• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 29 subscribers
  • Views 384 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.

PCM1681: Out of band noise at Fs

Jamie Robertson
Prodigy 10 points
Part Number: PCM1681

I have an audio player design up and running with the PCM1681. I'm running at 44.1kHz sampling rate with an 11.2896MHz master clock and using wide mode. Due to space and cost requirements, I'm attempting to use a passive RC low pass filter on the outputs with a cut-off frequency of around 28kHz. Noise floor and harmonic distortion in the audio pass band is excellent.

My test setup is that I'm playing audio and recording and/or analyzing  the output of the PCM1681 into a computer audio interface sampling at 96kHz. Below are two spectral plots of a full scale 1kHz sine wave. My questions is regarding the large out of band spikes that can be seen centered around Fs. These are present to some degree whenever any audio is playing, and are not present when feeding the DAC silence. The effect of my passive LPF is very small - when I remove the LPF these spikes are perhaps 3 or 4 dB higher. I've tried both sharp and slow roll-off for the internal filter with no noticeable effect.

(I see the same thing using multiple recording/analysis applications, including REW.)

My question is whether or not this is to be expected. I know that the datasheet recommends active filters due to out of band noise, but this seems higher than I was expecting. Is there perhaps something wrong with my testing methodology? If this is normal, can they really be reduced significantly by a 2-pole active LPF?

  • 0
    Prodigy 10 points

    In case it wasn't obvious, by "centered around Fs", I'm referring to the spikes at the far left of each plot at around 44kHz. Thanks.

  • +1 in reply to Jamie Robertson
    TI__Mastermind 25820 points

    Hello Jamie, As you already know when you have a sampling freq at Fs with input frequency of Fin, you create  aliases at Fs-Fn and Fs+Fin.  Looking at   plots you get for both 1KHz and 438Hz, this can be seen at the far right. These should be easily filter out with a LPF. As you increase the order of the filter, or use a passive filter with sharper roll off , more unwanted signals will be filter out.  These aliases  are out of band, since  you are filtering out anything beyond  28kHz ( assuming sharp role off). Make sure you are plotting  after the filter , not before the filter.

    in case you are interested in reading more about aliasing, here is a link for an article. 

    https://www.soundonsound.com/sound-advice/q-what-aliasing-and-what-causes-it

    Regards,

    Arash

  • 0 in reply to Arash Loloee
    Prodigy 10 points

    Thanks for that refresher on sampling theory. I guess I was puzzled for several reason; I assumed incorrectly that the DAC's internal filtering did more for this problem than it evidently does. Secondly, when I play broadband music rather than a sine wave, the aliasing still presents as defined spikes around Fs. I'll work on a better LPF filter.