1
$\begingroup$

I'm busy trying to implement a filter from this paper. I'm a bit confused with one particular step (30e in the paper).

Basically I need to find the zero frequency gain (K) of the following whitening/comb filter:

Comb Filter

The relevant steps in the algorithm are:

enter image description here

and

enter image description here

where $q^{-1}$ is the unit delay operator, $\omega_0$ is the fundamental freq of the comb and $\rho$ is a constant.

I'm not sure how to calculate K. Since the function isn't acting on anything, what should be delayed? The signal (at time t)? Or does passing a constant in place of the delay mean it should be replaced?

Any help is greatly appreciated!

EDIT

Some clarification on $q^{-1}$:

The right-shift operator $q^{-1}$ acts on a discrete-time signal f(k, T) as $q^{-1}$ f(k, T) = (k − 1, T). The operation $q^{-1}$ therefore shifts the discrete-time signal to the right by one sampling interval.

$\endgroup$
  • $\begingroup$ i would recommend that, instead of taking pics of equations, you should use $\LaTeX$ to write out the equations so that they can be edited. that said, please lose the $q^{-1}$ and replace with $z^{-1}$. then i might read the question, otherwise i won't bother to. $\endgroup$ – robert bristow-johnson Mar 10 '15 at 14:11
  • $\begingroup$ $q^{-1}$ is the symbol used in the paper, if you really want me to replace it I will, but does it really matter? $\endgroup$ – Pokey McPokerson Mar 10 '15 at 14:13
  • $\begingroup$ i use "$q$" for zeros and "$p$" for poles. i dunno whose book this came outa but the notation is not conventional. $\endgroup$ – robert bristow-johnson Mar 10 '15 at 14:17
  • $\begingroup$ I'd prefer to leave it as it is. I'm trying to understand what the author intended, so I think it would be best to use their words. Sorry you won't read my question. $\endgroup$ – Pokey McPokerson Mar 10 '15 at 14:56
2
$\begingroup$

With $A(q^{-1},t)$ as given in the formula you get

$$A(1,t)=\prod_{k=1}^n(1-2\cos(k\hat\omega_0(t))+1)\tag{1}$$

and

$$A(\rho,t)=\prod_{k=1}^n(1-2\cos(k\hat\omega_0(t))\rho+\rho^2)\tag{2}$$

So $K(t)$ is simply the quotient of (1) and (2).

$\endgroup$
  • $\begingroup$ Me and me colleague came to the same conclusion, I just found it confusingly written. Thanks for the answer. $\endgroup$ – Pokey McPokerson Mar 10 '15 at 15:18

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.