• State Resolved
  • Locked Locked
  • Replies 9 replies
  • Subscribers 63 subscribers
  • Views 1258 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.

TPS61221/TLV61225: high parasitic noise

Dmitry Vasilenko - Arrow
Expert 1380 points
Other Parts Discussed in Thread: TPS61221, TPS61220

Hi,

my customer evaluated TPS61221 for a regulator 3.3V / 70mA from a battery (0.7V). The power is produced for small RF key and the customer says that with TPS61221 the range of RF key dropped dramatically (by 3 times). Before they used AIC1642-33 and it looks much better in emitting parasitic noise. Alternatively they evaluate L6920DB (STM).

The customer did several experiments with TPS61221 and found out the EMI-noise of the product is too big in the range 100Hz-1GHz.

Is there any solution to overcome the problem?

Is TI avare of the problem?

  • 0
    TI__Guru**** 150830 points

    100 Hz to 1 GHz is quite a wide range!  The switching frequency of the TPS61221 will fall in that range as will every other switching power supply.  An equation for estimating the switching frequency is found on page 14 of the datasheet and can be easily changed by changing the inductance.

    What frequency was causing them problems?

    The other key thing is board layout.  Keeping the loop areas small and switching nodes small will reduce the EMI.  Follow the layout of the EVM.  They can also test with the EVM to see if using a different layout helps their situation.

    Since they have been able to use 2 other switching power supplies, they should be able to use this one also.

  • 0 in reply to Chris Glaser
    Expert 1380 points

    Chris,

    The RF frequency is 434MHz, the load is 1.5V 50mA and switchng frequency is about 1MHz

    I can send you schematics and layout via e-mail .

  • 0 in reply to Dmitry Vasilenko - Arrow
    TI__Guru**** 150830 points

    Ah, since the frequency of trouble is that high, it must be EMI from the inductor or SW node radiating.  The solution to this would be board layout and/on inductor selection.

    Are they using a shielded inductor?

    They could try a different inductance to change the switching frequency which may move the noise frequency but board layout and/or shielding of the switcher from the noise sensitive components are the best bet.

  • 0 in reply to Chris Glaser
    Expert 1380 points

    Chris,

    they have tried different inductors, power line filters, capacitors - no improvement. They insist that L6920 has same parameters, but much less noisy.

    I have layout of the product. Can you analyze it?

  • 0 in reply to Dmitry Vasilenko - Arrow
    TI__Guru**** 150830 points

    Yes, I can look over their layout.

    Have they tried different inductances?  Different inductor values will move the switching frequency of the IC around and may help.

    Have they tried a ferrite bead on the output?  It could dissipate any high frequency noise on the output.

  • 0 in reply to Chris Glaser
    Expert 1380 points

    Chris,

    What's your opinion regarding layout?

    I have discussed the issue with the customer once again. They have actually tried all the possible improvements (shielded inductor with different values, ferrite bead, evm instead of the TPS mounted on PCB), but did not see any significant effect. The engineer said they have discussed the issue earlier with your local TI application engineer. We desided to stop the improvement of schematic with TPS61221 and try another part: TPS61097. I have ordered EVM and samples. Hope this part is better.

    Note that the alternative from ST is not better: it is too hungry for current and consumes more than 20 uA, which is unacceptable for the application.

  • 0 in reply to Dmitry Vasilenko - Arrow
    TI__Guru**** 150830 points

    Dmitry,

    Please let me know if your testing with the TPS61097 is better than with the TPS61220.

    To further work the TPS61220 EMI issue, we need to figure out if it is radiated or conducted emissions that are affecting the output noise of the system.  Conducted would be noise on the 3.3V output of the device that could be filtered with ferrite beads, or affected with different operating frequencies by changing the inductance.

    Radiated would be noise due to the switching of the device and then radiating through the air.  This noise is reduced by shielding of the source or slowing down of the slew rate (most likely on SW) that causes the noise.

    At this point, I await your feedback on the TPS61097.

  • 0 in reply to Chris Glaser
    Expert 1380 points

    Chris, I promised you to give a final feedback on the tests of the TPS61097. The customer says the situtaion is the same as with TPS61220: too much radiated noise. As well they tried different techniques to reduce the noise:

    1) the used ferrites to reduce conducted noise

    2) moved the EVM away from the antenna and schielded (to reduce radiate noise)

    In both cases the result was the same: dramatic reduction of the wireless link distance (by factor 2-3).

  • 0 in reply to Dmitry Vasilenko - Arrow
    TI__Guru**** 150830 points

    Thank you for your reply and for continuing your testing.  A few more things to try on either IC:

    Have you tried adding a snubber to the L pin?  Here is an app note that discusses how to: http://focus.ti.com/lit/an/slva255/slva255.pdf

    Also, you can try shielding the IC and noisy nodes of the circuit board.  To do this, add some metal (copper, aluminum, etc.) over the IC package and the L node--cover as much as you can without shorting anything.  Then, Connect that metal to the PCB board ground.  This will help for radiated emissions.

     

    Looking at the efficiency graphs of the TPS61220 and L6920, the TPS61220 looks a little bit better over the whole range.  This could be due to faster switching times in our device, which would translate into slightly higher noise.