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.

TPS54326: Output ripple reduction

Part Number: TPS54326
Other Parts Discussed in Thread: TPS54425, TPS54325, TPS53511

Dear Sir/Madam,

I am using the TPS54326 and experience higher than expected output ripple (64mVpk-pk).
Input voltage is 12V.
Output voltage is 3.3V.
Load current is under 1A.

Circuit is similar to the following except:
Input capacitance is 47uF+47uF+22uF+10uF.
Output capacitance is 22uF+22uF+22uF

Any thoughts on how to reduce?

Regards Joe

  • Hi Joe,

    how are you calculating the expected ripple? The ripple in the waveform is around 60mV which is around your expected ripple. In any case, what is the exact output current when you measured this waveform? Do you mind also capturing the SW node in addition? Im trying to check if the device is operating in PFM when the ripple is usually much higher than in PWM. 

    Thanks,

    Amod 

  • Hi Amod,

    Thank you very much for the reply.

    The output ripple was measured at the Cout.

    The load current was measured as 200mA.

    The following is at the SW pins.

    Here is the data where I added external load for eco switchover (800ma external load).
    10mv pk-pk where HF noise is from the wires that go to the eload.

    The answer is probably not possible, but how to force the controller out of eco mode?
    If not possible are there other footprint compitiable parts that do not have this issue. 

    Regards Joe

  • Hi Joe,

    Yes this device is entering eco-mode. There are p2p compatible parts TPS54325 and TPS54425 which should allow you to stay in PWM and not have this high ripple during light loads. Please let me know if you can try either of those ICs on the same board. You will not need to make any BOM changes for this except the IC swap.

    Regards,

    Amod

  • Hi Amod,
    Thank you for looking into this.
    The part I am using is TPS54326RGTT (16-VQFN). I could not find any alternatives.
    Is there a way to not use eco mode?

    regards Joe

  • Hi Joe,

    I see. The TPS54325 would have been a perfect p2p replacement but only for the PWP package. Unfortunately, there is no p2p non eco-mode part for the TPS54326 with RGT package. Only other p2p device that has this package is TPS53511 but that device has eco-mode as well so that will not help. Is it possible to redo the layout with the other package or is this too far in the design stage?

    One hack is to increase the inductor to lower the output current at which the TPS54326 will enter PFM or eco-mode. If you check equation 1 on page 7 in datasheet (https://www.ti.com/lit/ds/symlink/tps54326.pdf#page=7), you will see the PWM to PFM transition current can be lowered by increasing the inductor. Of course, since this device is internally compensated, you can only increase the inductor so much before the converter is unstable. 

    Thanks,

    Amod

  • Hi Amod,

    Thank you. 
    The board is in production so change is not going to work at this stage.
    There is a 2512 (6432) current sense resistor added for isolation followed by 1uF capacitor.
    I tried a ferrite and 22uF and that made the ripple worse from 60mvpk-pk to 125mv pk-pk.
    I could use an inductor however with a switching frequency of 700KHz the inductor will need to be large.

    Regards Grant

  • Hi Grant,

    I am referring to the default inductor being used on the design and if that is increased, it will only help reduce the transition point of PWM to PFM. But it will not be able to get rid of it completely. It looks like you are referring to a second stage of ferrite bead + 22uF filter here - is that right? If you are open to additional filtering, you can consider using the same ferrite bead and cap combination mentioned in section 4.1 in this app note: https://www.ti.com/lit/an/slva549/slva549.pdf Alternatively, you can add an LDO in the path to remove high ripple but that is likely difficult now given this is in production already.

    Thanks,

    Amod