LM25011 Low Power Noise

 

You will need to check the Switch node to see if it is periodic and stable.  You will see lots of ringing on the Switch node as the current in the inductor falls to zero at light loads.  The buck inductor will ring out at resonance with the parasitic capacitance that exists from the Switch node to Gnd. Assuming that the switch node is stable and periodic; you could increase the buck inductor value to ensure that the part is not in DCM at the lower power levels?  You could add more output capacitance?  Or you can add an additional LC filter at the output?

 

Finally you could opt for a part that does not go into DCM mode (Forced CCM).  A part such as the LM5160?

Thanks Patel and David , I may have solved the problem differently .  I was on Webbench and noticed there is a filter you can select called “ Light Load”  I guess this was put there for people like myself that have lighter loads.   The Webbench came up with LMR14020 as the preferred part.  This part does not have a synchronous switch though.   When you run simulation the range of load it allows is 150 to 7.5 Ohm where as the LM25011 allowed 33 to 1.5 Ohms. 

My new question: Do you think the LMR14020 will work well at a light load and not have the same ringing problems as the LM25011?

Is the " Light load"  filter for this purpose?

Thanks, James Puritch

Hello James,
The LMR10420 is a non-synchronous buck convertor, and if the ringing you are referring to is the DCM ringing when the inductor current goes to zero amps, then yes, this will exhibit the same behavior at light loads. As mentioned in my previous email, look at the LM5160 for a forced continuous mode operation without DCM.

Also, the point at which the converter goes into DCM mode is related to the Switching frequency and the inductor value. High the Switching frequency the lower the load can be before you enter into DCM operation. The point at which it enters into DCM is called the point of "Critical conduction"
To calculate the Iout the point at which the converter is at Critical conduction is as follows.
DeltaiL/2=Iout (Critical Conduction)DeltaiL/2>Iout (DCM)
DeltaiL/2<Iout (CCM)
Where DeltaiL is the peak to peak ripple current in the inductor
DeltaiL=((Vin-Vout)/Lbuck)*((Vout/Vin)/Fsw)Where
Fsw is the selected switching frequency. Hope this helps?

OK, so based on this we should try to be as high a frequency as possible. I tried a bunch of combinations in an effort to get the LM25011 working, as we already have this on our board. I increased the frequency to 1.092Mhz, RT = 300K and increased the Inductance to 100uH. This worked well at no load and full load. I do see that the switch frequency is a little unusual, I am seeing a two equal on pulses with a short off between and then a long off between the next.

Hello James,

The variation in the Switch pulses should be fixed.  By increasing the Fsw and inductance, you have significantly reduced the ripple voltage at the feedback node.  A voltage of 30mV (of ripple) or higher is needed for stable operation (needed for the LM25011A: not needed for the LM25011).   

I direct you to section 8.2.2 in the datasheet under the subheading of Output Ripple Control (if using the LM25011A).  Look in this section and determine the option you have implemented.  Follow the design procedure therein to modify the ripple voltage at the feedback to an acceptable level.

 

Hope this helps?

 

 

Thanks David, this is all a little concerning are you sure the LMR14020 is not a better solution. This creating Vripple is kind of counter intuitive to the goal . IN the LMR14020 datasheet they show the DCM noise to be less than 10mVPP?

Hi James, Hi James,
I am sure the LMR14020 is a non-synch Buck Just like the LM25011. What I am not sure of is what you are seeing on the LM25011 compared to the LMR14020 and the cause of the differences. I was assuming this is just DCM ringing. However there could be a misunderstanding? What would help is reconfigure the LM25011 back to what you had and take some scope shots of what you are seeing and post for me to see. I would like to see the Switch node and the output voltage ripple (AC coupled) showing the noise that is concerning to you?

OK, not a problem as I have many boards unmodified. I am not sure I can attache images here? I have Patel's e-mail.

You can attach pictures here using the rich formatting tool but I have forwarded your pictures to David.