• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 2 answers
  • Subscribers 62 subscribers
  • Views 85 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.

TPS548A20: Output Capacitance

Mark Guastaferro
Genius 11095 points
Part Number: TPS548A20

I need a little help understanding how to determine the minimum amount of output capacitance for the TPS548A20.  The data sheet indicates the minimum capacitance should have the LC pole close to the zero frequency.  The design spreadsheet I found a link to on another forum post sets the minimum capacitance based on the LC pole < Fsw/30. 

With a design for Vin = 12V, Vout = 5V @ 2A, Fsw = 1 MHz, Webench selected 33uF +22uF output caps with L = 6.8 uH. However, the calculated min output capacitance Webench provides says the design needs 103.47 uF min output capacitance. The selected caps don't meet this recommendation.

If I use the data sheet equation with fz = 12 kHz (for Fsw = 1 MHz), Cout min = 27 uF (webench selected caps look good for this)

If I use the data sheet equation with fz = 6 kHz (for Mode pin set for slower RC time constant), Cout min = 103.4 uF.  This must be what Webench calculated but didn't follow when making the cap selection.

If I use design spreadsheet, Cout min = 3.35 uF.  Seems too low to me. Spreadsheet also has Cout max recommendation: Fcl < Fsw/100, which works out to 37 uF in my case. The spreadsheet includes a comment that Cout can be above this needs but to be checked for stability. 

I'm tempted to stay with the Webench selected values, but some clarification as to the best approach would be helpful. 

Another comment about the Webench results.  I found with different switching frequencies and inductors,  that the calculated minimum output cap the Webench reports sometimes uses the faster injected ripple RC while other times uses the slower RC times constant to determine Fz. The schematic for the Mode pin doesn't change.  It's not clear what criteria Webench is using for selecting the RC zero frequency.

  • 0
    TI__Genius 16640 points

    I can duplicate the webench calculations to obtain ~27uF.   I can not get the ~100uF result unless I increase the load step current.

    The design process is to select the inductance so that the ripple current is between 0.25 to 0.5 of output current.

    There are several calculations for Cout, for example ripple, transient and stability, so the required Cout can be very

    widely and will depend on these requirements.  

     

    Thanks for pointing out the mode pin component not changing with switching frequency.   I will submit the bug with the webench team and check if the RC is changing.

     

    I would use the webench result, because the webench model is derating the ceramic capacitor for dc voltage bias and checks the efficiency, power dissipation.

    The capacitor derating is manual in the .xls and does not estimate efficiency.

  • 0 in reply to David_Daniels
    TI__Genius 11095 points

    Hi David,

    Thanks for your comments.  I will stick to the Webench schematic results. But I'm still interested to understand why the .xls uses a different stability requirement.  Targeting the zero frequency seems like a better approach to me. I understand there is a wide range of values that work, just want to make sure the boundaries make sense. 

    Ask the Webench team to try this when they check for the RC bug:

    If I create a design from scratch with the conditions listed earlier, and then after an initial design is created I change Fsw to 1 MHz with ceramic caps only, the design will list min Cout at 27 uF and select appropriate caps.  L = 6.8 uH. Overall result looks good.

    If I then select an alternative inductor, still a 6.8 uH inductor but just a different option, the updated design still selects same output caps, but reported min Cout required now says 103 uF.  Then if I switch inductor back to the original inductor (still 6.8 uH), selected caps again remain the same but min Cout reported still says 103 uF.  This inconsistency leads to confusion. Webench has no option to select the RC time constant. 

  • 0 in reply to Mark Guastaferro
    TI__Genius 16640 points

    Let me review the excel, the datasheet is not explicit on the selection process so I think a similar devices procedure was used..