• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 62 subscribers
  • Views 164 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.

TPS54495: Excessive noise on Vout

Brenda Holmes
Prodigy 10 points
Part Number: TPS54495
Other Parts Discussed in Thread: TPS62135, TPS62912, TPS65400

I designed in TPS54495PWP on my latest board to create 5V and 3.3V from 12V, as seen below:

As designed, noise on both Vout's was coupling onto other parts of my board. 

For 5V out, base current draw is around 7mA.  When measured after FB11, there is a 1.3kHz sawtooth waveform with a Vpp of 135mV.  I experimented with different output capacitance and ferrite bead values, with the best improvement resulting from removing the ferrite bead, short across the pads, and up the output capacitance to 147uF.  The sawtooth frequency increased to 2.8kHz and Vpp of 110mV. 

For 3.3V out, base current draw is around 250mA.  When measured after FB8, there is a 38kHz sharktooth waveform with a Vpp of 70mV.  In this case, by removing and shorting the Ferrite bead and changing total output capaciatnce to 47uF, I was able to improve to  ~65kHz waveform at only 20mVpp. 

At this point, I would appreciate any guidance you could provide to improve the noise levels on these buses and any changes or improvements that are needed to the overall design. My design is very sensitive to audio frequencies, so minimizing or eliminating them is important. 

  • 0
    TI__Mastermind 24010 points

    Hi Brenda,

    Thank you for your question. Based on the current levels, the device is likely running in PFM where in general the output ripple will be higher. I see you are using only a 10uF output cap on both outputs in the first LC stage. The recommended is between 22uF-68uF caps and preferred to use ceramic caps. Are the output caps (C133, C102, C129, C101) all ceramic caps? It would be a good idea to add 2x or 3x 22uF ceramic caps (instead of 1x10uF C129, C133 ) on both the outputs to reduce the ripple at the output of the first LC stage. You are effectively increasing the output cap of the first stage when you short the ferrite beads and hence observing lower ripple. Do you have any waveforms of the output ripple?

    If you are looking for low ripple and noise, you can choose devices that run in forced PWM even at lighter loads which should keep your output ripple much lower than in PFM. Are you open to looking at other devices? Does the device have to be a dual output or can single output devices also work? 

    Thanks,

    Amod

  • 0
    Prodigy 10 points

    Hello Amod,

    Thank you for your feedback.  Here are my comments:

    1.  Yes, all output caps are ceramic.

    2.  For 5V, I did try changing C129 to 47uF and C101 to 10uF.  This would increase the capacitance of the first LC stage.  It did reduce the ripple compared to the original configuration (112mV Vpp, 2.6kHz), but not as significantly as removing the FB.  I will try the three 22uF to see if that changes the behavior.  However, I'm not hopeful this will reduce the ripple sufficiently.

    3.  Here are the waveforms.

    5V Unaltered 100mV/div, 500us/div

    5V Original, 100mV/div, 500us/div

    5V Modified, 50mV/div, 500us/div

    5V Modified (no FB, C129 = 47uF, C101 = 100uF), 50mV/div, 500us/div

    3.3V Original, 50mV/div, 20us/div

    3.3V Original, 50mV/div, 20us/div

    3.3V Modified, 50mV/div, 20us/div

    3.3V Modified (no FB, C133 = open, C102 = 47uF), 50mV/div, 20us/div

    4.  Yes, I am open to changing to another device, if that is what is deemed necessary.  The ripple/noise on 5V is too significant at this point for my overall design to function, so, unless this can be reduced significantly, I think I will need to change.  Is there a part you would suggest?  The 5V bus has very low default current draw, but needs to be able to supply peripherals that plug in on USB, which is why the 4A part chosen.  A dual is not required; the busses can be generated separately, and 3.3V could even come from an LDO if preferred.  The main thing I need is the low ripple.  Any suggestions?

  • +1 in reply to Brenda Holmes
    TI__Mastermind 24010 points

    Hi Brenda,

    Yes just adding the 3x22uF output cap will not get rid of the ripple significantly. You will also need a second stage LC filter such that the lower frequency  component from PFM operation can be attenuated. This may require a rather larger inductor and capacitor. I would recommend  using an LDO after the TPS54495 outputs. This will filter out low as well as high frequency noise from the output of the TPS54495. You can start a separate thread for LDO device suggestions.

    If you are looking for other buck converter only options, any device that functions in forced FPWM or CCM will reduce ripple significantly at lighter loads. Ideal part for low ripple is TPS62912/3 family which can deliver 2A/3A current. For the 4A, you can consider TPS62135/6 and use it in FPWM. Unfortunately, it looks like both these parts have zero inventory but you can check with TI customer support at https://ticsc.service-now.com/ for latest on current and future inventory.  One other part is TPS65400 which is however a quad channel (4 outputs) device that can also run in CCM only. But, since it has 4 outputs, I am not certain if that is required for your design.

    Hope this helps.

    Thanks,

    Amod