XO filters that sum to linearity - visualization

[detlev24]

Hello,

This is more of academic interest thus far, as I have been looking at SigmaStudio's "Crossover Filter" module. While the following filter types can be selected [in dB/octave]:

Linkwitz-Riley: 12/24/36/48
Butterworth: 12/18/24
Bessel: 12/18/24

I am missing other options, amongst these: Butterworth 6, 30, 36, 42, 48 dB/octave.

My interest is to find out which other combinations, besides (2x) Butterworth 18 or (2x) Linkwitz-Riley 24 or 48 dB/octave would sum to a linear frequency response - see examples attached. Is there another graphical tool available, that offers more freedom?

Later applications might include (digital) crossovers for subwoofers to the main loudspeakers; while steep(er) linear phase crossover slopes seem to be beneficial in the time-domain.

Thank you all!


. .

 

[bigjacko]

Are there any other crossover out there? Like elliptical or something?

 

[digitalfrost]

Well there are Neville-Thiele crossovers

http://www.acourate.com/freedownload/XOWhitePaper.pdf

https://sound-au.com/articles/ntm-xover.htm

 

[Justin Zazzi]

You can use SigmaStudio's "stimulus" and "response" blocks placed before and after a set of filters to see how they sum back together. Between the stimulus and response blocks you can place a collection of whatever you like to emulate whatever filter you can think of.

 

[Justin Zazzi]

There is another tool that I'm liking for this too. JL Audio released a new version of the TuN software that can simulate the transfer function of combining filters. It might not be as flexible as you need it to be, but it's really efficient and easy to use for what it does.

Discover TüN® 4 - JL Audio’s DSP tuning software

www.jlaudio.com

 

[DonH56]

You can modify most any filter topology to create a linear frequency response, I think, but I am not sure what you are defining as "linear". Flat magnitude, linear phase, ???

Linkwitz-Riley crossovers are comprised of cascaded Butterworth stages to provide flat magnitude response through the crossover region while maintaining the Butterworth characteristics. Elliptical filters can provide linear phase and higher roll-off at the cost of ripple (magnitude variation) in the pass band and stop band. Chebyshev filters can also provide steeper roll-off than Butterworth at the cost or ripple in either the pass band or stop band. Bessel filters provide excellent phase (pulse) response but slow roll off. Etc.

There was a speaker touting its use of elliptical crossovers but I cannot for the life of me remember which one. Most manufacturers do not say much about their internal crossover designs; it is part of the "secret sauce".

Linear phase means constant group delay, i.e. all frequencies are delayed equally, and thus optimal pulse response (time-domain response). Except for bandwidth changing, a pulse from a linear-phase filter looks the same as it did going in. There is a lot of debate about how much that matters in an audio system, with a body of research saying it really doesn't unless it is really bad. Pulse integrity is critical in things like radar systems (where I was weaned) but is not something I have worried much about in my audio system. I have had speakers that had great impulse response and poor and think, at least for myself, that other factors dominate their sound.

Filter design is the subject of many texts and college course (I have a number of both) and is a pretty big topic to tackle on an Internet forum thread. There are online filter calculators that let you compare responses but I have not looked for them in a while.

HTH - Don

 

[abdo123]

Well there are Neville-Thiele crossovers

http://www.acourate.com/freedownload/XOWhitePaper.pdf

https://sound-au.com/articles/ntm-xover.htm

Do you have any idea where i can read more about R type Bessel filters? I couldn't manage to replicate this in VituixCAD.

 

[fpitas]

Some pertinent reading:

RANE Commercial - Knowledge Base

Rane Commercial

www.ranecommercial.com

 

[fpitas]

There's an elephant dancing in the corners of this room. You can get Butterworth (for example) to sum flat easily enough on-axis. Just put the two drivers in quadrature, 90 degrees apart. But, then you end up with a big +3dB maxima lobe pointing upward, or downward in the vertical plane, depending on the polarity of the drivers. The point here is that the on-axis is an important consideration, but the polar response matters, too. Unless you listen in an anechoic chamber