Crossover Debate: Bessel vs. Butterworth

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jump to Latest
J

Jmmlc

R.I.P.
Joined 2005
Re: Re: Re: Crossover Debate: Bessel vs. Butterworth

Hello,

You are perfectly right.

Eventually one can argue that at this highr frequencies, a given variation of phase has far less influence on the total pulse reponse than at lower frequency. Even a variation of 180degrees from 4000Hz to 20000Hz seems to be inaudible...

Best regards from Paris, France

Jean-Michel Le Cléac'h



pooge said:


This is not entirely correct. High pass filters cannot have flat time response. So even an isolated HP Bessel does not have a flat time delay like its low pass counterpart.



http://users.ece.gatech.edu/~mleach/papers/phasedist.pdf
Click to expand...
 
M

MarcMTL

Member
Joined 2008
Thanks for all the responses guys. But I seem to be getting in over my head here with all this talk of phase response and such.

What I'm asking is, are you better off going with a Butterworth that typically offers a flat magnitude response, or the Bessel that offers a linear phase response?

I'm pretty new at this, so I'm still learning here. What I've seen is that phase response is kind of like voodoo; some believe in it and others don't!

I'm really interested in all this stuff, so whatever I can get from your guys is a gain for me.

What is the importance of linear phase response? Should I concentrate more on flat magnitude response?

Thanks Guys!

Marc
 
planet10

planet10

frugal-phile™
Joined 2001
planet10-hifi.com
Paid Member
MarcMTL said:
What I'm asking is, are you better off going with a Butterworth that typically offers a flat magnitude response, or the Bessel that offers a linear phase response?
Click to expand...

In some sense what is being said, is that in the light of the impedance of a driver, and the FR of the driver, neither. Text book doesn't work on a speaker. You can use either as a rough starting point, then you have to work from there.

Me, i just try not to bother with crossovers at all... a good full-range (particularily for a 1st project) can get you a long ways with minimum loss of hair. Or build something where someone else has already figured the XO out. My maxim is that XOs are evil (certainly the weakest part of a loudspeaker)

dave
 
PB2

PB2

Member
Joined 2004
linkedin.com/in/petebasel
S.P. Lipshitz and J. Vanderkooy, "A Family of Linear-Phase Crossover Networks of High Slope Derived by Time Delay", J. Aud. Eng. Soc, vol 31, pp2-20 (1983 Jan/Feb.).

The SPICA TC-50 was the first that I know of to use a Bessel low pass with a roughly first order high pass to provide an approximately linear phase response, and yes there is ripple in the passband. I happen to be looking at these speakers:
http://baselaudiolabs.googlepages.com/MR-TC50-REB.html

Pete B.
 
F

fredex

Member
Joined 2005
Re: Re: Crossover Debate: Bessel vs. Butterworth

planet10 said:
What the best XO is depends on everything...
dave
Click to expand...

True

gedlee said:
To talk about a crossover as if it were a simple electrical problem is simply absurd. The crossover is a complex system of electro-acoustic summations of multiple non-coincident and non-ideal sources that occurs in a full three dimensional field. The correct crossover to yield the best impulse response, both on and off axis, is never going to have a simple electrical specifcation. Its not going to look anything like an ideal "Bessel" or Butterworth" or anything else you want to call it.
Click to expand...

True.

Without a lot of additional information there is no correct answer to the question implied by the title of this thread.
 
Dr. Ram

Dr. Ram

Member
Joined 2006
audesine.com
MarcMTL said:
What is the importance of linear phase response? Should I concentrate more on flat magnitude response?
Marc
Click to expand...

Marc, if you are not already in an opinion camp about phase, I would suggest you give phase response less priority and concentrate on flatness of magnitude response (and aim for a system with decent *transient* response). However, it is still important that you have your crossover design done in a manner that the summations do *take into account* the phase behavior of the xover alignment so that the predicted amplitude response is accurate.

Further, do note that the enclosure itself will have an alignment. If that alignment is not a low-transient one (i.e., has a Butterworth-type Q factor) then there is not much point in having a crossover which is Bessel-like because you really cannot get the same transient behavior out of a system whose enclosure AND crossover were both Bessel-like.

HTH.

-Ram
 
J

john k...

Member
Joined 2004
PB2 said:
S.P. Lipshitz and J. Vanderkooy, "A Family of Linear-Phase Crossover Networks of High Slope Derived by Time Delay", J. Aud. Eng. Soc, vol 31, pp2-20 (1983 Jan/Feb.).

The SPICA TC-50 was the first that I know of to use a Bessel low pass with a roughly first order high pass to provide an approximately linear phase response, and yes there is ripple in the passband. I happen to be looking at these speakers:
http://baselaudiolabs.googlepages.com/MR-TC50-REB.html

Pete B.
Click to expand...


The L + V work is not the same as the SPICA approach. In fact, L+V showed in another paper that linear phase crossovers can not be obtained by using driver offset, though, as the SPICA work showed a good approximation can be obtained.

Once again, see my web page , the section, Approximately Transient Perfect Crossovers using Time Alignment, about 2/3 the way down.
 
D

dlr

Member
Joined 2005
speakerdesign.net
john k... said:



The L + V work is not the same as the SPICA approach. In fact, L+V showed in another paper that linear phase crossovers can not be obtained by using driver offset, though, as the SPICA work showed a good approximation can be obtained.

Once again, see my web page , the section, Approximately Transient Perfect Crossovers using Time Alignment, about 2/3 the way down.
Click to expand...

Hi John,

I've got one of the most recent incarnations of Firefox. Your web page graphics overlap the text pretty badly. I see that you're using Yahoo's editor, SiteBuilder. It's apparently got compatibility issues. Having a paid site, you can use an ftp client to upload pages that you create on your PC off-line. It would be helpful to those who eschew MS products.

If you come to DIY NE this year, I can show you what I use (available for free on the web) for ftp. You'll probably want a better editor. I use a straight-forward html type, not a GUI, but you should be able to find a good one for that. The built-in editors such as Yahoo just don't provide quality at times.

Dave
 
PB2

PB2

Member
Joined 2004
linkedin.com/in/petebasel
john k... said:



The L + V work is not the same as the SPICA approach. In fact, L+V showed in another paper that linear phase crossovers can not be obtained by using driver offset, though, as the SPICA work showed a good approximation can be obtained.

Once again, see my web page , the section, Approximately Transient Perfect Crossovers using Time Alignment, about 2/3 the way down.
Click to expand...


I am aware of both of your points. I never said that the two were closely related. I'm not an advocate of phase linear designs, just find them interesting.

It's all an approximation anyway as far as speakers go, IMO.

Pete B.
 
D

dlr

Member
Joined 2005
speakerdesign.net
planet10 said:


I seem to recall this a problem going back a ways... iCab, Safari, Firefox & Camino.

dave
Click to expand...

Yeah, I recall comments and John replying that he used the web site software. I had no problem with a slightly older version of Firefox IIRC. Now I have a problem with the page.

I find it easier to use an off-line package to create pages rather than try to work on-line with the supplied tools such as that offered by Yahoo. Since I may see John, I thought I'd put a bug in his ear again. ;)

Dave
 
poptart

poptart

Member
Joined 2004
I think the problem with John's site has to do with the font size you've set as the default in your prefs. When I first go there it looks fine but if I say "make text bigger" it overlaps.

John, do you use a transient perfect crossover in your speakers or is this just a curiosity? I'm not very up to date on these things but I've heard other people say group delay in inaudible.
 
J

john k...

Member
Joined 2004
First the web site. I'm too dumb to learn any other web publishing software but SiteBuilder (or too lazy :)).

Second, I agree that it may be related to the font size used in the web browser. Dave, could you check that? Also, what page(s) are you referring to? Just let me know one so I can check it. I will down load the newest version of FireFox and check it out.

As for TP speakers. I have build a few in the past but the NaO series is not TP. The ITCA will be, if I ever get it completed. I have vacillated on making the ICTA open baffle or sealed box. At this time I'm looking at sealed box design again.
 
D

dlr

Member
Joined 2005
speakerdesign.net
john k... said:
First the web site. I'm too dumb to learn any other web publishing software but SiteBuilder (or too lazy :)).

Second, I agree that it may be related to the font size used in the web browser. Dave, could you check that? Also, what page(s) are you referring to? Just let me know one so I can check it. I will down load the newest version of FireFox and check it out.
Click to expand...

It does appear to be the font. On this PC I've got a default of Arial 14, but the option was set to let the web page specify a font. the page that you linked here on transient analysis displays fine on this PC. I'll have to check my PC at home to see what setting I have. Normally I don't change it, but I have a larger monitor set for higher resolution, so I may have bumped up the default font size.

Looking at the page source I see that the page specifies Arial 14. I must have bumped it up at home.

Dave
 
gedlee

gedlee

Member
Joined 2004
gedlee.com
phase_accurate said:


The whole speaker-design is an approximation: Frequency response is, off-axis response is, distortion is and time-respone is an approximation as well.

But the time-response can be made as accurately as a flat frequency response can be made.

Regards

Charles
Click to expand...


Isn't the point to find out which "approximations" or trade-offs make the most audible difference. They are most certainly not all the same in audibility.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Top Bottom