• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 30 subscribers
  • Views 301 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.

PCM5100A: Distortion

Kuramochi Tadahiko
Guru 28715 points
Part Number: PCM5100A


Hi,

My customer is evaluating PCM5100A.

The output signal is distorted at a higher frequency as attached.

PCM5100A distortion.xlsx

Is this waveform valid?

If no, what should we check?

Best Regards

Kuramochi

  • 0
    TI__Guru 54923 points

    Hi Kuramochi-san,

    This looks normal to me.  The digital interpolation filter brings the sampling to 384ksps, so when you have a 20kHz tone you would see about 19 samples per period.  That looks like what you are seeing.

    Thanks,

    Paul

  • 0 in reply to Paul_Frost
    Guru 28715 points

    Paul-san,

    Thank you for your support.

    >The digital interpolation filter brings the sampling to 384ksps, so when you have a 20kHz tone you would see about 19 samples per period.

    I'm very sorry to trouble you, but could you explain to me why we see 19 samples per period by the digital interpolation filter of 384ksps?

    And, is an adjustment of the external output LPF the best solution for this distortion?

    Best Regards,

    Kuramochi

  • 0 in reply to Kuramochi Tadahiko
    TI__Guru 54923 points

    Hi Kuramochi-san,

    The device features a digital interpolation filter that will upsample the input data to a higher sample rate.  For example, if you are using a 48kSPS input to generate a 10kHz sine wave, then the raw digital data would have 48kSPS/10kHz = 4.8 samples per period of the sine wave.  If you were to directly graph that, it would not look alike a sine wave at all, it would look more like a square wave. 

    The interpolation filter will upsample the 48kSPS to 384kSPS, so now you would have 384kSPS/10kHz = 38.4 samples per period.  If you graphed that you would see something that looks much closer to a sinewave.

    When you have a 20kHz sine wave, it would be 384kSPS/20kHz = 19.2 samples per period.  

    The behavior you are seeing is still not a problem.  The first order filter behavior you are seeing is still attenuating the harmonics on the output, which are already well beyond the audible range.  If you measure the THD+N at this condition, you should still see good results.

    Thanks,

    Paul