I'm trying to solve this exercise:

Consider a High-Pass filter with a cutoff frequency of $W=2\pi$

Calculate the output signal if the input signal is $x(t)=\displaystyle \frac{\sin(4\pi t)}{\pi t}$

The fourier transform for the input signal is:

$X(\omega)= \left\{ \begin{array}{lcc} 1 & \text{if} & -4\pi \lt \omega \lt 4\pi \\[2ex] 0 & \text{otherwis}e \\ \end{array} \right.$

Now, the output signal is : $Y(\omega)=H(\omega)X(\omega)$

$Y(\omega)= \left\{ \begin{array}{lcc} 1 & \text{if} & 2\pi \le |\omega| \le 4\pi \\[2ex]0 & \text{otherwise} \\ \end{array} \right.$

I can get here, but the solution to the exercise converts this function with inverse fourier transform, giving the result of $x(t)=2\displaystyle \frac{\sin(\pi t)}{\pi t}\cos(3\pi t)$.

I don't know how did they get to that conclusion, I have tried to search for it, but i don't know how to look for that.

I hope that the question is clear. Sorry if I'm lacking some details, it's my first time asking this type of questions.


2 Answers 2


What you get as $x(\omega)$ is two $\mathrm{sinc}(\pi t) = \displaystyle\frac{\sin(\pi t)}{\pi t}$ shifted by $+\omega_0$ and $-\omega_0$ in frequency ($\omega_0 = 3 \pi$). By the frequency shift property of the FT:

$$ \mathcal F\left\{x(t) \cdot e^{i\omega_0 t}\right\} = X\left(e^{j(\omega - \omega_0)}\right) $$

Then, remember $e^{i\omega t} = \cos(\omega t) + i \cdot \sin(\omega t)$. Therefore:

$$ x(\omega) = \sqcap_{2\pi} (\omega - \omega_0) + \sqcap_{2\pi} (\omega + \omega_0) $$

Doing inverse: $$ \begin{align} x(t) &= \mathcal F^{-1}\left\{x(\omega)\right\} \\ &=\mathrm{sinc} (\pi t) \cdot \big(\cos(\omega_0 t) + i \cdot \sin(\omega_0 t)\big) + \mathrm{sinc} (\pi t) \cdot \big(\cos(-\omega_0 t) + i \cdot \sin(-\omega_0 t)\big) \\ &= 2 \cdot \mathrm{sinc} (\pi t)\cdot \cos(\omega_0 t) \\ &= 2 \cdot \frac{\sin(\pi t)}{\pi t}\cdot \cos(3 \pi t) \end{align} $$

since $\cos(-x)=\cos(x)$ and $\sin(-x)=-\sin(x)$


So you have to take the inverse Fourier transform of the output of the highpass filter which you have correctly defined. Instead of directly computing the result $y(t)$ by evaluating the inverse Fourier transform integral; $$y(t) = \frac{1}{2\pi} \int_{2\pi < \lvert \omega\rvert <4\pi} {1 e^{j\omega t} d\omega }$$ you can also use the linearity property of the Fourier transform to directly write the answer, without explicitly evaluating the integral.

More specifically, you can see by inspection of $Y(\omega)$ that $$ Y(\omega) = Y_{4\pi}(\omega) - Y_{2\pi}(\omega),$$ where $Y_{4\pi}(\omega)$ corresponds to the frequency response of an ideal lowpass filter with cutoff frequency $W_1 = 4\pi$ radians per second, and $Y_{2\pi}(\omega)$ corresponds to the same type of filter with a cutoff frequency $W_2 = 2\pi$ radians per second. Since you already know the inverse Fourier transform of the ideal zero phase lowpass filter with a cutoff frequency $W$ radians per second and a gain of unity as given by: $h_{W}(t) = \displaystyle \frac{\sin(W t)}{\pi t}$, therefore you conclude; $$ y(t) = h_{4\pi}(t) - h_{2\pi}(t) = \frac{\sin(4\pi t)}{\pi t} -\frac{\sin(2\pi t)}{\pi t} $$ What remains is trigonometric simplification; (At this point I've to note that from a number of possible trigonometric simplifications (which are all equivalent as a correct answer) I'll choose the one that yields the provided answer, and how do I decide on that is by inspection of the provided answer using educated guess again):

\begin{align} y(t) &= \frac{1}{\pi t} \big( {\sin(4\pi t) - \sin(2\pi t) } \big)\\ &= \frac{1}{\pi t}\big( {\sin(3\pi t + \pi t) - \sin(3\pi t -\pi t) }\big)\\ &= \frac{1}{\pi t}\bigg( {\sin(3\pi t) \cos(\pi t) + \cos(3\pi t) \sin(\pi t) - \big[ \sin(3\pi t) \cos(\pi t) - \cos(3\pi t) \sin(\pi t)\big]}\bigg)\\ &= \frac{1}{\pi t}\big( {\sin(3\pi t) \cos(\pi t) + \cos(3\pi t) \sin(\pi t) - \sin(3\pi t) \cos(\pi t) + \cos(3\pi t) \sin(\pi t) }\big)\\ & = \frac{1}{\pi t} {2\cos(3\pi t) \sin(\pi t) }\\ &= 2\frac{\sin(\pi t)}{\pi t} {2\cos(3\pi t) } \end{align}

Note that another perspective in the inspection of $Y(\omega)$ may result in the conlcusion as @oxuf did, and that interpretation can provide a more simpler and elegant procedure of obtaining the same result, I belive.


Your Answer

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

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