0
$\begingroup$

I am trying to take a signal $F(t)$ that has been sampled at some time DelT, I then wish to pass this signal through a channel $H(s)$. To do this I am sampling my signal $H(s)$ at DelT time intervals then finding the Z-transform $H(z)$. How would I go about applying this in matlab?

Currently I am using c2d to convert s-domain to z-transform for my channel but am unsure how I can go about the convolution with my sampled signal which is just an array of numbers. I have tried to take the fft of sampled $F(t)$ then use freqz of $H(z)$. To begin with I am just trying to apply a simple time delay $e^{-st}$ to the signal.

$\endgroup$
  • 1
    $\begingroup$ Do you understand the physical meaning of $z$ (or $z^{-1}$) in a transfer function in the Z-domain? $\endgroup$ – fibonatic May 14 '16 at 10:32
0
$\begingroup$

There's a lot of confusion in your question. You refer to "a channel $H(s)$", then to "my signal $H(s)$", so for the sake of clarity, let's properly define:

  1. Input signal - continuous time $f(t)$, impulse sampled $f[n]=f(n\Delta T)$, Fourier transform of continuous time signal $F(s)$, Z-transform of impulse sampled signal $F(z)=F(e^{s\Delta T})$.
  2. Continuous time channel - impulse response $h(t)$, transfer function (Laplace transform of impulse response) $H(s)$.
  3. Discrete time channel - discrete impulse response $h[n]$, transfer function (Z-transform of discrete impulse response) $F(z)$.

Note the definitions above follow a convention:

  1. time-domain signals/functions are lower case ($f, h$); frequency-domain signals/functions are upper case ($F, H$).
  2. arguments of continuous signals/functions are enclosed in parentheses $(t)$; arguments of discrete signals/functions are enclosed in brackets $[n]$.

Then, you write:

(...) z-transform for my channel but am unsure how I can go about the convolution with my sampled signal which is just an array of numbers

It seems you do not understand the basics of LTI systems. Note that $H(s)$ and $H(z)$ are frequency-domain and $f(t)$ and $f[n]$ are time-domain.

Convolution makes sense when the two convolved functions are in the same exact domain (continuous/discrete time/frequency), although for the usual purposes the two convolved functions should both be in the continuous/discrete time domain.

If you're dealing with a discrete-time signal ("signal which is just an array of numbers"), then you might want to convolve $f[n]$ with $h[n]$. If you already have $H(z)$, then an inverse Z-transform on $H(z)$ would produce $h[n]$.

$\endgroup$

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.