Filter design with 'freqsamp' method?

On Jul 2, 4:42=A0pm, spop...@speedymail.org (Steve Pope) wrote:
> Pete Fraser <pfra...@covad.net> wrote:
> >"Steve Pope" <spop...@speedymail.org> wrote in message
> >> A good thing to remember is that any filter that is as
> >> selective as a given Butterworth filter will have a
> >> similar RMS delay spread as the Butterworth filter.
> >> So no free lunch on phase issues in many cases.
> >But I'm doing a symmetric FIR with the Butterworth
> >amplitude response.

so then you'll get even *more* phase shift delay.  (Butterworths are
minimum phase and linear phase is not minimum phase unless it's a wire
or a simple gain.)

> Linear phase is overrated. =A0:-)

yeah, and what Steve said.

r b-j

On Jul 1, 10:59=A0am, Fred Marshall <fmarshallx@remove_the_xacm.org>
wrote:
>
> r b-j,
>
> Well, maybe I've had it wrong all these years but I'd say that the
> windowing method starts with N frequency samples where N is the length
> of the filter you want.

so then, since the DFT and iDFT are bijective (i love using fancy-
pants words), why window?  if h[n] has N samples and N degrees of
freedom, so does H[k].  you specify your N frequency samples and you
can hit it perfectly with no windowing.

so, that seems curious to me.

r b-j

Steve Pope wrote:
> A good thing to remember is that any filter that is as
> selective as a given Butterworth filter will have a
> similar RMS delay spread as the Butterworth filter.
> So no free lunch on phase issues in many cases.

Pete Fraser replied:
>But I'm doing a symmetric FIR with the Butterworth
>amplitude response.

Then robert bristow-johnson said:
> so then you'll get even *more* phase shift delay.  (Butterworths are
> minimum phase and linear phase is not minimum phase unless it's a wire
> or a simple gain.)

True, but Steve and I were talking about RMS delay spread.

Pete 

On Jul 2, 7:55=A0pm, "Pete Fraser" <pfra...@covad.net> wrote:
>
> ... but Steve and I were talking about RMS delay spread.
>

"spread" from what?  zero?  or some mean delay?  how is the mean
defined?

"RMS delay spread" hasn't been a parameter i've been familiar with.
is it a measure of overall phase nonlinearity?

just curious.

r b-j

robert bristow-johnson  <rbj@audioimagination.com> wrote:

>On Jul 2, 7:55&#4294967295;pm, "Pete Fraser" <pfra...@covad.net> wrote:

>> ... but Steve and I were talking about RMS delay spread.

>"spread" from what?  zero?  or some mean delay?  how is the mean
>defined?

>"RMS delay spread" hasn't been a parameter i've been familiar with.

>just curious.

Interestingly some of the literature defintions of RMS delay spread
I disagree with, but it basically goes something like this: if the impulse
response is:

      (sum over i) (h(i) * (t - tau(i)))

then the average delay is

     ave = (sum over i)(h(i)*tau(i) / (sum over i)(h(i))

and the RMS delay spread is

    sqrt((sum over i)(h(i)*((tau(i) - ave)^2)).

>is it a measure of overall phase nonlinearity?

No, the phase can be linear but you can have a large RMS delay spread.

I'm interested if anyone doesn't like the expression I just gave above.
I've seen expressions that were not equivalent to this (I'm going to
say by Spencer).


Steve

Oops, let me fix this.  I left out some absolute values.

the impulse response is:

      (sum over i) (h(i) * (t - tau(i)))

then the average delay is

     ave = (sum over i)(abs(h(i))*tau(i) / (sum over i)(abs(h(i)))

and the RMS delay spread is

    sqrt((sum over i)(abs(h(i))*((tau(i) - ave)^2)).



S.

On Fri, 2 Jul 2010 20:37:14 -0700 (PDT), robert bristow-johnson
<rbj@audioimagination.com> wrote:

>> ... but Steve and I were talking about RMS delay spread.
>>
>
>"spread" from what?  zero?  or some mean delay?  how is the mean
>defined?
>
>"RMS delay spread" hasn't been a parameter i've been familiar with.
>is it a measure of overall phase nonlinearity?

Possibly useful:  see US Patent 5375067 for a discussion of Central Time and RMS
Delay.

Greg

On Sat, 03 Jul 2010 06:48:57 -0400, Greg Berchin <gberchin@comicast.net.invalid>
wrote:

>Possibly useful:  see US Patent 5375067 for a discussion of Central Time and RMS
>Delay.

I gotta stop typing before my brain is awake.

Make that Central Time and RMS *Duration*.

On Jul 3, 6:48=A0am, Greg Berchin <gberc...@comicast.net.invalid> wrote:
> On Fri, 2 Jul 2010 20:37:14 -0700 (PDT), robert bristow-johnson
>
> <r...@audioimagination.com> wrote:
> >> ... but Steve and I were talking about RMS delay spread.
>
> >"spread" from what? =A0zero? =A0or some mean delay? =A0how is the mean
> >defined?
>
> >"RMS delay spread" hasn't been a parameter i've been familiar with.
> >is it a measure of overall phase nonlinearity?
>
> Possibly useful: =A0see US Patent 5375067 for a discussion of Central Tim=
e and RMS
> Duration.
>


Greg, i didn't know about this patent of yours.  good show!

well, i'm still trying to pin these guys down.  if their r.m.s. delay
parameter ("spread" or whatever it's called) is a measure of deviation
from phase linearity, then nothing can beat a phase linear FIR.  but
if it's a measure of phase delay or group delay (from zero), then a
minimum-phase filter will beat linear-phase.

that's all i wanted to say about it.

r b-j

robert bristow-johnson  <rbj@audioimagination.com> wrote:
>well, i'm still trying to pin these guys down.  if their r.m.s. delay
>parameter ("spread" or whatever it's called) is a measure of deviation
>from phase linearity, then nothing can beat a phase linear FIR.  but
>if it's a measure of phase delay or group delay (from zero), then a
>minimum-phase filter will beat linear-phase.

It is neither.

Steve