Re: Lost with direct-form filter equations (inverting)

On Mar 15, 12:13 pm, "Andor" <andor.bari...@xxxxxxxxx> wrote:
On 15 Mrz., 15:51, Jerry Avins <j...@xxxxxxxx> wrote:

Andor wrote:
Oops. I should first learn how to use Greg's method before posting.
After using it correctly, it came up with the exact but unfortunately
unstable inverse ...

Another option would be to use one of the methods for IIR design
posted in another thread ("Generating 1/f^2 and 1/f^3 noise"). Or to
reflect the unstable pole about the unit circle. This should give
exact magnitude but result in a phase error.

Does phase error have any significance for noise?

Why noise? The filters are supposed to be used for de-emphasis of an
analogue emphasis system. In that case, phase matters.

yeah, but if the pre-emphasis was not minimum-phase (i.e. zeros
outside the unit circle which would become nasty-ass poles outside the
unit circle in the de-emph filter), this pre-emphasis has no
realizable inverse (meaning, in my book, that it was not well
designed).

now if phase is that important (i've argued both sides of this "is
phase audible" issue, BTW), then, if you're willing to put up with
some throughput delay, a decent approximation to the inverse filter
can be done with an FIR using any of these FIR design methods (and
that trick in the dspguru tricks page about using PM to design for
magnitude *and* phase).

otherwise (if the phase relationship between frequencies is *not* all
that critical), just reflect the nasty poles into the interior of the
unit circle.

dunno if i can add anything else.

r b-j


.

Relevant Pages

  • Re: Strange integral
    ... yes, I mean exact it ... get a convergent series? ... both poles are inside the unit circle. ...
    (sci.math)
  • Re: Minimum-Phase Systems Characteristics
    ... responses don't really have poles or zeros, ... get a real impulse response) and a constant phase response of -45°.. ... phase spectrum is zero. ... Any zero inside the unit circle will cancel out the ...
    (comp.dsp)
  • Re: Simple question about minimum phase filters
    ... definition from Oppenheim and Schafer says all poles and zeros must be ... stable causal filter. ... zeros can also be on the UC, ... assume that a zero exactly on the unit circle ...
    (comp.dsp)
  • Re: pole zero stuff
    ... > Also confused about unit circle and IIR filters. ... Draw your poles and zeros. ... to your expected IIR filter response? ...
    (comp.dsp)
  • Question about poles/zeros and WMF
    ... z=, if Sh has any additional poles at all, then it ... will have poles outside the unit circle. ... the sentence "The equivalent channel response for the WMF is Sh" ...
    (comp.dsp)