• State
  • Locked Locked
  • Replies 4 replies
  • Subscribers 99 subscribers
  • Views 426 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.

DM365 power supply noise and how it relates to the audio codec/microphone subsystem

Chris Reed
Prodigy 100 points

When we connect a bench supply to the VDDA18_PLL pin we see at least 3dB performance improvement in the Audio Codec/Microphone subsystem PSNR.   When we play back the audio that has been captured by microphone subsystem we can hear the improvement.

In our DM365 based design we have a 1.8V, 1.35V, and 3.3V switch mode regulators that provide power to the DM365.  The 1.8V supply feeds all the 1.8V pins except for VDDA18_VC (which is the audio codec power pin).  The VDDA18_PLL and VDDMXI pins have ferrite beads between the pin and the 1.8V rail.  Hence we can externally power the  VDDA18_VC, VDDA18_PLL and VDDMXI pins.

Questions:
1. What is the relationship between VDDA18_PLL and the Audio Codec subsystem?  In other words, would you expect low frequency noise on the VDDA18_PLL pin to cause audible noise in the Video Codec?
2. If you expect that noise on the VDDA18_PLL will cause microphone/codec subsystem noise, is it acceptable to tie the VDDA18_PLL, VDDMXI, and VDDA18_VC pins together and power then off the same linear regulator?  If so, how much current will these pins draw?
3. What other power/ground signals are going to cause noise in the microphone subsystem?  For example the VSSA18_VC is the voice codec ground pin.  I have VSS_MX1 tied to the board ground because I am using an external oscillator.  Do I need to do something special with this ground?  Does the VDDA18_PLL pin(s) have similar grounds that need to be isolated?

  • 0
    TI__Genius 9775 points

    Hi Chris,

     

    I've checked with one of our designers and he recommends using a decoupling capacitor between VDDA18_VC and VSSA18_VC.  A recommended value is 10uFand 0.1uF (two capacitors).

     

    The designer also recommends using a dedicated power supply to obtain best performance (as you have performed with your experiment).  Adding an LC (RC) low pass filter between the VDDA18_V and the regulator is also recommended.

     

    His last recommendation is to use a notch filter in the ADC path to reject the undesired frequency.

     

    Given all this information, I think you probably have some issues with your board routing that is coupling noise into your signal.  You need to make sure you do a good job isolating both the signals and the power supply.  You could have some noise getting coupled over the signal trace going into your amplifier, power supply to the VDDA18_VC trace (in case there is no power plane--power planes are always recommended), or ground planes.  Do you have any switching signals going through the same ground path that your microphone system is using?

     

    The *_VC power signals on the data sheet are the ones we need to ensure are clean and isolated for best performance.

  • 0 in reply to Ivan Garcia
    TI__Guru* 84110 points

    Chris Reed,

    Any new information? Or hopefully you have eliminated your noise problems.

    Please let us know if we can offer more help.

    Regards,
    RandyP

  • 0 in reply to RandyP
    Prodigy 100 points

    Randy,

    I have a 1.8V linear regulator that is providing power VDDA18_VC.  This regulator only provides power to this pin.  This regulator has a PSRR of 70dB which should eliminate almost any 60Hz noise from getting into the Codec subsystem.  I also have the codec ground (VSSA18_VC) tied to the main power connector ground; basically in a type of star topology. 

    Here is some more detail:

    1. We have found that if we tie the VDDA18_PLL pin to the same regulator that is powering the voice codec that we get about 2dB better PSNR in our Codec subsystem.  Hence the questions as to if/how the VDDA18_VC and VDDA18_PLL power supplies are related.

    2. The only thing that really eliminates the 60 Hz noise is to power the board off a bench supply or a battery. 

    3. If I replace the microphone with a 2K ohm resistor the 60Hz noise is eliminated.  The 2K ohm resistor essentially biases the microphone input around Vdd/2.

    Moving forward:

    1. Hence it looks like there are two sources of 60 Hz noise.  One is within the microphone subsystem and the other is in the wall brick supply.  I have tried all sorts of experiments to eliminate the noise on the microphone subsystem and a low pass filter on the power supply is not really practical. 

    2. At this point we have enabled the Codec's notch filter and set the notch to have a cuttoff frequency of 60Hz. 

  • 0 in reply to Chris Reed
    TI__Guru* 84110 points

    Chris,

    Sounds like you are making good progress, even if not completely satisfying.

    At this point I am stepping out of my realm of knowledge, but I figure I will make the comments I can think of. Your knowledge and experience on these board-level issues are surely greater than mine, but maybe I will raise some dust in the room.

    For any noise problem, there have to be three components: 1) sourcing transmitter, 2) transferring medium, and 3) receiving antenna. For your case, you seem to have found that the sourcing transmitter is somewhere between the wall socket and the power rail(s) on the board (brick, external wiring, power supply components, board wiring). And the receiving antenna must be the physical microphone or its wiring or the first stage amplifier.

    To track down the noise medium, have you tried shielding with grounded aluminum foil? You could put this carefully around the wall brick and wires, and also around the microphone and its wires. It would be harder to do this safely to shield the board components, but maybe some method could be worked out with some form of insulation around foil or sheet metal, with grounding if possible.

    These may be in the list of things you have tried already.

    Best of luck to you,

    Regards,
    RandyP