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.

multiplexer sampling anti aliasing filter

Other Parts Discussed in Thread: ADS131A04

Hi,

Recently I'm working on a project and I used a mux before one channel ADC.There is some problem with anti aliasing filter. For example, sigma delta adc has modulate frequency Fd, Data rate Frequency Fdr and my signal bandwidth is Fc. Max bandwidth of anti aliasing is Fd/2 according to Nyquist sample rate and this is the problem bother me.In multiplexer existence,I think the max anti aliasing bandwidth is Fd/2N because for each channel it's like down sampling every N samples. It's also like for each channel the sample rate is Fd/N except that the sampling rate is Fd in 1/N cycle sampling period and 0 in the other (N-1/)N period. Am I right?

Best Regards!

  • Hi There!

    Welcome to our e2e forum!  If you have a specific part number in mind, please do let us know.  In the meantime, this article has some anti-aliasing basic tips for you.

  • Hi Hendrick!

    I am using ADS131A04 but only using one channel.I put an anti aliasing filter between the ADC and multiplexer for low cost purpose, otherwise I will have multi signal condition chains. This article does not cover the case of multiplexer and I think it may be different from the traditional one channel sampling plan because of the irregular sampling for each channel. The modulator frequecy I chose is 4.096M Hz so the max anti aliasing bandwidth is 2.048M Hz.There are 9 channel in my signal chain and the throughput samples is 2000 for each channel so the channel changing frequecy is 2000*9 = 18kHz, adc output data rate is 128k and we assume we will always get valid data sample of last data in one channel sampling perid . For the purpose to reach 0.01% error in analog signal there should be 9.2RC time for anti aliasing. Total settling time should be lower than 1/18K =55us. The settling time before the anti aliasing filter is about 20us, therefore the settling time of anti aliasing filter should be lower than 35us. So RC time is 35us/9.2 = 3.8us and the bandwidth of that RC anti aliasing filter is 41k. This filter has -34db attenuation at 2.048M Hz.The problem is that whether for each channel modulator frequecy is down to 4.096M/9 = 455K or it is still 4.096M? Even if the modulator frequecy is equivalent down to 455K, can it be treat as normal sampling not a irregular one?

    Best Regards!

     

  • Hello,

    Are you considering changing the MUX after each ADC sample, or are you collecting 2000 samples from one channel and then switching the MUX to another input?

    If you are holding the MUX constant for 2000 samples before switching, I believe it makes sense to design your anti-aliasing filter according to the ADC's output data rate and modulator sampling frequency, ignoring the MUX. The 2000 samples that you collect will be output from the ADS131A04 at the configured data rate (fDR). The corner frequency of the anti-aliasing filter should be set about one decade beyond fDR/2 (the data rate Nyquist frequency). This should give you enough attenuation at the modulator sampling frequency (4.096 MHz) to reduce out-of-band noise.

    If you plan to switch the MUX after each ADC sample, you must keep in mind that there is a required settling time through the digital filter. This is due to the filter's inherent response to sudden input transients and requires some deterministic amount of time for the output data to be considered settled.

    We also have additional online training on this subject found here: https://training.ti.com/delta-sigma-adcs-aliasing

    Best regards,

  • Hi,

     I'm changing the MUX after each ADC sample.The data sheet shows digital filter settling time is 3 times data rate period which is 23us and that is short enough for my system. The only problem is still whether I should design anti aliasing filter by modulator sampling frequency(fD) or modulator sampling frequency over N( fD/N, N is the channel number).

    Best Regards!

  • Hi,

    You would base the design of the anti-aliasing filter on the modulator frequency. The ADC is sampling the input at the modulator frequency. The ADCs are independent, and N (the channel number) doesn't play into the calculation. 

    Incidentally, we've basically recommended using a filter bandwidth of 10x of the data rate. This gives a wide enough bandwidth so it doesn't give a large dropoff for the normal frequency response of the device. 


    Joseph Wu