A fundamental question on 0-phase filter

[ZedToe]

Hi,
Thanks for your concern in advance.

I was told that a zero-phased filter Hzp(z) can be used to 'off-line'
filter a time sequence x(n). Since its response Hzp(w) is real, so its
output Y(w) = Hzp(w)X(w) should preserve the phase information of
X(w).

However when I convolute the x(n) with hzp(n), then hzp(n) has no
difference with its linear phase counterpart. Suppose x(n) is a sine
wave, a phase difference is observed in y(n).

Am I wrong in express 'Y(w) = Hzp(w)X(w)' because its a noncausal
filter? and what is the application of zerophase filters?

Thanks.

Zedtoe

[Rick Lyons]

On 19 Aug 2003 07:41:44 -0700, [email protected]
(ZedToe) wrote:

>Hi,
>Thanks for your concern in advance.
>
>I was told that a zero-phased filter Hzp(z) can be used to 'off-line'
>filter a time sequence x(n). Since its response Hzp(w) is real, so its
>output Y(w) = Hzp(w)X(w) should preserve the phase information of
>X(w).
>
>However when I convolute the x(n) with hzp(n), then hzp(n) has no
>difference with its linear phase counterpart. Suppose x(n) is a sine
>wave, a phase difference is observed in y(n).
>
>Am I wrong in express 'Y(w) = Hzp(w)X(w)' because its a noncausal
>filter? and what is the application of zerophase filters?
>
>Thanks.
>
>Zedtoe

Hi Zedtoe (humm, ... zedtoe, wonder what that means?)

using a nonlinear-phase IIR filter, you:

1. filter a time sequence,
2. flip (reverse in time) the filter's output sequence,
3. filter (with the same IIR filter) the flipped sequence,
4. flip the 2nd filter's output.

[-Rick-]

[Scott Gilbert]

[email protected] (Rick Lyons) wrote in message
>
> using a nonlinear-phase IIR filter, you:
>
> 1. filter a time sequence,
> 2. flip (reverse in time) the filter's output sequence,
> 3. filter (with the same IIR filter) the flipped sequence,
> 4. flip the 2nd filter's output.
>

How far past the end of your signal do you drag the filter in step 1?
Do you repeat the signal, or stuff zeros after the end?

Thanks in advance,
-Scott

[Jerry Avins]

Scott Gilbert wrote:
>
> [email protected] (Rick Lyons) wrote in message
> >
> > using a nonlinear-phase IIR filter, you:
> >
> > 1. filter a time sequence,
> > 2. flip (reverse in time) the filter's output sequence,
> > 3. filter (with the same IIR filter) the flipped sequence,
> > 4. flip the 2nd filter's output.
> >
>
> How far past the end of your signal do you drag the filter in step 1?
> Do you repeat the signal, or stuff zeros after the end?
>
> Thanks in advance,
> -Scott

Keep collecting output until it pretty much peters out.

Jerry
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯ ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯