• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 46 subscribers
  • Views 656 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.

Sigma Delta, what happen to Data output when input frequency/2 is close to sample rate (ADS1274)

richard payne
Genius 4640 points
Other Parts Discussed in Thread: ADS1274

This query relates to fundamental Sigma Delta as I have read several material and could not get answer to determine the sample rate setting for signal detection.

I have source sigmal between 1Khz and 4Khz from the vibration sensor, I'm looking into Sigma Delta as alternative to SAR (ADS8631) because of digital filter feature which removes upper frequency (there will be 1-2 pole low pass analogue filter before the ADC, which set to f3dB=4Khz).

I plan to set ADS1274 to high resolution mode so it over sample 128 time the data rate. I'm actually only interested 16 bits rather than 24 bits,

I could set the data rate to 8KSPS or 12KSPS or 16KSPS and so on. The constraint that the DSC (320F28835) being a real time processing speed so lower sample rate give extra overhead in DSC processing (ie FIR and FFT). NB: I cannot use other higher performance DSC (which most are BGA) 

One issue concern me that in SAR, obviously with data rate of 8KSPS

  • test signal-A has 4Khz sinewave with 0 phase (0ᵒ)  ,      I expects =/-Vpp output data stream
  • test signal-B has 4Khz sinwave with π/2 phase 45ᵒ)       I expects +/-0.71*Vpp output data stream
  • test signal-C has 4Khz sinwave with π/2 phase (90ᵒ)     I expect 0Vpp output data stream

the question that I unsure how SD behave differently from SAR,

I read material that say because of over-sampling, it does not care about phase shift of sinewave which lead to zero output as in SAR. 

Can you shown (or demonstrate) what the data stream look like or FFT look like when input signal is slighly less than data rate / 2.

 

 

  • 0
    TI__Mastermind 23275 points
    Hi Richard,

    I just realized we also owe you an answer to this question. I apologize for the delay. I am not sure though I can sufficiently answer your question.

    First of all I wanted to note that you should set the 3-dB frequency of your analog low-pass filter higher than 4kHz if you want to measure a 4kHz input signal without any attenuation.
    Also you should set the data rate higher than 8kSPS. When running at 8kSPS the digital filter of the ADS1274 will attenuate a 4kHz signal by 5dB to 6dB already (see figure 64 in the datasheet).

    Are you applying all three test signals at the same time to the ADC? If so, then I would expect the output code to show a result which is equal to the superposition of the three signals.
    In case you apply the signals individually then I would expect the output to read the same result for each signal. Means a 4kHz signal with amplitude of +/-Vpp.

    Regards,
  • 0 in reply to Joachim Wuerker
    Genius 4640 points
    Thank for reply and happy new year

    Consider the I/P signal is 4KHz and sample rate is 8KHz where it sample peak to peak sinewave, however if you phase shift the I/P signal, it can sample DC to DC sinewave, especially if 4KHz waveform could be 3.95KHz. This is a point I was trying to get across, do you catch my drift?

    By having sample rate of 16Khz or 24Khz with more capture point per sinewave, it can capture I/P waveform regardless of phase, although max peak to peak remains dependent on phase where it capture that peak sinewave.

    I have done experiment on AD7732 and found x2 sample rules rather strange but various textbox and app note, this is poor solution since it not immune to signal phase shift where it no longer capture peak to peak waveform. I'm tad surprised this topic is not fully discussed or made clear in collage or high school.

    R.
  • 0 in reply to richard payne
    TI__Guru 66506 points

    Hi Richard,

    Happy New Year to you as well.

    Going through your initial post, I believe we are misunderstanding your concern with the phase of your input signal. Can you explain again how test signals A, B, and C are being measured in your system? Are they singled-ended or differential? Are they being measured individually, or is one signal driven to INP and another to INN for the same ADC channel?

    Unlike some SAR ADCs that multiplex different input signals to a single converter, the ADS1274 has 4 simultaneously-sampled differential channels. As Joachim mentioned, this allows you to measure each signal individually without altering the signal amplitude. Measuring the signals together somehow would yield a superimposed result. The maximum peak-to-peak sine wave that the ADS1274 can measure will not be related to phase - rather, it is only dependent on the signal common-mode and amplitude relative to the ADC reference voltage, VREF.

    To ensure that the ADS1274 digital filter does not attenuate your input signal, make sure that you are sampling fast enough such that 0.453 * fDATA is greater than your bandwidth of interest.

    Did this answer your question?

    Best Regards,