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.

LM2840-Q1: Webench frequency response

Part Number: LM2840-Q1

The frequency response of my Webench design for this switching regulator is not great.

With 5.7-15V input voltage range, output voltage of 3.3V and max load 0.04A.

The original design chosen by Webench gave phase margin of 45.37 deg and crossover frequency of 3.86 kHz. Following this I made a few changes to output capacitance (lowered from 120uF x2 tantalum to 47uF x1 ceramic with lower ESR). This produced a phase margin of 39.21 deg and crossover frequency of 11.53 kHz.

Is 39.21 deg phase margin going to cause stability issues? Normally we look for >45 deg at least but >60deg is preffered.
Is 11.53 kHz cross over going to cause poor transient response? Normally we look for 10% of switching frequency although in this case with 1.25MHz switching it may not be possible. At least 30kHz is what we have on our other regulators.

It seems there is compromise between phase margin and crossover frequency as the output capacitance is varied. Which is preferred, higher phase margin or higher crossover frequency?

There is no average model to enable frequency response to be confirmed with simulation. Is Webench accurate for this?

  • Hi Kiran,

    Thank you for reaching out.  I took a look and it seems the L and Cout are big in the Webench model. To get higher crossover frequency, you may consider to reduce Cout and L, but since the loop compensation is internal, you may try a few different combinations. 

    You are right that there may be some limitations by the internal compensation, but the circuit can be made stable if the design guideline is followed. 

    Regarding the tradeoff, I would suggest to get to a good enough crossover frequency then get a higher phase margin.  39deg phase margin should be okay, although you may see some more overshoot and undershoot in a step load but the circuit should be stable. 

    Regarding the Webench model, it has been validated by our engineer when creating the model. However, all models have some approximations here or there during linearization. The advice here is use it to get a good start, and always verify and fine tune your design by experiment.

    Best Regards,

    Youhao

  • The problem with lowering inductance and capacitance is that the converter enters discontinuous mode. Is this a problem?

  • Hi Kiran,

    Strictly speaking, DCM is a valid mode. 

    Thanks,

    Youhao