I have an analog, continuous impulse response $$h_a(t)=u(t)\cdot\sum_{n=1}^4A_ne^{s_nt}$$, and by sampling it with the usual method I get $$h[n]=\Delta th_a(n\Delta t).$$ Now, that $\Delta t$ is a bit problematic for me, because upon performing the z-transform of $h[n]$ I get $$H(z)=\sum_{n=0}^\infty h[n]z^{-n}\\=\sum_{n=0}^\infty\sum_{k=1}^4\Delta tA_k\Big(\frac{e^{s_k\Delta t}}{z}\Big)^{n}\\=\Delta t\sum_{k=1}^4A_k\sum_{n=0}^\infty\Big(\frac{e^{s_k\Delta t}}{z}\Big)^{n}\\=\Delta t\sum_{k=1}^4A_k\sum_{n=0}^\infty\beta^n \\=\Delta t\sum_{k=1}^4A_k\frac{1}{1-\beta}$$ $$=\sum_{k=1}^4\frac{\mathbf{\Delta t}A_k}{1-e^{s_k\Delta t}z^{-1}}$$

with $\beta\triangleq\frac{e^{s_k\Delta t}}{z},|\beta|<1$. If I implement such transfer function in an IIR discrete-time filter, each coefficient in the difference equation contains $\Delta t$, except for the coefficients in the feedback part. This results in an incorrect amplitude, when compared to the plot of $h_a(t)$. Therefore, I wonder, since not including $\Delta t$ makes the discretized $h[n]$ practically equal to $h_a(t)$ (in the sampling points), why would I use that coefficient at all? Or better, am I supposed to use it and since I get different results am I doing something wrong? I remember it was necessary for some definition (DFT or autofunctions), but I don't see any practical use of it.

Sorry if it's a stupid question, I hope I haven't missed anything.


Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.