise on LNK320x series

Posted by: saiknaram on 07/22/2022

Hello PI Engineers, I need your help with an audible noise problem. I have been struggling to first reduce the 12V ripple (now good) and second reduce the sharp persisting audible noise at various frequencies due to different inductors being used. Please reference the attached PNG file for schematic and Excel file for the loading current values per program. 

I am trying to tune L201 and U200. 

Loading scenarios:

1) ATTiny microcontroller with 3 LEDs and 5 buttons on 3.3V and relay on 12V. Total hitting close to 42 mA on 12V at most. - This is the X1 program in the excel sheet

2) ESP32 based microcontroller with BT and WiFi on 3.3V and with RGBW LED and relay on 12V. Total at around 185 mA on 12V. - This is the X2 program in the excel sheet

For scenario 2, I have attached 3 pictures at the different loading levels due to the ESP32 micro at power on, standby and WiFi pairing. 

I initially started with an LNK3206 and used 680 uH, 820 uH, 1 mH and 1.8 mH as trials. All of them are noisy and change tone as the inductor is changed. This was over 5 dB of noise relative to ambient with each inductor change. And that can be head at 6 inches from the ear.  

Then I realized through trial and error that using an LNK3202 + 1.8 mH inductor, loading scenario 1 is resolved. But it did not help in scenario 2 as a high current mode during Wifi pairing will cause the regulator to reset.

However, the LNK3202 chip is almost silent compared to the LNK3206 and LNK3205 versions as well. What is the reason? I refuse to believe that I need a varnished inductor when the LNK3202 option is almost working with a drum core inductor. Should I pay attention to the freewheeling diode along with the feedback diode?

 

 

Files

Attachment Size
WiFi Pairing.png 34.71 KB
Standby.png 81.86 KB
PoweronPeak.png 134.12 KB
PowerIntegrationsLNK3202.PNG 49.04 KB
PowerSupplyLoading-Excel.PNG 44.29 KB

Comments

Submitted by saiknaram on 07/23/2022

Can anyone please help?

Submitted by PI-Lupin the Third on 07/26/2022

Hi saiknaram,

Your request has been already delegated to one of our technical support. Thanks for reaching out to us.

Regards,

PI-Lupin the Third

 

Submitted by PI-BERJ on 07/26/2022

Hi saiknaram,

Please try to access PI Expert Online and input your design parameters to determine the recommended component values. I have also taken the liberty of generating a design example for your application. See attached.

Based on this, you would need at least a size 5 (LNK3205) to cater to the maximum load. Size 2 (LNK3202) will not work due to peak current limitations. It is also recommended to increase the input caps to 10uF or go full-bridge if possible.

Audible noise heavily depends on how tight the windings are on the inductor and varnishing (or applying an epoxy) usually helps.

You will also want to check on the device switching frequency if there is 'pulse bunching' involved which generates a subharmonic audible frequency component. The more 'uniform' the switching frequency is, the better.

I would also suggest using the recommended inductance value (680uH for LNK3205) vs a higher inductance. This is to force the device to operate at the highest frequency and away from the audible frequency range.

Regards,

Attachment Size
LinkSwitch-TN2 Buck Design.pdf 473.89 KB