Difference between Linkwitz Riley and Butterworth Crossover? - AVS Forum
Forum Jump: 
 
Thread Tools
Old 08-13-2014, 08:17 AM - Thread Starter
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 25,132
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 240 Post(s)
Liked: 1179
Difference between Linkwitz Riley and Butterworth Crossover?

I've been playing around with my miniDSP and trying to decide if I want to attempt my own DIY project with active crossovers.

I've always seen these terms, but I don't specifically know what they mean or what the difference is. Anyone care to give me a quick explanation 101 on the differences between Linkwitz Riley, Bessel, and a Butterworth crossover ?

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is offline  
Sponsored Links
Advertisement
 
Old 08-13-2014, 08:25 AM
AVS Addicted Member
 
Bill Fitzmaurice's Avatar
 
Join Date: Nov 2009
Location: New Hampshire
Posts: 10,506
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 1769
Read this:
http://www.bcae1.com/xoorder.htm
zeus33 likes this.

Bill Fitzmaurice Loudspeaker Design

The Laws of Physics aren't swayed by opinion.
Bill Fitzmaurice is online now  
Old 08-13-2014, 08:29 AM
Advanced Member
 
Bassment's Avatar
 
Join Date: Aug 2013
Location: Ontario, Canada
Posts: 921
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 425 Post(s)
Liked: 171
Butterworth for odd filters, Linkwitz for even. A Linkwitz is just cascaded butterworth filters. Bessel filters are for keeping an even phase/group delay.

Each of the filters has a different Q value, which give different amplitude responses. BW is Q or 0.49, LR is 0.71 (which is 0.49^2), which is because it's just a double BW filter.

LR filters are -6 dB at crossover, so they sum to flat (0 dB). BW are -3 dB, so they sum to +3 dB.

Scott Simonian and Mfusick like this.
Bassment is offline  
Old 08-13-2014, 11:12 AM
AVS Special Member
 
zeus33's Avatar
 
Join Date: Dec 2012
Posts: 1,692
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 364 Post(s)
Liked: 269
Quote:
Originally Posted by Bill Fitzmaurice View Post

Great link. Thank you Bill.
zeus33 is offline  
Old 08-13-2014, 11:17 AM - Thread Starter
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 25,132
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 240 Post(s)
Liked: 1179
so if I wanted to cross over say (2) 15" woofers to (2) 6" midranges and then to a single CD/SEOS horn I would theorectically use a 3 way Linkwitz ?

Or is this something I'd play with and measure ?

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is offline  
Old 08-13-2014, 11:27 AM
Advanced Member
 
Bassment's Avatar
 
Join Date: Aug 2013
Location: Ontario, Canada
Posts: 921
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 425 Post(s)
Liked: 171
Yes, you'd need to measure. But in general you would do a low pass on the 15", say 300 hz, a high pass on the mids at 300 hz, a low pass on the mids at say 1300 hz, and a high pass on the tweeter at 1300 hz. A 2nd order Linkwitz would be a good place to start. What SEOS/CD though? You probably don't need mids if your CD and waveguide can go down to 600 hz around, and if your woofers can go up to 1000 hz around.

You're doing active right?
Bassment is offline  
Old 08-13-2014, 11:41 AM - Thread Starter
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 25,132
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 240 Post(s)
Liked: 1179
Yes active.

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is offline  
Old 08-13-2014, 11:48 AM
AVS Special Member
 
tuxedocivic's Avatar
 
Join Date: Feb 2010
Location: Ladysmith, BC
Posts: 5,176
Mentioned: 13 Post(s)
Tagged: 1 Thread(s)
Quoted: 620 Post(s)
Liked: 700
The filter you program into the mini actually has nothing to do with the XO you're trying to achieve. You might use a BW1 to get an acoustic LR6 if the driver has a natural steep slope on it. You also might put the filter at 700hz to achieve a 500hz XO. You also need to consider many more things.

Can I suggest starting with something much simpler. Active is not really any different than passive. You need to follow the same process.
tuxedocivic is online now  
Old 08-13-2014, 12:00 PM - Thread Starter
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 25,132
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 240 Post(s)
Liked: 1179
Well my first project is going to be a 2 way crossing a single woofer to a single CD.

Then I think I'll buy a cheap mid and try adding that.

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is offline  
Old 08-13-2014, 04:07 PM
AVS Special Member
 
tuxedocivic's Avatar
 
Join Date: Feb 2010
Location: Ladysmith, BC
Posts: 5,176
Mentioned: 13 Post(s)
Tagged: 1 Thread(s)
Quoted: 620 Post(s)
Liked: 700
Well that's certainly a much better way to start. Smaller and simpler is even better. Learning anything is about baby steps. Maybe start with a dome and paper 5.25" woofer. Compression drivers aren't easy to tame.
pgwalsh likes this.
tuxedocivic is online now  
Old 08-14-2014, 04:18 AM
AVS Special Member
 
Martycool007's Avatar
 
Join Date: Jun 2011
Posts: 3,692
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 280 Post(s)
Liked: 174
Quote:
Originally Posted by Mfusick View Post
Well my first project is going to be a 2 way crossing a single woofer to a single CD.

Then I think I'll buy a cheap mid and try adding that.

I am getting ready to dive into active speaker building myself. I already have my speaker parts. Just waiting until I can afford the wood, wire, speakon connectors and another iNuke1000dsp, which is what I am using as the active crossover devise, and it also has the benefits of being an amplifier too!
Martycool007 is offline  
Old 08-18-2014, 05:16 AM - Thread Starter
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 25,132
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 240 Post(s)
Liked: 1179
I've always seen the iNUKES used for subs but never saw anyone comment on their sound quality for full range.

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is offline  
Old 12-30-2014, 06:44 PM - Thread Starter
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 25,132
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 240 Post(s)
Liked: 1179
Quote:
Originally Posted by Bassment View Post
Butterworth for odd filters, Linkwitz for even. A Linkwitz is just cascaded butterworth filters. Bessel filters are for keeping an even phase/group delay.

Each of the filters has a different Q value, which give different amplitude responses. BW is Q or 0.49, LR is 0.71 (which is 0.49^2), which is because it's just a double BW filter.

LR filters are -6 dB at crossover, so they sum to flat (0 dB). BW are -3 dB, so they sum to +3 dB.

So revisiting this a little...


Planning to finish up my 2 way TD15 with DNA360/SEOS15 build pretty soon. I am wondering if there is any specific place I should start or test for an active crossover.

I was thinking 950hz might be a good spot, but honestly I'm not sure. I was planning to tinker a little. Suggestions ?

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is offline  
Old 12-30-2014, 07:33 PM
AVS Special Member
 
augerpro's Avatar
 
Join Date: Aug 2005
Location: Denver, Colorado
Posts: 1,726
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 19 Post(s)
Liked: 29
Ideally you would cross where the directivity is basically the same for the horn and woofer, while allowing the drivers to operate in their comfort zone. Then listen and go from there.

Since there is really no connection between the filter type and crossover frequency you set in the minidsp, and the actual acoustic crossover slope and frequency of the loudspeaker, you should measure and model using software like PCD and Holmimpulse.
augerpro is offline  
Old 12-31-2014, 10:01 AM - Thread Starter
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 25,132
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 240 Post(s)
Liked: 1179
Quote:
Originally Posted by augerpro View Post
Ideally you would cross where the directivity is basically the same for the horn and woofer, while allowing the drivers to operate in their comfort zone. Then listen and go from there.

Since there is really no connection between the filter type and crossover frequency you set in the minidsp, and the actual acoustic crossover slope and frequency of the loudspeaker, you should measure and model using software like PCD and Holmimpulse.
I was checking out Xsim: http://libinst.com/Xsim/XSimSetup.exe

and WinPCD: http://www.speakerdesign.net/WinPCD/installation.htm

I'd like to figure out how to properly design a crossover- and play around and learn rather than just take what other's have done.

I know @bwaslo did some good work on these- so my thinking was I could copy him. I know I would get a reasonable result.

But I would like to also learn how to measure and design, figure out the polars, etc.. I also wonder if I should be putting something in line with the tweeter to keep from frying them?

Necessary on an active?

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is offline  
Old 12-31-2014, 11:00 AM
Senior Member
 
Nate Hansen's Avatar
 
Join Date: Aug 2012
Location: Rochester, MN
Posts: 255
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 76 Post(s)
Liked: 58
Quote:
Originally Posted by Mfusick View Post
I also wonder if I should be putting something in line with the tweeter to keep from frying them?

Necessary on an active?
Not absolutely necessary, but I do just to be on the safe side. Just make sure it's large enough that it doesn't effect the low end response of the CD. Another route would be to use a small cap (1uF or so) that will knock down the extra midrange energy, but then it's always effecting the response and you're limited on what you can do.
Nate Hansen is offline  
Old 12-31-2014, 11:04 AM - Thread Starter
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 25,132
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 240 Post(s)
Liked: 1179
I want to do my crossover active. but I would also want to protect the CD from damage. I would imagine a DNA360 might be harsh and quite loud before I damaged it though..

-

"Too much is almost enough. Anything in life worth doing is worth overdoing. Moderation is for cowards."
Mfusick is offline  
Old 12-31-2014, 11:07 AM
Advanced Member
 
Samps's Avatar
 
Join Date: Feb 2014
Location: Pittsburgh, PA
Posts: 529
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 200 Post(s)
Liked: 66
What kind of measurement setup are you using to test your results?
Samps is online now  
Old 12-31-2014, 11:16 AM
AVS Special Member
 
Jay1's Avatar
 
Join Date: Aug 2002
Location: West Valley, AZ
Posts: 4,199
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 89 Post(s)
Liked: 133
I've never done an active setup, but I imagine with a CD you would want to put a resistor in parallel with the +/- terminals in order to squash the impedance peaks at FS. Usually tweeters will "squak" at you if you play them into uncompensated impedance peaks. Something like a 15 ohm resistor should have a minimal effect on everything else.
Jay1 is offline  
Old 12-31-2014, 11:22 AM
Advanced Member
 
mtg90's Avatar
 
Join Date: Aug 2011
Location: Millbrook, IL
Posts: 962
Mentioned: 14 Post(s)
Tagged: 0 Thread(s)
Quoted: 108 Post(s)
Liked: 269
It's a fairly straightforward process. You first start with raw driver measurements, best to get the speaker elevated and outdoors for cleanest measurements. You don't have to worry too much about measuring off axis with this combo to determine the range of directivity overlap before starting crossover work. 900-1100hz should be about the range you want to crossover to get a good transition there. As long as your level is not too high you don't need any protection in front of the tweeter for measurements, compression drivers should handle 90dB sweeps no problem. Now being active you might want something there to protect from any DC thumps from the amps getting through to the CD so might as well stick a decent sized cap out in front, say 15-20uF. If you do use a cap in the final design you will want to get the driver measurements with it in place. The mic should probably be 1.5m away from the baffle for this speaker and placed between the two drivers. Once there measure the individual response of the woofer and compression driver. I usually run leads from the drivers out the box at some spot and use alligator clips on the ends of the speaker cable from my amp to connect to those when I am measuring.

One important bit of information needed before you start the crossover design is to know the acoustic offset of the drivers, this is the difference in the amount of time it takes for the sound from one driver to reach you vs. the other. For this design the compression driver is behind the woofer so it's sound will lag behind the woofers. Now there are different methods to measure this, I use REW for all my measurements and to measure this using REW I will while taking initial driver measurements wire the two driver together in parallel and measure them like that. I also take a second measurement with the drivers in parallel but with the tweeter wired out of phase. It is important that the mic stays in the same position while taking all of these initial measurements.

With that info you can import into you active crossover design software of choice, I don't do active stuff often so I have only played around with the active crossover design settings in Jeff Bagby's PCD. In that software the first step besides importing the driver measurements is to determine the acoustic offset using the parallel measurements. You compare the simulated summed driver response to the actual measured summed response while adjusting the Z offset of the woofer until they match. It should be somewhere between 2-4" in front of the compression driver.

Now you are ready to play with the active settings and get things EQ'ed and see how the different crossover slopes effect the response and phase. When you have something that you think looks good plug those settings into your active crossover and take a measurement things line up and if they do have a listen. Then it's the fun loop of adjust, measure, listen and repeat until you have it dialed in to your liking.

I kind of breezed over it, you could fill a book with trying to cover every detail but hopefully you can see the steps to get there.
Scott Simonian, Mfusick and Face2 like this.
mtg90 is online now  
Old 12-31-2014, 12:08 PM
AVS Special Member
 
augerpro's Avatar
 
Join Date: Aug 2005
Location: Denver, Colorado
Posts: 1,726
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 19 Post(s)
Liked: 29
mtg90 covers it pretty well so I'll just give a couple thoughts. Try to use a software that uses a dual channel MLS measurement. This way you can measure actual relative phase and not have to reverse engineer your acoustic offsets. Just saves time and takes out the guess work.

The parallel resistor on the CD is good suggestion. I used to do this but never got around to checking whether it made an audible difference, thought that would probably be amp dependent.

I personally wouldn't bother with a protection cap, but if you do, if the crossover modeling software can do both active and passive (like Soundeasy) than you can measure without the cap. When you model the crossover you can model what size cap provides protection and still works with the active filters to accomplish your target response.

I would stick with LR acoustic response, probably LR4 at first, maybe LR6. Once the driver responses are on target adjust the delay to get the deepest reverse null in the modeler. Swap polarity back to normal and you should have good phase integration.

You can load 4 crossover versions into the minidsp, so model a few different crossover points, or the same point but different slopes. Flip back and forth to see which you like. This step is invaluable.
augerpro is offline  
Old 12-31-2014, 12:57 PM
AVS Special Member
 
augerpro's Avatar
 
Join Date: Aug 2005
Location: Denver, Colorado
Posts: 1,726
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 19 Post(s)
Liked: 29
BTW the most critical piece is the measurement. You must start with good data. Also look into Bagby or Soundeasy's method on producing quasi-anechoic frequency responses down to 10hz. The basic idea is to take a nearfield measurement of the woofer, add the effect of baffle diffraction to the response, measure the port and sum with the woofer response, take a far field measurement of the woofer, now splice the far field with the constructed near field response at a frequency that depends on your signal gating.
augerpro is offline  
Old 12-31-2014, 03:40 PM
Senior Member
 
Nate Hansen's Avatar
 
Join Date: Aug 2012
Location: Rochester, MN
Posts: 255
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 76 Post(s)
Liked: 58
Quote:
Originally Posted by Jay1 View Post
I've never done an active setup, but I imagine with a CD you would want to put a resistor in parallel with the +/- terminals in order to squash the impedance peaks at FS. Usually tweeters will "squak" at you if you play them into uncompensated impedance peaks. Something like a 15 ohm resistor should have a minimal effect on everything else.
Is this really necessary with an active crossover? I remember reading that it wasn't, but I don't remember why. I use compression drivers in an active setup and there's no "squak" to be heard.



I'll second what augerpro said about getting good data. Garbage in = garbage out. Like mtg90, I find I get the best results outdoors, but if you can't do that gated indoor measurements will suffice for the xo. When I do this I clear the room out as much as I can, get the speaker on a stand to get the design axis halfway between the floor and the ceiling, and I've got several batts of bonded logic I like to throw on the floor so it's about 2' deep. Also, having the speaker and mic at a diagonal across the longest dimension of the room will give you the greatest distance to the side walls. With an 8' ceiling I get about 3ms of clean impulse. This short gate will give the appearance of a high pass much higher than the speaker actually rolls off, and has the effect of smoothing the data <1khz.

I'll also second what augerpro said about using dual channel measurements, though I use REW which is a swept-sine measurement and not MLS. This makes it easy to look at the phase relationship of the drivers and determine exact delays. The reverse null method works with LR type crossovers but I like to see good phase tracking on either side of the xo, and I can calculate the delay by the difference in phase angle at the xo. For a beginner, maybe the reverse null method is best .

Another nice this about REW is the eq section. It has most of what you need to "simulate" the filters you will be using. ARTA is good for designing crossovers, because it has an overlay feature that will overlay various xo filters (LR, BW, etc) at whatever level and frequency you choose.
Nate Hansen is offline  
Old 12-31-2014, 03:47 PM
AVS Special Member
 
Jay1's Avatar
 
Join Date: Aug 2002
Location: West Valley, AZ
Posts: 4,199
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 89 Post(s)
Liked: 133
I've never ran active so I cant speak to what you are capable of doing. If you can control the drivers impedance, or have the response sharply cut before the impedance issues then it wouldnt be necessary.

I only deal with passives and it's a common requirement for that process. Feel free to ignore my comments
Jay1 is offline  
Old 12-31-2014, 04:44 PM
Senior Member
 
Nate Hansen's Avatar
 
Join Date: Aug 2012
Location: Rochester, MN
Posts: 255
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 76 Post(s)
Liked: 58
I don't want to ignore you.......I just want a better understanding of the situation, and it's relevant to the op.

Here's a thread from DIYA: http://www.diyaudio.com/forums/multi...ver-right.html Check out speaker dave's post #14 . For those that don't know, speaker dave is David Smith formerly of JBL, McIntosh, Snell and I believe now at Bose. The jist is......impedance compensation is not necessary in an active xo, though "might" be beneficial for certain tube amps.
Nate Hansen is offline  
Old 12-31-2014, 05:25 PM
AVS Special Member
 
Jay1's Avatar
 
Join Date: Aug 2002
Location: West Valley, AZ
Posts: 4,199
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 89 Post(s)
Liked: 133
With respect to Speaker Dave he was talking about a zobel which flattens the top of a drivers impedance, and can be used to bring a rising response down and/or simplify passive circuits.

The impedance peak(s) directly related to a drivers Fs are not going to be effected by a zobel usually.
Jay1 is offline  
Old 12-31-2014, 05:59 PM
Senior Member
 
Nate Hansen's Avatar
 
Join Date: Aug 2012
Location: Rochester, MN
Posts: 255
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 76 Post(s)
Liked: 58
Clearly I know little about passive crossovers. Thanks for the clarification, I'll keep reading.
Nate Hansen is offline  
Old 12-31-2014, 06:01 PM - Thread Starter
AVS Addicted Member
 
Mfusick's Avatar
 
Join Date: Aug 2002
Location: Western MA
Posts: 25,132
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 240 Post(s)
Liked: 1179
Active !!!!
Mfusick is offline  
Old 12-31-2014, 06:02 PM
AVS Special Member
 
augerpro's Avatar
 
Join Date: Aug 2005
Location: Denver, Colorado
Posts: 1,726
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 19 Post(s)
Liked: 29
Thanks Nate for the link. Gotta love diyaudio. Just as vicious as AVS but with better articulated positions

Mfusick I just got feedback from Jeff Bagby that ARTA and HOLMimpulse will do dual channel MLS. From your posting over in the HTPC forum I know you have the technical ability and like to really jump in, I would suggest getting Soundeasy. Not free, but by far the most fully featured software out there. If you do get it let me know and I'll email you some quick setup info.
augerpro is offline  
Old 12-31-2014, 06:08 PM
AVS Special Member
 
augerpro's Avatar
 
Join Date: Aug 2005
Location: Denver, Colorado
Posts: 1,726
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 19 Post(s)
Liked: 29
Regarding the shunt resistor in an active crossover, I got the idea from Lynn Olson I think. Or at least from his monster thread at diyaudio. It is only for compression drivers, since their impedance swings are pretty extreme. I doubt a modern solid state amp would have a problem, and as I said never did an A/B to see if it was an audible improvement. But someday when you have these speakers up and running it is easy enough to tie it into the CD wires with a switch in your lap and flip back and forth to see if you hear an improvement.
augerpro is offline  
 
Thread Tools


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off