0
$\begingroup$

Suppose the CTFT of continuous-time input $x_c(t)$ to an LTI system is $X_c(j\Omega)$ and that of its continuous-time output $y_c(t)$ is $Y_c(j\Omega)$. We have, $$X_c(j\Omega) = 0,\phantom{1}\text{for } |\Omega| \ge \frac{\pi}{T_1}$$.

The LTI system samples the signal (C/D operation at sampling rate $1/T_1$) and passes it through the discrete-time system with response $H(e^{j\omega})$. The result of this DT system is then passed through an ideal D/C system (operating at rate $\frac{1}{T_2} = \frac{1}{2T_1}$) to yield $Y_c(j\Omega)$.

I want to find out the overall frequency response of the system. I can write

$$Y_c(j\Omega) = \begin{cases}\frac{T_2}{T_1} H(e^{j\Omega T_2})X_c\left(j\Omega \frac{T_2}{T_1}\right),\phantom{1}\text{for } |\Omega| < \frac{\pi}{T_2}\\0,\phantom{1}\text{for } |\Omega| \ge \frac{\pi}{T_2}\end{cases}$$

Since I get $X_c\left(j\Omega \frac{T_2}{T_1}\right)$ in the above expression and not $X_c\left(j\Omega \right)$, I am not sure how to find the overall frequency response. Is it possible to express $X_c\left(j\Omega \frac{T_2}{T_1}\right)$ in terms of $X_c\left(j\Omega \right)$ using some continuous-time filter?

$\endgroup$
  • $\begingroup$ why is the "ideal D/C system" running at half the sample rate ($ \frac{1}{T_2} = \frac{1}{2} \cdot \frac{1}{T_1} $ that the subsystems preceding it on the signal chain? how is the portion of the DTFT $Y(\omega)$ for $\frac{\pi}{2} < |\omega| < \pi$ defined? $\endgroup$ – robert bristow-johnson Oct 24 '16 at 23:39
  • 1
    $\begingroup$ BTW, a system that samples a continuous-time system is not LTI. it is LTI if the time variance is limited to integer multiples of the sampling period. $\endgroup$ – robert bristow-johnson Oct 24 '16 at 23:41
1
$\begingroup$

Is it possible to express $X_c\left(j\Omega \frac{T_2}{T_1}\right)$ in terms of $X_c\left(j\Omega \right)$ using some continuous-time filter?

an LTI system cannot do that. continuous-time "filters" need not be LTI but usually we mean LTI when we consider "filters" in analog or digital filters.

you can change the spectrum of $X_c\left(j\Omega \right)$ to $X_c\left(j\Omega \frac{T_2}{T_1}\right)$ by speeding up (or slowing down) the time index of $x_c(t)$ to $x_c\left(t \frac{T_1}{T_2}\right)$.

it appears you want a pitch shifter or similar. is that it?

$\endgroup$
  • $\begingroup$ My question was poorly worded. As you pointed out, this is not even LTI as the D/C runs at half the sample rate. $\endgroup$ – ryan80 Oct 25 '16 at 11:36

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.