• State
  • Locked Locked
  • Replies 16 replies
  • Subscribers 65 subscribers
  • Views 2008 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

TPS54140A: Ripple in very small load operation

Toshiya Asai
Prodigy 180 points
Part Number: TPS54140A
Other Parts Discussed in Thread: TPS54140, TPS54160, TPS54160A

I have a ripple problem with TPS54140A in very small load (<50mA).
The ripple wavefroms are (1) shown below.
So I selected a low inductance to reduce the ripple.
(2) is the circuit diagram, and (3) shows the waveforms(IL and Vout) captured from this circuit.
I wonder the circuit (2) is a correct solution for a very small load application.
If there's a problem or an advice, Please let me know. It would be very helpful.

Regards,

(1) Comparison of different inductance value (when no load attached)



(2) Circuit



(3) Comparison of different Loads with inductance=4.7uH (above circuit diagram)




  • 0
    TI__Guru 74162 points
    Hello,

    I forwarded your enquiry to the product expert. In the meantime, I took a look at the schematic and noted that this is a 1.5A converter. Thus, it runs in DCM ("pulse skip ECO mode" as described in the datasheet) at light loads.

    What is your output voltage ripple spec?

    Regards,
    Tim
  • 0 in reply to Timothy Hegarty
    Prodigy 180 points

    Hi Tim,
    Thank you for your reply.

    ＞What is your output voltage ripple spec?

    ±5% is desirable.

    Regards,

  • 0 in reply to Toshiya Asai
    TI__Guru 74162 points
    Increasing the output cap reduces the voltage ripple increase related to the triangular inductor current pulses at light load. Using a higher inductance also helps.

    Regards,
    Tim
  • 0 in reply to Timothy Hegarty
    Prodigy 180 points
    ＞Using a higher inductance also helps.

    According to the waveforms in the [ (1) Comparison of different inductance value (when no load attached) ], the smallest inductance case gives the best result.

    Regards,
  • 0 in reply to Toshiya Asai
    TI__Guru 74162 points
    That may be correct - the energy per pulse (0.5*L*I^2) is increased if the inductance is higher.

    Have you tried increasing Cout? This should reduce the output voltage ripple amplitude (i.e. for a given packet of energy supplied to the output cap via the inductor).
  • 0 in reply to Timothy Hegarty
    Prodigy 180 points

    ＞Have you tried increasing Cout?

    Here's the comparison of different value of Cout.
    The difference is cycle time of ripple waveform.

    Regards,

  • 0 in reply to Toshiya Asai
    TI__Guru 74162 points

    Looks like you're getting a lot of switching pulses when the IC comes out of sleep mode. Can you send the layout for review? In particular, the FB resistor placement and FB trace length are important as these are factors in how well the FB voltage tracks VOUT.

  • 0 in reply to Timothy Hegarty
    Prodigy 180 points

    Hello,

    Here's the layout. I think the feedback line is not so bad.

    Regards,

  • 0 in reply to Toshiya Asai
    TI__Guru 74162 points
    The feedback trace is too close to the inductor and is likely picking up noise from the high-voltage slew rates of the SW node. A small cap from FB to GND may help (47pF). Otherise, I recommned a layout change to resolve the issue.

    Regards,
    Tim
  • 0 in reply to Timothy Hegarty
    Prodigy 180 points

    Hello,

    ＞A small cap from FB to GND may help (47pF).

    I added 47pF on FB. but I think its waveform is almost same as previous one.This waveform comparison is shown below.

    ＞I recommned a layout change to resolve the issue.

    I moved the inductance a little(～5mm) to prevent the interference with FB line. But its waveform is almost same.

    Regards,

  • 0 in reply to Toshiya Asai
    TI__Guru 74162 points
    Not sure if moving the inductor will help as the SW copper is still adjacent to the FB components and trace. I suspect the noise coupled to FB from SW is causing additional pulses when the converter starts to switch, thus increasing the output ripple.

    Regards,
    Tim
  • 0 in reply to Timothy Hegarty
    Prodigy 180 points

    Thank you for your advice.

    ＞ the SW copper is still adjacent to the FB components and trace.

    I modified the board to remove the SW copper adjacent to the FB components.
    But the ripple still exists.

    Regards,

  • 0 in reply to Toshiya Asai
    TI__Expert 6860 points

    THe peak to peak ripple that you are seeing is little higher than expected .  Here is the waveform of 3.3V output at light load taken from TPS54140 EVM .

    Basically , When the COMP pin lowers to 500 mV typical, the device enters Eco-mode operation(pulse Skip).When the lower inductance is used , the device is out

    of pulse skip mode at much lower current due to higher peak current (which results in higher comp value) .

    ---Ambreesh

  • 0 in reply to Ambreesh Tripathi
    Prodigy 180 points

    Thank you for your feedback.

    (1) TPS54140,TPS54140A
    I have used TPS54140(non-A version) in our previous design.
    In case of TPS54140, there's no large ripple like the current design. I tested the previous board today, with no load.
    Both circuits are almost same.
    But I have not replaced TPS54140A on the current board with TPS54140, because it's hard to replace with my hand. (especially GND-PAD)

    (2) Regarding as comp value, I evaluated the current design (shown at the top of this thread) using Bode plot tool of WEBENCH. Phase/Gain margin is OK.
    Are there any problem about using the above circuit ?

    Regards,
    -------------------------------------------
    ★Two weeks have gone from my reply above.
    I am still expecting someone's response regarding this issue.

    Regards,

  • 0 in reply to Toshiya Asai
    TI__Expert 6860 points
    1) TPS54160 and TPS54160A should have samilar output ripple perfromance . There is just Enable threshold improvement in the A version .
    2) I simulated the design and the closed loop performance looks OK


    What is the output load current you are designing for ? If you are designing for low current application , you can use lower inductance value and that way the device will come out of ECO mode at much lower output current . Hence your implementation with 4.7uH inductance should work well .

    --Ambreesh
  • 0 in reply to Ambreesh Tripathi
    Prodigy 180 points


    > 1) TPS54140 and TPS54140A should have samilar output ripple perfromance . There is just Enable threshold improvement in the A version .

    I think thers's a trivial difference of behavior in very small load.

    > you can use lower inductance value and that way the device will come out of ECO mode at much lower output current . Hence your implementation with 4.7uH inductance should work well.

    Thank you for your confirmation.

    Regards,