• State Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 1 answer
  • Subscribers 63 subscribers
  • Views 3526 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.

TPS54225 maximum ripple current?

Olli Männistö
Prodigy 150 points
Other Parts Discussed in Thread: TPS54225, SWITCHERPRO, TPS54325

The datasheet for TPS54225 staters the output inductor ripple must not be more than 40% at maximum load.

(http://focus.ti.com/general/docs/lit/getliterature.tsp?genericPartNumber=tps54225&fileType=pdf October 2010 page 12)

It is also says in the same paragraph

"Larger ripple current increases output ripple voltage, improves S/N ratio and contributes to stable operation.
Smaller ripple currents result in lower output voltage ripple. When using low ESR output capacitors output ripple
voltage is usually low, so larger ripple currents are acceptable."

This seems a little counterintuitive. Is >40% ripple going to cause instability due to D-CAP2  control's compensation or other ill effects?

In my application I'd have 1.8V 900mA and using 6.8uH inductor would result in about .5 ripple (.45 ripple, .9 iout). Components are otherwise happy with this arrangement due to ceramics.

  • 0
    Prodigy 150 points

    In fact switcherpro starts complaining if you try to use inductor >6.8uH in this design. So it seems either the datasheet is wrong or switcherpro is wrong.

    Other parameters are 12Vin, 22uF 16V ceramic Cin (about 10uF eff), 2x22uF ceramic Cout.

  • 0
    TI__Guru**** 191445 points

    While all that is in general true, let me see if I can explain what you really need.  Lets start with DCAP control.  In DCAP the on time is set by the Vout to Vin ratio.  The off time is semi-hysteretic.  During the on time, the inductor current is increasing and teh output voltage ripple component is increasing also.  Now during the off time, teh ripple voltage is decreasing along with the inductor current.  The controller compares the falling output ripple component to the internal reference.  Whent the output ripple falls to the reference, the off time is completed and teh next switching cycle starts.  The controller requires a relatively large ripple component on the output for proper operation (low inductor ( high percentage ripple current) value and high ESR for the output cap).

    Now DCAP2 such as TPS54225 is designed for low ESR output caps.  there is an internal, proprietary circuit to inject some ripple directly at the VFB pin.  So you still need some amount of ac ripple current in teh inductor for proper operation.  From a stability standpoint, you are more concerned with the LC corner frequency.  The LC corner frequency must be high enough that the phase does not reduce to zero prior to crossover, and low enogh that the crossover frequency is well below one half te switching frequency.  This constrains the inductor and capacitor values for the output filter. So for a given output voltage and output capacitance, there is a particular inductor value that gives the best stability.  That value is independant of the output current (and percent of ripple current).  These recommended component values are shown in the EVM users guide. 

    So for your design a 3.3 uH inductor would be ideal.  Eventually we will update teh datasheet to reflect this.

  • 0 in reply to JohnTucker
    TI__Guru**** 191445 points

    Switcherpro should have selected 3.3 uH for your design.  Let me know if this is true. Ignore the datasheet equation for inductor value.  The equations for peak and RMS currents are correct,

  • 0 in reply to JohnTucker
    Prodigy 150 points

    With regards to stability, the datasheet does not consider this at all. In fact I've been reclutant to implement DCAP2 control in our product as I have designed various SMPS circuits over a decade and so far I've not seen a free lunch.

    So in fact there is stability criterion, we just do not know about it. There's no way to determine the feedback loop crossover or phase margin with the information given in the datasheet.

    Now that you pointed me to the EVM manual, I can at least make some informed decisions WRT component values. I told switcherpro to use 4.7uH as I'm using this inductor  on another design and I usually try minimize stock lines. In any case, switcherpro feels <6.8uH is OK for 1.8v / 2x22uF output capacitor so it should work OK. 2x22uF was chosen for the same reason. So it's just a fluke I'm using compatible values to start.

    Wrt changing output capacitor values, if I go for 22uF instead, I take I should use 6.8uH inductor for best effect? For 100uF that would work out to 1,5uH. Assuming the relationship is directly determined by the resonant frequency.

    Switcherpro recommends 2.2uH inductor for this design if I let it make the decisions.

  • 0 in reply to Olli Männistö
    TI__Guru**** 191445 points

    The TPS54225 family of DCAP2 is meant to be used with the fixed values as recommended in teh users guide, since there is no provision for external compensation.   The recommended values are known to produce stable circuits, but of course there are other possible solutiuons.  Initially it was thought that the loop measurement would not be valid as DCAP2 is semi-hysteretic. In fact DCAP 2 loop performace can be measured (and therefore modeled), we just do not support that at this time.  The academic study and model are not yet published or peer reviewed.  You can break the loop in circuit and make your own measurements and determine if  your desired output filter meets your stability criteria.  Make sure to break the loop with both VFB and VO on one side and Vout on teh other.

    I tried your design in switcherpro. I got 3.3 uH as I predicted.  I know that is what it should do, because switcherpro is hard coded to use teh values from teh users guide tables.  In switcherpro you can always make manual component changes.  For other parts, the loop is re-calculated.  For these DCAP2 parts, switcherpro does not model or check the loop response.

  • 0 in reply to JohnTucker
    Prodigy 150 points

    JohnTucker said:
    I tried your design in switcherpro. I got 3.3 uH as I predicted.  I know that is what it should do, because switcherpro is hard coded to use teh values from teh users guide tables.  In switcherpro you can always make manual component changes.  For other parts, the loop is re-calculated.  For these DCAP2 parts, switcherpro does not model or check the loop response.

    Did you try the desktop or online version? On the desktop I'm pretty sure I got 2.2uH suggested. I'll re-check that in Tuesday if the values were right (1.8 Vo 44uF Co).

    Desktop model started complaining about keeping inductor <= 6.8uH which would make sense. 2x difference in single component value shouldn't compromise the design too much, assuming the original values provide ample phase margin. Of course I have no idea how complicated the Dcap2 feedback loop turns out to be. If it's something like SEPIC with IIRC 6 zeros/poles..

    In any case the datasheets should be revised ASAP as they give incorrect info about selecting component values. That's what made me suspicious in the first place, interchangeable component values and no feedback loop considered..

  • 0 in reply to Olli Männistö
    TI__Guru**** 191445 points

    I used the on line version to get 3.3 uH. The desktop should be the same.  It is supposed to be just a straight look up table.  You enter in teh output voltage and it picks the components based strictly on that.

     DCAP2 mode is not really all that complex, but It is a proprietary control method, so there is not any of outside scholarly literature like you will find for voltage and current mode control,. You can't go to a text book to find the small signal model.  I do not personally have the bandwidth to go of fand develope one.  I think someone in Japan is working on this though. 

    As far as the datasheet is concerned, it is somewhat difficult to make changes after the literature is released, and it would be the responsibility of the systems engineer.  I knowthat is on the big list of things that need to be done, but we are currently in a push to finish our part releases before teh end of the year.  I doubt you will see anything on that before January.

  • 0 in reply to JohnTucker
    Prodigy 150 points

    I just checked it, desktop version recommends 2.2uH for TPS54325, 12Vin, 1.8Vout, 0.9A. For TPS54225, 3.3uH is recommended.

    I assumed they're the same device except for the current capability, but recommended inductor type(s) appear to be different. Looks like the internal compensation/amplifier circuit is different.

  • 0 in reply to Olli Männistö
    TI__Guru**** 191445 points

    Yes there are subtle differences in the power stage that may require different output filter components for teh same condtions for different devices in the product family.