TPS54140_ Bad Ripple and transient Loading

Eq 55 is for switching losses.  It only a first order approximation, but it is correct.  It should not have any relation to ripple or transient response.  For most designs, if you are using ceramic output capacitors, output ripple voltage should be very low.  Do not confuse "swirching noise" with output ripple.  Transient response is dependent on output capacitance and ESR.  Depending on your load step requirements you may need to increase your output capacitance or possibly increase the loop bandwidth.  Can you provide waveforms that illustrate your concerns along with your schematic?

7713.R78 project-a.pdfDear John,

Yes, we are using ceramic capacitor and we have a low ripple voltage at full load (only 20mVpp). The ripple problem varies from unit to units from 100mA up to 350mA which varies from 30mVpp up approximately 60mVpp. Our maximum limit is 50mVpp. With regards to transient response, yes it depends on the output capacitance level and esr, and also depends on the LIR of the inductor and loop bandwidth. Increasing the loop bandwidth will not help anymore as the TI IC is so sensitive to noise (you can see this by putting your finger tip on top of the IC on the demoboard when it is operational and monitoring the output ripple voltage). Increasing also the capacitor will not help due to cost impact on the design. For your reference, see attached files.

Do note that the schematic values are not updated and changes are as follows. R2 = 31.6kΩ R6 = 52.3kΩ C6 = 2.2nF

Conditions: Output Voltage = 3.3V Output Current = 500mA Input Voltage = 4.75Vdc – 34Vdc Switching Frequency = 500kHz4670.R78 project-a.pdf3482.ripple.pdf0601.Schematic.pdf

Looks to me like you are skipping pulses or unstable.  Above 400 mA, your output waveform looks normal.  You are switching at 500 kHz, so at 10 usec/div, you should see 5 cycles per division.   The lower current waveforms do not display that, or are unsynchronized with respect to trigger.  Can you provide waveforms of the PH node?  They will need to be well synchronized, and I may need to see both zoomed in (2 usec/div or so) and zoomed out (20 usec/div).  Any scale that gives an accurate picture.  Also it would be helpful to see the loop response at 400 mA.

Dear John,

Pls see attached waveforms for Vout and PH at different conditions. Thank you.

Warmest regards,

Joy4544.ripple_with_ph.pdf

Dear John,

would you be able to advise me base on the eariler email for waveform for Vout and PH.0257.ripple_with_ph.pdf

Pls see attached.

We are currently on extended holiday.  But I have looked at your design.  You are skipping pulses, but the required pulse width is above the min controllable on time.  It looks like a stability issue. You can try changing C8 to 120 pF.  Let me know if that works.  If not you will have to take some loop measurements for further analysis.