This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM3478: Output noise of booster circuit

Part Number: LM3478

Thank you for your help.
I am making 24V to 35V with a circuit like the attached circuit diagram.
This circuit is a circuit generated from WEBENCH in 2017.
(VinMin = 20.0V, VinMax = 28.0V, Vout = 35.0V, Iout = 4.0A)
At light and no loads, the 35V output looks like the attached waveform.

Question 1. Please tell me the reason why the 35V output is not constant and looks like the attached waveform.

Recently, when I simulated again with WEBENCH, I noticed that C108 in the attached circuit diagram changed from 10nF to 10pF.
When the constant was changed to 10pF, the frequency of voltage drop of the attached waveform tended to decrease.

Question 2. Why does the 35V output voltage drop frequency decrease when C108 is set to 10pF?

Question 3. What are some ways to eliminate the voltage drop in the attached waveform?

Thank you.

  • Hi Kiyoshi,

    Thank you for using LM3478. The output waveform is normal for the light load and no load operation. This is caused by the controller skipping pulses due to the minimum on time limit. The 10nF filter capacitance is too big for the ~600kHz switching frequency in your case. After this capacitance is reduced, the switch can be turned on more easily, which explains why the pulse skipping happens less. To eliminate this, you can add a dummy load in addition to the external load. You can also increase the inductance value to make the pulse skipping happen at a lighter load condition.

    Thanks,

    Yinsong

  • Hi Yinsong,

    Thank you for your support.

    I'm thinking of adding a dummy load,
    How much dummy load do I need?
    What is the recommended resistance or current value?

    Thank you very much.

  • Hi Kiyoshi,

    Please use the equation below as a starting point. You will still need bench test to confirm and also please remember to consider all the tolerances.

    Io=((TonVin)^2 * fs)/(2*L*(Vo−Vin))

    Thanks,

    Yinsong

  • Hi Yinsong,

    Thank you for your support.

    If I use the equation you contacted,
    I understood that the additional load current required can be estimated.

    I don't understand the meaning of the equation,
    Could you give me a brief explanation?

    Thank you very much.

  • Hi Kiyoshi,

    The idea behind is that as the load is reduced, the boost starts to operates in DCM. In DCM, the duty cycle reduces as the load continue to reduce. Eventually, the duty cycle reduces to the point where the minimum on time is reached. And the pulse skipping takes over after that. So the dummy load has to ensure it's above that critical point.

    Thanks,

    Yinsong

  • Hi Yinsong,

    Thank you for your support.

    After that, test using the actual machine and
    I will decide on an additional load current.

    Thank you very much.