Clay wrote:


>>There's no practical use for 24th-order cross-over filters. You aren't
>>likely to build anything that is better than 2nd-order Linkwitz-Riley.
>>
>>Jerry
>>--
> 
> 
> There can be an instructive point to making a 24th order Bessel
> filter. But as Vladimir and Jerry say, the actual application is not
> likely to be very practical in there is any real benifit over using a
> 2nd order.

One of my areas are the consumer and car audio applications. Of course 
we experimented with the different crossover filters, and came to the 
conclusion that the IIR filters of the 4th order are adequate and 
sufficient for the most of applications. The type of the filter doesn't 
really matter, Butterworth is not any better or worse compared to Bessel 
or LR.


> I recall years ago when I hooked up a dsp eval board with a
> good 16 bit codec to be in the loop between my preamp out and my main
> amp in. Playing with different filters was fun and instructive.

Most of the observable effects are caused by the difference of the 
delays in the high and low paths, especially if the long FIR filters are 
used. If the delays are set equal, there is basically no matter what are 
filters.


> balanced spectrally.  I believe the OP must be thinking in terms of
> only band A's power can go to band A's speaker. He doesn't realize
> that only a modest amount of rejection of out of band energy is all
> that is needed.


If the company A makes a filter of the 4th order, and the company B 
makes the filter of the 5th order, every customer knows that the product 
of the company B is better. Isn't it? Going down this road, we can reach 
to the filter of the 24th order. (Probably, the numbers 25 or 64 would 
look better).  Specmanship is very important although it can be meaningless.


Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

http://www.abvolt.com

On Jul 2, 12:30 pm, Jerry Avins <j...@ieee.org> wrote:
> gangadhar.m wrote:
> > I am implementing a Crossover in c sharp platform. I have already
> > implemented 24th order Butterworth and Linkwitz Riley filters.
>
> > Now I need to implement 24th order Bessel filter with cascaded biquad
> > sections. Can any of you help me with the procedure?
>
> There's no practical use for 24th-order cross-over filters. You aren't
> likely to build anything that is better than 2nd-order Linkwitz-Riley.
>
> Jerry
> --
> Engineering is the art of making what you want from things you can get.
> =AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=
=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=
=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF

Hello Jerry and others,

There can be an instructive point to making a 24th order Bessel
filter. But as Vladimir and Jerry say, the actual application is not
likely to be very practical in there is any real benifit over using a
2nd order. I recall years ago when I hooked up a dsp eval board with a
good 16 bit codec to be in the loop between my preamp out and my main
amp in. Playing with different filters was fun and instructive. There
was even a neat effect resulting from "brick wall" filters. I
rememember doing things like putting the audio through a 127 tap
brickwall FIR filter to see how it sounds. Well if the audio is cut
off abruptly in frequency (cutoff somewhere auditorily midband), it
just sounds weird. It is sort of like the brain is expecting to hear
some of the frequencies that were so completely removed. I think there
is a pschoacoustic phenominon where audio needs to be somewhat
balanced spectrally.  I believe the OP must be thinking in terms of
only band A's power can go to band A's speaker. He doesn't realize
that only a modest amount of rejection of out of band energy is all
that is needed.

Clay

gangadhar.m wrote:
> I am implementing a Crossover in c sharp platform. I have already
> implemented 24th order Butterworth and Linkwitz Riley filters.
> 
> Now I need to implement 24th order Bessel filter with cascaded biquad
> sections. Can any of you help me with the procedure?

There's no practical use for 24th-order cross-over filters. You aren't 
likely to build anything that is better than 2nd-order Linkwitz-Riley.

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

On Jul 2, 10:30 am, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:

> > To implement a Bessel filter, the difference with a Butterworth
> > structure is
> > that a Bessel filter has a feedback path.  The book "Numerical Recipes
> > in C"
> > should cover this.
>
> ?????? I woudn't even comment on that. Is it a monday morning hangover?
> Julius, your postings are much better as a rule.

Vladimir, I am disappointed in you.  One gets what one pays for,
right? ;-)

Better to let the original poster have something to play with, since
he gave
no indication of experience nor ambition.

Julius

julius wrote:

> On Jul 2, 1:38 am, "gangadhar.m" <gangadha...@jasmin-infotech.com>
> wrote:
> 
>>I am implementing a Crossover in c sharp platform. I have already
>>implemented 24th order Butterworth and Linkwitz Riley filters.

24th order Butterworth or L/R is nightmare from the numerical standpoint 
and nonsense from the practical standpoint.

>>Now I need to implement 24th order Bessel filter with cascaded biquad
>>sections. Can any of you help me with the procedure?
>>
>>Regards
>>Gangadhar
> 
> 
> Are you asking about how to implement a Bessel filter given the
> coefficients,
> or how to generate the coefficients?
> 
> To implement a Bessel filter, the difference with a Butterworth
> structure is
> that a Bessel filter has a feedback path.  The book "Numerical Recipes
> in C"
> should cover this.

?????? I woudn't even comment on that. Is it a monday morning hangover?
Julius, your postings are much better as a rule.

> 
> To actually find the coefficients, this is covered in many signal
> processing
> books, such as one by Proakis/Manolakis.

A digital Bessel is rather non-trivial design, and it is not a basic 
textbook issue.


Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

http://www.abvolt.com

On Jul 2, 1:38 am, "gangadhar.m" <gangadha...@jasmin-infotech.com>
wrote:
> I am implementing a Crossover in c sharp platform. I have already
> implemented 24th order Butterworth and Linkwitz Riley filters.
>
> Now I need to implement 24th order Bessel filter with cascaded biquad
> sections. Can any of you help me with the procedure?
>
> Regards
> Gangadhar

Are you asking about how to implement a Bessel filter given the
coefficients,
or how to generate the coefficients?

To implement a Bessel filter, the difference with a Butterworth
structure is
that a Bessel filter has a feedback path.  The book "Numerical Recipes
in C"
should cover this.

To actually find the coefficients, this is covered in many signal
processing
books, such as one by Proakis/Manolakis.

Julius

I am implementing a Crossover in c sharp platform. I have already
implemented 24th order Butterworth and Linkwitz Riley filters.

Now I need to implement 24th order Bessel filter with cascaded biquad
sections. Can any of you help me with the procedure?

Regards
Gangadhar