Say the analog signal $x(t)$ and its spectrum $X_a(\Omega)$. After sampling with frequency $F_s$ or sampling period $T$ we get

$$x[n] \triangleq x(nT) = x\left(\frac{n}{F_s}\right)$$

and its spectrum $X_d(\omega)$.

The relationship between the two spectrums is

$$X_d(\omega) =\frac1T \sum ^{\infty}_{n=-\infty}X_a\left(\frac{\omega + 2\pi n}{T}\right)$$ $$\omega=T\Omega$$

For band-limited analog signal having spectrum in range $[-B,B]$. $X_a(\Omega)=0$ for $|\Omega| \geq B$ if $F_s$ is above Nyquist rate then $$X_d(\omega) =\frac1T X_a(\Omega)$$

For digital signal, its frequency is restrict in $[-\pi,\pi]$ but not the case of analog signal. The equation above means that $X_d(\omega)$ is not restricted in $[-\pi,\pi]$

  • $\begingroup$ The DTFT always gives you something in $[-\pi,\pi]$. It's part of the definition -- it maps a function from $\mathbb{Z}$ to $\mathbb{C}$ to $\mathbb{C}$-valued functions on the unit circle. $\endgroup$ – Batman Dec 16 '15 at 21:21
  • $\begingroup$ Please change the Lower index of summation into "negative infinity" because when I try to do so this robocop says it is less than 6 characters and considers it as a password cracking cyber-terrorist attack on the web site possibly... $\endgroup$ – Fat32 Dec 18 '15 at 0:37

The DTFT $X_d(\omega)$ of a discrete-time signal is periodic with period $2\pi$, so it's sufficient to consider the interval $[-\pi,\pi]$. "Restricted" to this interval means that due to the inherent periodicity there can be no new frequency components outside the interval $[-\pi,\pi]$. Of course, $\pi$ corresponds to half the sampling frequency.

If $X_a(\Omega)$ is the spectrum of a band-limited continuous-time signal, and if the sampling frequency is greater than twice the band-width, then the equality


holds in the interval $[-\pi,\pi]$. Outside that interval, $X_a(\Omega)$ is zero, and $X_d(\omega)$ is continued periodically.

  • $\begingroup$ Sorry because my choice of word. I means if $X_d$ is just scale of $X_a$ then if $X_a$ is defined in $[-B,B]$ and $B>\pi$ then $X_d$ is not periodic in $[-\pi,\pi]$ $\endgroup$ – aukxn Dec 16 '15 at 16:18
  • $\begingroup$ @aukxn: $X_d(\omega)$ is always periodic. The second equation in your question holds in the interval $[-\pi,\pi]$. $\endgroup$ – Matt L. Dec 16 '15 at 16:26
  • $\begingroup$ @MattL. Otherwise said, if the DTFT co-efficients are known then $X_a\left(\dfrac{\omega}{T}\right)$ is the scaled version of DTFT co-efficients $X_d(\omega)$ by a factor $T$ How is mapping done now? From $[-\pi, \pi]$ to which frequency range? $\endgroup$ – jomegaA Feb 10 '20 at 13:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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