We have an acquisition hardware in our lab that acquires and digitizes data from a source generating band-limited signal in the MHz range. It samples the signal at a rate of 4 times the center frequency $f_0$ of the signal. In the documentation, they say that this specific sample rate makes it easier to convert the RF samples to demodulated IQ samples.

How would $f_s=4f_0$ make the demodulation easier? I have been using Hilbert transform to achieve the demodulation (the old-fashioned way). But, if there is any shortcut offered by this specific sample rate, that would be really helpful.


  • $\begingroup$ What is the bandwidth of your signal? $\endgroup$
    – Gillespie
    Jul 5, 2022 at 3:03

1 Answer 1


To get IQ data from a bandlimited signal centered at f0, with a bandpass well inside f0-f0/2 to f0+f0/2, and using a sample rate of fs=4*f0, you can simply keep 2 successive samples out of every 4, and ignore the rest. Take every 4th sample, put them in the I channel, take the samples right after those and put them in the Q channel, and throw away the 3rd and 4th sample out of every 4 samples (I Q ignore ignore I Q ignore ignore I Q ignore ignore ... etc.). The result is an IQ signal at a sample rate of f0.

  • 3
    $\begingroup$ This is a very poor method since it is ignoring half the signal samples. Instead, the ignored samples should be used to enhance the SNR. $\endgroup$ Sep 20, 2019 at 17:54
  • $\begingroup$ @hotpaw2, you mind clarifying the notations in the first and last line regarding center frequency vs cut-off frequency? The $f_0$'s. $\endgroup$
    – Gilles
    Feb 10, 2021 at 12:42
  • $\begingroup$ If the signal is bandlimited to a bandwidth narrower than fs/8, then 3/4ths of the samples can be reconstructed from every 4th (according to Nyquist-Shannon-Kotelnikov), so there is no loss throwing away half the samples. $\endgroup$
    – hotpaw2
    Jul 8, 2022 at 15:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.