PDA

View Full Version : Fourier transform of a modified impulse train


naimad
Jul26-07, 03:57 PM
I need to find the Fourier Transform (FT) of:

x(t)=\sum^{\infty}_{n=-\infty}((-1)^{n}\delta(t-nT))

Not really sure how to solve this problem, so any help will be appreciated.

Also, if you guys know a good reference for non-uniform sampling and reconstruction, please post it.

WolfOfTheSteps
Jul26-07, 08:35 PM
I need to find the Fourier Transform (FT) of:
x(t)=\sum^{\infty}_{n=-\infty}((-1)^{n}\delta(t-nT))



Note: I think my answer below is correct, but I've never found a way to check my answers when finding the FT.... So, if someone with more experience could verify this, I would appreciate it!

Notice first that all the impulses at odd values of n are being subtracted, while all the impulses at even values of n are being added. This suggests that we write x(t) as follows:

x(t)=\sum^{\infty}_{n=-\infty}\delta(t-2nT) - \sum^{\infty}_{n=-\infty}\delta(t-(2n+1)T)

=\sum^{\infty}_{n=-\infty}\delta(t-2nT) - \sum^{\infty}_{n=-\infty}\delta(t-2nT-T)

Now let

\hat{x}(t) = \sum^{\infty}_{n=-\infty}\delta(t-2nT)

This is a standard impulse train with period 2T. From any table of basic Fourier Transforms:

\hat{X}(j\omega) = \frac{2\pi}{2T}\sum^{\infty}_{n=-\infty}\delta\left(\omega-\frac{2\pi n}{2T}\right) = \frac{\pi}{T}\sum^{\infty}_{n=-\infty}\delta\left(\omega-\frac{\pi n}{T}\right)

Now notice that x(t), above, can be written as:

x(t)=\hat{x}(t)-\hat{x}(t-T)[/itex]

We already know the fourier transform for \hat{x}(t); now we just need to use the time shifting property to find the transform for \hat{x}(t-T):

[tex]FT \{\hat{x}(t-T)\} = e^{-j\omega T}\hat{X}(j\omega)

Now, by linearity:

X(j\omega)=\hat{X}(j\omega)-e^{-j\omega T}\hat{X}(j\omega)
=\hat{X}(j\omega)[1-e^{-j\omega T}]

And since we are dealing with an impulse train, the only values of \omega we have to deal with are those at:

\omega=\frac{\pi n}{T} (since the impulse will be 0 everywhere else).

From this we can write:

e^{-j\omega T}=e^{-j\pi n }=(-1)^n

So our final answer will be:


X(j\omega)=\frac{\pi}{T}\sum^{\infty}_{n=-\infty}\delta\left(\omega-\frac{\pi n}{T}\right)[1-(-1)^n]

naimad
Jul27-07, 02:16 PM
Thanks WolfOfTheSteps, I checked it out with a friend and it seems to be correct.

Still, if anybody knows a good reference for non-uniform sampling and reconstruction, please post it.