• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 31 subscribers
  • Views 3427 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.

help on reducing conducted EMI from TPA3118D2

Chi-wen Wang
Intellectual 900 points
Other Parts Discussed in Thread: TPA3118D2, TPA3118D2EVM

We used 2 TDA3118D2 ICs to design a 2.1 speaker system.  We did some EMI pre-scan and found serious conducted EMI issues.  When we put both TPA3118D2 ICs in shut-down mode, these conducted EMI noises did go away so the source of such noises was confirmed to be from TPA3118D2.

We pretty much followed TPA3118D2EVM design, except that we use Panasonic EXC-ELDR35V bead instead of inductor.  Both TPA3118D2 ICs operate in BD Modulation mode and AM2, AM1, AM0 are set to 0,0,0 so the switching frequency is around 400K Hz.

Attached are our audio amp design and the conducted EMI measurement.   Any help will be highly appreciated.

  • 0
    TI__Expert 7925 points

    Hello, Chi-wen. Sorry for the delay in my reply. I have a few ideas you might try to reduce this conducted noise. Unfortunately, ferrite bead filters open up a can of worms when it comes to EMI considerations. The ferrite bead filter on the audio outputs causes ringing on the PWM edges in the 1 MHz to 30 MHz range, which is bad for conducted noise pollution.

    You may try

    1. decreasing the size of the ferrite beads to reduce their ringing.
    2. reducing the size of the filter caps from 1 nF to 470 pF or less.
    3. either eliminating the C–RC snubbers following your ferrite beads, or changing the filter to something like an 18 Ω/330 pF design or smaller on the caps. The snubbers could be dumping some noise into ground.
    4. adding a ferrite choke between the power supply and the PVCC input to the amp.
    5. running your speaker wires through a torroid to act as a choke along the output wire.
    6. adding more ground vias on your board between ground planes (assuming it's a two-layer board).

    Please let me know if you have any success with these ideas.

    Good luck!
    Matt

  • 0 in reply to Matt Lauer
    Intellectual 900 points

    Hi Matt,

      Thanks a lot for your suggestions.

      Regarding 4. We already have a common mode choke between DC Jack and PVCC.  We have tried to increase the inductance of this common mode choke and it didn't make any difference.  Do you suggest to have another ferrite choke near the PVCC supplier to TPA3118D2?

      Regarding 5. We did try that but it didn't help.

      Another interesting find is that we disconnected speaker wires from all three channels (Left, Right and Woofer) and the noise dropped totally because the load was removed from the system.  When we connected to woofer driver, the noise was pretty load.  However, when we connected either Left or Right driver, the noise went up to the roof.  It seems to us that the circuitry for left and right drivers generates most of conducted noise.  We haven't figured out the reason for this yet.

      We will try your suggestions and let you know if we make any progress.

  • 0 in reply to Matt Lauer
    Intellectual 900 points

    Hi Matt,

      We have done several experiments.  However, the measurement of conducted EMI noises doesn't make much differences.

    Experiment 1: C-RC filter is change from 3.3 Ω/1 nF to 18 Ω/150 pF.  Here is the scope capture of the output.

    Experiment 2: C-RCsnubber is totally removed

    Experiment 3: C-RC filter is change from 3.3 Ω/1 nF to 18 Ω/680 pF.  Here is the scope capture of the output.

    BTW, our current design sets the switching frequency to 400K Hz.  (AM0=0, AM1=0, AM2=0.)

    Here are a few more questions:

    1. If w change to a higher switch frequency, will that help?

    2. If we just connected to the woofer speaker output, which uses a single TPA3118D2 in parallel BTL mode.  If we only connect woofer output, there is much less noises compared to left and right channels in BTL mode.  Does this make any sense to you?

    3. Even though removing C-RC snubber hasn't help in reducing conducted EMI noises, it does reduce the heat dissipation from PCBA.  Is this because of increasing R value of the filter?

  • 0 in reply to Chi-wen Wang
    TI__Guru**** 189399 points

    Hi, Chi-wen,

    I was looking at this issue, and it seems like you have a really big spike on the switching waveform. This is usually due to poor high-frequency decoupling of the power supplies and/or layout in general.

    You might want to grab one of our EVMs and see how it compares to your PCB layout.

    Unfortunately, most of our customers power our amps with SMPSes, so we typically don't get a lot of questions about conducted EMC.

    -d2

  • 0 in reply to Don Dapkus
    Intellectual 900 points

    Hi Matt & Don,

      Thanks a lot for your suggestions.  We ended up asking the AC adapter supplier to use a bigger inductor and filter out more noises.  This is the easiest solution for us.