I’m given a signal $x(t)$, it's convolved with $h(t)$ and sampled at rate T=1. The result is called $\tilde{x}[n]$.

For $$h(t) = \begin{cases} 1 & -0.5<t\le 0.5 \\ 0 & \text{else} \end{cases},$$ find $\tilde{X}^f(\theta)$ as an expression of $X^f(\theta)$.

My attempt

I noticed that I can write for a sampled signal $$\tilde{X}^f(\theta) = \frac{1}{T} \sum_{k=-\infty}^{\infty} \tilde{X}^F\left(\omega = \frac{(\theta - 2\pi k)}{T}\right).$$

All that I had left to do was simplify using the convolution to multiplication relation and this identity:

$$\begin{cases} k & |t|<t_0 \\ 0 & \text{else} \end{cases} \to 2kt_0 \text{sinc}(\omega t_0),$$ I substituted $\tilde{X}^F(\omega)$ into the expression.


$$\sum_{k=-\infty}^{\infty} X^F(\theta - 2\pi k)\operatorname*{sinc}(0.5(\theta-2\pi k))$$

whereas what they got: $$\sum_{k=-\infty}^{\infty} X^F(\theta - 2\pi k)\operatorname*{sinc}\left(\frac{\theta-2\pi k}{2\pi}\right)$$

I'm losing my mind over the $\pi$ in the denominator, why is it there?

  • 1
    $\begingroup$ I have no idea what this notation means. What is "$X^f(\theta)$"? What is $\theta$? Is $X(\cdot)$ being raised to the $f$ power? Why not just use standard electrical engineering notation? $\endgroup$ Commented Apr 20 at 21:35
  • $\begingroup$ this is the notation I was taught for DTFT, $$X^f(\theta) = \sum_{-\infty}^{\infty}x[n] e^{-j\theta n}$$ $\endgroup$ Commented Apr 21 at 11:51

1 Answer 1


It is hard to follow what you did as your nomenclature is inconsistent and you didn't explicitly define any of it.

I believe the bit you got wrong was the Fourier transform of $h\left[t\right]$, with the correct expression for the more general case of a rectangular pulse of height $k$ centred on $t=0$ with width $t_0$ being: $$\frac{k}{t_0}{\rm sinc}\left[\frac{\omega}{2\pi t_0}\right]$$ Where $\omega$ is the frequency (radians per unit time).

Here is a full solution assuming that $X^f$ is the Fourier transform of the signal prior to convolution, $\bar X^f$ is the final discrete time Fourier transform, and $\theta$ is the frequency (radians per unit time):

Firstly, the Fourier transform of $h(t)$ is: $$H\left[f\right]={\rm sinc}\left[f\right]$$ Where $f$ is the frequency (cycles per unit time).

Secondly, using the Poisson summation formula and that convolution in the time domain is multiplication in the transform domain: $$\bar X^f\left[2\pi f\right]=\sum^\infty_{k=-\infty}X^f\left[2\pi\left(f-\frac{k}{T}\right)\right]{\rm sinc}\left[f-\frac{k}{T}\right]\\ =\sum^\infty_{k=-\infty}X^f\left[2\pi\left(f-k\right)\right]{\rm sinc}\left[f-k\right]$$ Where I have used that the sampling interval, $T$, is equal to $1$, and $\theta=2\pi f$.

Finally, rewriting in terms of $\theta$: $$\bar X^f\left[\theta\right]=\sum^\infty_{k=-\infty}X^f\left[\theta-2\pi k\right]{\rm sinc}\left[\frac{\theta}{2\pi}-k\right]\\ =\sum^\infty_{k=-\infty}X^f\left[\theta-2\pi k\right]{\rm sinc}\left[\frac{\theta-2\pi k}{2\pi}\right]$$

This matches the correct expression that you were given.


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.