• State Resolved
  • Locked Locked
  • Replies 10 replies
  • Subscribers 65 subscribers
  • Views 866 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

LM25574 Inverting Buck-Boost--Webench problems?

Marc Schuman
Intellectual 750 points
Other Parts Discussed in Thread: LM25574

I'm trying to use Webench to design an inverting buck boost power supply with the LM25574.  Input 8-14V, Output -9V @ 100 mA, Fsw 700 kHz.  Webench gives me a design that is topologically correct.  But many of the component values seem questionable.  For example, the values given for Cramp and Rramp are about twice what one would think from the datasheet discussion (for a non-inverting design).  The value for Css doesn't make any sense.  The datasheet does not give a design procedure for an inverting configuration, so I have no way of verifying what Webench is doing.  My main concern is getting the compensation components correct....  How can I proceed with this?

  • 0
    TI__Genius 15810 points

    Hello Marc,

    Have you prototyped the webench design and are seeing issues with the operation? It would be helpful to know your design and BOM that webench generated to see if the BOM  was not corrected properly.

    Regards,
    Akshay 

  • 0 in reply to Akshay Mehta
    Intellectual 750 points

    I don't want to spend the time and expense on prototyping a design that looks pretty clearly to be incorrect.  Could you provide a manual design procedure for the Inverting Buck-Boost configuration of the LM25574?

  • 0 in reply to Marc Schuman
    TI__Genius 15810 points

    Marc,

    We do not have a ready design procedure for buck-boost with us as yet. But we are working on putting together a board and calculating certain component values to check the stability. We should have the information soon.

    Thank you for the patience.

    Regards,
    Akshay 

  • 0 in reply to Akshay Mehta
    Intellectual 750 points

    Akshay, thanks for your help!  I am attaching the schematic that Webench produced for me.

    Note that the Webench value for Css is 10 times what the datasheet seems to suggest.  Also the values for Cramp and Rramp are 2x what the datasheet says (for a buck design at least).  Perhaps these differences are needed for the inverting configuration; I don't have the information on the internal design of the IC to determine that.  But the Css value seems likely to be an error.  And the loop compensation values seem to provide much higher gain-bandwidth than I would expect to be appropriate from what (admittedly little) I know.

    7120.webench_design_1222828_61_947586274-negative.pdf

  • 0 in reply to Marc Schuman
    TI__Genius 15810 points

    Marc,

    There was an error in the calculations for Cramp and the Css value limits seemed to not be calculated properly. To recalculate the Css value, you could just use the equation #10 in the datasheet. We put together some calculations and came up with some BOM components that we have tested on the bench. The new values for Cramp, Rramp, Rcomp, Ccomp, and Ccomp2 are as follows:

    Cramp = 500pF
    Rramp = 200kohms
    Rcomp = 37.4kohms
    Ccomp = 6.8nF
    Ccomp2 = 68pF

    The load transient test of no load to 100mA results in less than 100mV of voltage overshoot as seen here:
     T

    This test was performed at only room temperature and exhibits good stability. 

    We also obtained a couple bode plots of the circuit using the new compensation component values. One was taken with Rramp connected and one was taken without Rramp connected. The attached pdf file has two plots appropriately labelled. The plot without Rramp shows some amount of peaking at about 100Khz. With the addition of Rramp, extra slope is generated and this helps in reducing that.
    5025.LM25574 bode plots 8Vin -9Vo 100mA.pdf
     

    The crossover frequency of 7Khz is close to the target value of 6Khz. The target value is obtained by calculating the frequency of the right half plane zero caused by the inductor. This zero does cause the gain to increase by 20dB/dec but at the same time reduces the phase like a pole would. To have a stable response, the system crossover is designed to be a decade away from the RHPzero. The RHP zero frequency for your design using 8Vin min and 100uH is 60Khz. 

    I hope this information is helpful for you.

    Regards,
    Akshay 

  • 0 in reply to Akshay Mehta
    Intellectual 750 points

    Hi Akshay,

    Thanks, that is extremely helpful.  The compensation that you used looks much more like what I would expect.  Can you provide the full schematic, so I can make sure the other values are what I'm thinking?  And/or verify that L=100uH, Rfb=10.5k/1.65k, Cout=22uF?

    Do you feel that 100uH is a good inductor value for this design?  I will in fact sometimes be operating at 10 to 20MA load.

    Are / will the corrections to the calculations be put into Webench?

  • 0 in reply to Marc Schuman
    TI__Intellectual 2930 points

    Hi Marc,

    We've been working with Akshay and have identified the errors in WEBENCH and should have them fixed promptly.

    Best regards
    Tommy 

  • 0 in reply to Marc Schuman
    TI__Genius 15810 points

    Marc,

    The only BOM components I changed were listed in the previous message. The schematic stays the same and L and Cout values were also the same. Here are the part number I used for L and Cout:

    Cout = C3225X7R1C226M
    L = DR74-101

    The 100uH inductor, I believe, is based on a 30% ripple current and would help in operating in CCM. It should be fine to use this value. 

    Regards,
    Akshay 

  • 0 in reply to Akshay Mehta
    Intellectual 750 points

    Akshay, I just want to verify that the feedback resistors you used were (approximately) 10.5k and 1.65k, since that impedance is a critical part of the compensation.

    When driving a light load the converter will be in DCM.  You don't see any stability concerns in that case?

    Thanks again for your help!

    Marc

  • 0 in reply to Marc Schuman
    TI__Genius 15810 points

    Marc,

    The resistor values were the same too. Upper was 10.5k and lower was 1.65k ohms. The inductor value should be fine for your operation. In the plot that I sent before you will notice that Vout is in regulation even at no load. 

    Regards,
    Akshay