I have found an exercise on a signal processing book that asks to compute the Fourier series of a function by using its Fourier Transform, let:

$$ x(t) = \sum_{n=-\infty}^{\infty} \Lambda \left( \dfrac{t-nT}{T/2} \right) $$

I did compute the FT of this function, I used that a trianglePulse can be decomposed into two convoluted rectangularPulse. It led to:

$$ x(t) = \Pi \left( \dfrac{t}{T/2} \right) \star \Pi \left( \dfrac{t}{T/2} \right) \star \sum_{n=-\infty}^{\infty} \delta (t-nT) $$

Thus, the Fourier transform is:

$$X(\omega)= \dfrac{4}{\omega^2} sin^2 \left( \dfrac{\omega T}{4} \right)\cdot \dfrac{2\pi}{T} \sum_{n=-\infty}^{\infty} \delta \left( \omega - n\dfrac{2\pi}{T} \right) $$

Where: $$FT\left[\sum_{n=-\infty}^{\infty} \delta (t-nT) \right] = \sum_{n=-\infty}^{\infty} e^{-jnT\omega} = \dfrac{2\pi}{T} \sum_{n=-\infty}^{\infty} \delta \left( \omega - n\dfrac{2\pi}{T} \right)$$

I know that FT is a generalization of the Fourier Series. But I don't know how to continue in order to find the Fourier series. What should I try to do to solve the exercise? (I just need to know how to continue, it's not necessary that you do it for me).

  • $\begingroup$ There should be some complex exponential terms in $X(\omega)$ coming from the fourier transform. Real and imaginary parts combine to give cosine and sine terms. $\endgroup$ – Paul May 6 '15 at 15:32
  • $\begingroup$ Please, check my edit. I obtained a complex exponential. Should I decompose it into $\sin$ and $\cos$ ? Thank you $\endgroup$ – Airman01 May 6 '15 at 15:40
  • $\begingroup$ This seems correct :) $\endgroup$ – krvolok May 6 '15 at 15:43

Once you have obtained the Fourier Transform:

\begin{equation} \mathcal{F} \big\{ x(t)\big\} = X(\omega) = \dfrac{4}{\omega^2} \sin^2 \left( \dfrac{\omega T}{4} \right)\cdot \dfrac{2\pi}{T} \sum_{n=-\infty}^{\infty} \delta \left( \omega - n\dfrac{2\pi}{T} \right) \end{equation}

You can take into consideration that for any function $b(\omega)$ and a certain pulse delay $\omega_0$ you have:

\begin{equation} b(\omega) · \delta(\omega-\omega_0) = b(\omega_0)\delta(\omega-\omega_0) \tag{1} \end{equation}

Substituting $\omega_n = n\dfrac{2\pi}{T}$ we can rewrite you Fourier transform as following:

\begin{equation} X(\omega) = \dfrac{4}{\omega^2} \sin^2 \left( \dfrac{\omega T}{4} \right)\cdot \sum_{n=-\infty}^{\infty} \dfrac{\omega_n}{n} · \delta \left( \omega - \omega_n \right) \end{equation}

Now considering (1):

\begin{equation} X(\omega) = \sum_{n=-\infty}^{\infty} \dfrac{4}{\omega_n^2} ·\sin^2\Big(\dfrac{\omega_nT}{4}\Big)·\dfrac{\omega_n}{n}·\delta \left( \omega - \omega_n \right) \end{equation}

Working it out we finally obtain:

\begin{equation} X(\omega) = \sum_{n=-\infty}^{\infty} \dfrac{2T}{n^2\pi} ·\sin^2\Big(\dfrac{n\pi}{2}\Big)·\delta \left( \omega - n\dfrac{2\pi}{T} \right) = \sum_{n=-\infty}^{\infty} \alpha_n·\delta \left( \omega - n\dfrac{2\pi}{T} \right) \end{equation}

Where $\alpha_n = \dfrac{2T}{n^2\pi} ·\sin^2\Big(\dfrac{n\pi}{2}\Big)$ denotes the amplitude for each sample $n$.

  • $\begingroup$ Thanks @luikn! It was helpful and now I could check that the exercise result was correct. I would vote your answer, but I can't yet. Thanks $\endgroup$ – Airman01 May 6 '15 at 18:59
  • $\begingroup$ Don't worry, we are here to help each other :) @Airman01 $\endgroup$ – lucasrodesg May 6 '15 at 21:25

If you have a $T$-periodic function $x(t)$, its Fourier series is given by

$$x(t)=\sum_{k=-\infty}^{\infty}c_ke^{j2\pi kt/T}\tag{1}$$

where $c_k$ are the complex Fourier coefficients. Taking the Fourier transform of (1) gives

$$X(j\omega)=2\pi \sum_{k=-\infty}^{\infty}c_k\delta\left(\omega-\frac{2\pi k}{T}\right)\tag{2}$$

So, given the Fourier transform (2), you can directly read off the Fourier coefficients $c_k$.

  • $\begingroup$ Wow alright. So does the problem reduces to saying: $ c_k = \dfrac{4}{T\omega^2} \sin^{2} \left( \dfrac{\omega T}{4} \right) $ ? (Being that the coefficient in the Fourier Series) $\endgroup$ – Airman01 May 6 '15 at 16:22
  • $\begingroup$ @Airman01: Yes, with $\omega=2\pi k/T$ because $f(\omega)\delta(\omega-\omega_0)=f(\omega_0)\delta(\omega-\omega_0)$. $\endgroup$ – Matt L. May 6 '15 at 20:18

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.