• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 63 subscribers
  • Views 633 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.

Original question:

TPS43061: Ripple noise on Output

TPS43061: High amplitude ringing at the output

Lenio Cacula
Genius 16580 points
Part Number: TPS43061

I measured excessive ringing at the output of the EVM and wanted your help to either correct a possible mistake in the way I'm testing it or to provide clarification to understand the discrepancy I see against the documentation.

Another observation is that a board sitting at ambient temperature seems to have smaller ringing than that of a board with which I loaded the output with higher current, even for a short period of time. Would the MOSFET's or the chip temperatures have an influence in the results?

This experiment is for confirming an observation from a customer, who seems to have reached comparable findings with the ones I tested with the TPS43061EVM-198. I made sure to reproduce the nominal values of input voltage and load to eliminate possible issues with the compensation in the circuit. I therefore applied 9V to the input and measured a well-regulated 15VDC on the output as expected. I have applied different loads and made sure to use a calibrated tip&barrel oscilloscope probe to collect the data. 

The documentation for the EVM estimates that the voltage ripple should be in the order of 70mV when operating in CCM and about 10mV when operating in DCM. Page 8 of the EVM user's guide represents in Figure 9 a ringing of 16V (!!!) peak to peak, which makes me wonder if the indication of 10V/div should actually have been 10mV/div. My measurement however is the following with a load of 1A (CCM):

I appreciate any comments you might have about these findings. Thank you!

  • 0
    TI__Guru* 77065 points

    Hi Lenio,

    Thank you for posting.  First, where do you get 16V p-p ripple voltage?  Do you mean C1 curve?  It is the switch node waveform, not the output ripple.  C3 is the output ripple, and it may be measured with probe bandwidth turned on (mostly 20MHz bandwidth).

    What you show in your scope picture may be real:  ringing at about 30MHz, which can be attenuated by adding some high frequency ceramic apacitors (0.01uF and 0.1uF) in parallel to the output capacitors.  These caps position should be at the output port such that the ripple (ac signal) can be confined inside. 

    Thanks,

    Youhao Xi

  • 0 in reply to Youhao Xi
    TI__Genius 16580 points

    Can you please address my question as to why the documentation shows a much smaller ripple for the EVM? I assume the charts show how the EVM performs out-of-the-box, without adding extra components to it. Also, I did add capacitors and was able to reduce the ripple, but my best effort only got me to 100mVpp. Is that what we should expect out of this solution?

  • 0 in reply to Lenio Cacula
    TI__Guru* 77065 points

    Hi Lenio,

    Because you said you used the tip and barrel probe, so I guess the engineering who wrote the user guide may have turned on the scope bandwidth to 20MHz when capturing the waveform.  I am looking for an EVM and need to run some test myself to see if it is the case.  What I mentioned previously is a general approach to attenuate high frequency ripples.  If capacitor alone does not help much, a ferrite bead may need to be added.

    By the way, your labeled your trace as VDS but you said it was the output ripple, so I believed the label was a typo.  Could you confirm?

    Thanks,

    Youhao

  • 0 in reply to Youhao Xi
    TI__Genius 16580 points

    HI Youhao,

    I appreciate the effort into reproducing the experiment on your end!

    I confirm VDS is a typo, or rather lack of attention about changing the oscilloscope channel label from the previous measurement. The picture I show is on the output of the switcher, by connector J2. I tested few spots along the way such as on top of the output caps, test points, and J2 without seeing any significant difference. Evidently if the probe is leaning towards the inductor the results are much worse, but what I show in the post is actually the smallest measurement I could get.

    Thank you!

  • 0 in reply to Lenio Cacula
    TI__Guru* 77065 points

    Hi Lenio,

    Just let you know that I could not find an EVM here in our lab, so I ordered the EVM and will test it once receiving it. 

    Thanks,

    Youhao

  • 0 in reply to Youhao Xi
    TI__Guru* 77065 points

    Hi Lenio,

    Sorry for the delay.  I just received the EVM and measured on the bench.  I can confirm your results, although not exactly the same but the amplitudes are close.  I believe the engineer who wrote the user guide turned on the 20MHz bandwidth on the scope.

    If you measure directly on C5, you can find the ripple is about 1/3. This means the output capacitor is not effectively used.  The PCB power traces should have been improved to force the current to flow through each capacitor. 

    Anyway, these are high frequency switching hashes (> 30MHz) and it should be easily attenuated by a ferrite bead on the output rail.

    Thanks,

    Youhao