• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 47 subscribers
  • Views 645 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.

ADS1675 Anti-Alias Filtering

John Foster
Prodigy 30 points

Other Parts Discussed in Thread: ADS1675

I have a good understanding of anti-alias filtering for SAR converters.  For the ADS1675, I am uncertain.

I need X dB of alias rejection at 100kHz when operating ADS1675 at 32MHz with 32:1 decimation (1MSPS).  Do I need X dB of attenuation at 400kHz (as would for a SAR converter at 1MSPS), or at 32MHz - 100kHz, or somewhere in between?  In other words, does the sigma-delta actually relax my filtering requirements?

  • 0
    TI__Mastermind 23235 points

    Yes - one of the advantages of the delta-sigma converters are that due to the over-sampling, the anti-aliasing filter can be relaxed.  The actually sampling is occurring at fMOD (which equals fCLK for this part) vs the actual datarate.  The attenuation level would depend on your requirements and any potential signals that may alias.  Generally,  40dB or higher attenuation is adequate, but you should set this based on your system.

  • 0 in reply to Greg Hupp
    Prodigy 30 points

    Thank you Greg, that is what I thought.

    My application is actually clocking at 20.48MHz, 32:1 decimation, for 640KSPS data rate.  I need 130 to 140 dB alias protection (yes, I meant 130+ dB) for desired input frequency up to 60 kHz.  My SAR design used cascaded 5 pole filters (buffer amp in between) to give me >140 dB at 580 kHz (sample clock - 60 kHz).  We implemented it and it worked.  So my question is do I design for 140 dB at fMOD - 60 kHz (20.42 MHz for this design, only 4 poles total)?  Or do I need something like 60 dB at 580 kHz (fMOD/32 - 60 kHz).  Desired signal < 60 kHz of 100nV with potential interferer 70 kHz to 40 MHz of 200mV (that's 126 dB).

    The datasheet is unclear (note 1 to tables 4, 5, and 6).  Is the answer the combination of the converter's digital filtering and the analog anti-alias filter?

    Regards

  • 0 in reply to John Foster
    TI__Mastermind 23235 points

    Mr Foster -

    For reference, a quick search pulled this up: http://en.wikipedia.org/wiki/Delta-sigma_modulation#Analog_to_digital_conversion

    Delta-sigma converters are also know as over-sampling converters since the actual sample rate is much faster than the output datarate.  To clarify the previous statement, the fMOD (or modulator frequency) is actually the sampling rate of the converter.  The output datarate is just a decimated version of that frequency.  So for filter purposes (in this case the anit-alias filter), the sampling frequency = the modulator frequency.  For the ADS1675, the modulator frequency is fixed and equal to the input clock frequency.  The decimation ratio determines the actual output datarate.

    Reading the notes in the mentioned tables, it could be a little confusing.  The note is written more to the general statement of Nyquist in that aliasing of signals occur at Fs/2.  To be more correct and clear it should reference fMOD instead of fDATA