LM5164: Audible noise from inductor

Just for clarification: The mentioned duty cycle is the duty cycle at which the load switches not the dc/dc converter.

Hi Roger,

Thank you very much for posting and for the clarification in your second post. Can you add more output capacitors such that the step load response over and undershoot can be reduced?  Also check your 22uF capacitor datasheet, because I worry its effective capacitance would be reduced a lot under dc voltage bias of Vout.  

Let me know if you see any improvement.

Thanks,

Youhao

Hi Youhao,

Thank you for answer. I've tried with other capacitors 10uF/25V and assembled a maximum of 60 uF (6x 10 uF) and a parallel 100uF/35V electrolytic capacitor. The ripple is getting smaller but but the noise doesn't disappear. Thanks also for the hint about the reduction of capacity depending on the bias voltage.

I have also tested with higher and lower values for the following capacitors C205, C206, C208, C209 but with the same result - the noise doesn't disappear.

I have another question regarding the schematic from the Webench design. There is just a 2.2uF capacitor but with a 220 mOhm resistor in series, what is the reason for this resistor? Why not connect the capacitor directly to the 12 V output voltage?

Kind regards,

Roger

Hi Roger,

Sorry to hear that the audible noise is still there.  Can you show the new scope picture after you added the capacitors?  Could you ask the customer to try a different vendor's inductor?

Could you show me your Webench design? I can create a design but I don't know your operating conditions so I don't see that 2.2uF and 220mOhm.

Thanks,

Youhao

Hi Youhao,

Here is a scope image when using 6x 10 uF MLCC capacitors. From the test with 6x 10 uF and an electrolyte cap I don't have a scope image but the noise stays the same.

I will try it with a different inductor and also with 2.2 uF and 220 mOhms as in the Webench design. But I must first order this components.

Here are my settings for the Webench Design:

Then in Webench I change the output ripple setting to "Reduced Ripple Solution":

For L1 I choosed an alternative part (CDRH127/LDNP-560MC from Sumida)

Then my design looks like this

Thanks again for your great support!

Kind regards,

Roger

Hi Roger,

Thank you for the info.  I see.  Cff is to inject noise to FB needed for the COT operation.  ESR in the circuit also generate ripples for FB to use. 

The noise should be related to the step load response, and do you want to try with a smaller inductor (I mean smaller inductance but higher Isat spec)? I might be wrong, but I am thinking of speeding up the step load response and a larger inductor would slowd down the response.  Just for a quick test with 50% or the inductance? 

Thanks,

Youhao

Hi Youhao,

Thanks again for your feedback.

I've just ordered a few 100V capacitors and also 68uH and 33uH (with higher Isat) inductors. I will get back to you as soon as I have tried them out...

I don't really understand how the mentioned COT operation with the ripple FB is working. Do you know a good source or AN to learn this?

Kind regards,

Roger

Hi Roger, 

COT is a hysteretic control scheme. Its operating principle is like this:  The circuit will turn on the FET for a fixed time (if VIN is stable), during which Vout (and FB) will rise. After the FET is off, Vout (and also FB) will decrease.  The IC waits until the FB pin voltage drops to reach the reference voltage to turn on the FET again.  The FB ripple voltage should be at least 20mV to get a stable operation.   

Hope this clarifies,

Youhao

Great now I understand, thanks!

Kind regards,

Roger

Please remember to post here AFTER you tested with new parts.

Thanks,

Youhao