• State Resolved
  • Locked Locked
  • Replies 1 reply
  • Subscribers 46 subscribers
  • Views 639 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.

reducing A2D noise

george dorian
Intellectual 940 points
Other Parts Discussed in Thread: ADS1281, ADS1282

Hello all,

I'm not huge in A2Ds and have some Qs about reducing noise.  I’m applying for a job to do care and mantance on seismographs in California (for UCB, if case you’re wondering).  The seismographs are read by (delta sigma I assume) 24bit A2Ds

OK, Signal to Quantization Noise Ratio is: SQNR = 20log10(2^n)

Where n is the number of A2D bits.  So with a 24 bit A2D you will have about 144dB signal above quantization error (one least sig bit).

Now, what happens to the SQNR if you do some averaging?  I think you get 3dB improvement per 8 over samples.  Is that right?

And what happens if you right shift out the least sig bit (or two)?  Do you just get a 23 bit A2D (with about 138dB SQNR) or do you actually get better SQNR?  

Is it true that you still need an analog anti-aliasing filter even if you do it by code?

BTW, I have read some info on suggested in this forum.

thanks george

  • 0
    TI__Guru* 97375 points

    George,


    For more information about some of our seismic ADCs you might look into the ADS1281 and ADS1282. Both datasheets have some information on operation and the internal workings of the ADC. Also each of these devices have an EVM with a user guide. You might want to look at the user guides as well. There's more information about the features and schematics for the evaluation boards.

    As for your question, noise will generally reduce by root 2 of the number of sample if you are averaging. As an example, if you have some data that has a particular noise, the noise will reduce by a factor of 2 for every 4 samples averaged. Noise will reduce by a factor of 4 for every 16 samples averaged.

    However, if you talk about quantization noise, you are generally talking about modeling the error of the quantization, (the error between the signal and the nearest point of quantization) as a noise. Most of the time when we talk about Delta-Sigma ADCs, there is a quantization error for a 1-bit DAC feedback that is oversampled and they use that quantization noise to show the noise shaping through the modulator and the theoretical noise that you might get out of the ADC.

    In your case you talk about the quantization noise for a 24 bit system which would be the error between the signal and the quantization. This is just the error, so you wouldn't think of this as something you could average.

    I'm not sure what you are asking when it comes to bitshifting. The SNR metric usually compares some large signal to the noise floor. When it comes to the best value of SNR, you usually use a full scale input to compare with the noise. Just by bitshifting, I suppose you restrict your full scale input so I would guess you would have less SNR.

    Generally, you need an anti-aliasing filter and cannot remove it through code. If you have high frequency noise that is out of band, this noise is aliased down to a lower frequency. At that point, you can't separate noise from signal.


    Joseph Wu