• State
  • Locked Locked
  • Replies 4 replies
  • Subscribers 30 subscribers
  • Views 458 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.

MSPM0G3507: ADC HW Averaging, Oversampling Understanding

Stefano Andrenacci
Prodigy 40 points
Part Number: MSPM0G3507

Tool/software:

Hello,

I'm trying to understand ADC HW Averaging of MSPM0G3507 a little better.

I'm trying to sample a signal at 20kHz frequency: in this case the Nyquist frequency should be 10kHz.

This signal is conditioned by an analog front end stage with an antialiasing filter.

If I implement hardware averaging with 16x oversample, am I actually sampling the signal at 20kHz x16 = 320kHz?

So, if I need at least 60dB of alias attenuation at Nyquist frequency, can I realize the low pass anti aliasing filter so that I have -60dB at 160kHz? Or do I need to have -60dB at 10kHz?

Also, after oversampling, MSPM0G3507 allows to perform averaging: I'm not sure if I understand how decimation is actually a pass-bass filter in this context, and what will be the cutoff frequency of such filter.

Thanks in advance for your replies,

Stefano

  • 0
    Prodigy 40 points

    Sorry, maybe I didn't give a proper explanation: I am using a timer at 20kHz to trigger ADC sampling & conversion of a channel. So I have a sampling time of 50us.

    Then I implemented ADC HW Averaging 16x: from my understanding the signal is sampled 16 times and each sample is summed into an accumulator. The result is then averaged.

    This "oversampling" happen in a much shorter time than my sampling time of 50us: is it ok to consider this as oversampling, giving a 20kHz*15 = 320kHz sampling frequency? If this is true I can relax my analog antialiasing filter.

    Thanks again,

    Stefano

  • 0 in reply to Stefano Andrenacci
    TI__Genius 9810 points

    Actually the normal sampling is happening in a much shorter time than 50us. It may cost only a few us, for example 2us. Then the rest 48us ADC does not sample at all.

  • 0 in reply to Yuhao Zhao
    Prodigy 40 points

    Yep, exactly, what I was wondering is what's the Nyquist Frequency in this case. I fear it's still 10kHz (half of 20khz), considering that hw averaging should be represented, in time domain, as a train of impulses close to the 20kHz sampling frequency. In frequency domain it will also be a train of impulses though, so I'm not sure if it operatively change anything for the antialiasing filter. Do you think that's the case?

  • 0 in reply to Stefano Andrenacci
    TI__Genius 9810 points

    Yes I think so. It should be still 10 kHz.