I have a couple of observations. The first is that there are advantages of particular filter shapes, there is no mention of the advantages or disadvantages of a Likwitz-Riley filter with regards to directivity or summation. Typically Linkwitz-Riley is recommended for the scenarios it gets recommended for due to the effect it has on the polar response through the crossover when the source of the lower frequencies are from below the source of the higher frequencies. If we only look at phase and magnitude graphs from one position it’s easy to think that this is all that matters.
I now use the Linkwitz transform to produce high pass filters for speaker drivers. This works for woofers, midrange and tweeter equally well. A second order Linkwitz transform allows you to perfectly compensate for the natural speaker highpass Q and F3 and effectively move it to the desired second order highpass frequency and Q that you desire for your crossover. If you then want a fourth order crossover, simply cascade another 2nd order high pass filter to achieve that or any other result. No guess work or tweeking required. You simply measure the existing response (F3 and Q) and compute the filter that actually compensate the driver perfectly with the Linkwitz transform.
Question: When you say 'polarity' inversion... I assume that means to only one of the two drivers. Is the result the same if you invert the polarity of the woofer, as if you had inverted the polarity of the high frequency driver? Or is there an industry standard that one always inverts? Can you explain?
Polarity inversion is applied to one of the two outputs you plan to combine. No industry standard. My personal preference is to invert whatever output will have the least effect on the rest of the system.
I can only speak from a car audio perspective... All relevant persons with any knowledge (like Andy Wehmeyer from Audiofrog) always talk about acoustical alignments when refering to LR24db filters. They always say use whatever filter you need to get the slope in the crossover to match the acoustical LR24db alignment as close as you can, or, linearize the driver in question as flat as possible at least one octave below/above (depending if talking about Lowpass or Highpass filter) the desired crossover point and then use an electrical LR24db filter.
@@nathanlively Both works equally well in a car audio environment. When you have a perfect linear acoustical response an electrical 24db LR filter gives you an acoustical 24db LR filter. But there may be reasons to choose other electrical filter slopes and do the EQ afterwards. Maybe you need too much boost in the electrical domain to get a linear response, especially when you're at the stretch of the passband and in the natural roll off of the speaker f.e. Therefor you always measure first and look how "bad" the acoustical response is and then simulate different electrical slopes and the necessary EQ work in REW. Putting this into WinISD f.e. can show you if you exceed the Xmax of the driver in question or not, gives you the estimated power needed in the passband, shows you an estimated group delay graph... Based on all this you can make decisions about electrical filters to suit your acoustical environment.
Excellent video. You have quite a knack for really getting the information across without over complicating the message. That really helped me have my ah-ha moment by rethinking all of my past experiences in tuning car audio. Thank you.
I think your title is a little misleading… it should read stop recommending generic filter types… And then “use whatever filter type you need to make the crossover fit the crossover curve type and slope you require”
Thanks Adam. Great point. The message is really that there is no one filter that will work all of the time under all circumstances. How about: 1. For every season, there is a filter. 2. There is no one right filter. 3. The best filter achieves the desired result.
Superb! I should had seen this before I tried out DSP. Some DSP software recommends all LR4 filters and equalization. Looks great on graph, but sounds horrible to the ears.
Hello Nathan, do you think this method is applicable for 'in room' measurements at listenning position (As I saw that the measurements you are using are quasi-anechoid) ? Or it's more a PA way of working ?
Hey Julien, I definitely prefer near anechoic for this work, but the method will work with any data. The only thing that changes is my confidence level. :) That being said, can you not start by taking a measure very close to the speaker to remove as much of the room as possible?
Hey Julien, I have updated the article and attempted to answer your question there. Let me know what questions come up: www.sounddesignlive.com/stop-recommending-linkwitz-riley-filters/
In fact I'm asking myself if it's relevant to use anechoic data for in-room alignement of sub + main in a Small room, because for bass till Fs we know that the room dominate ... Actually, I'm testing some way of doing that very pragmatically: I measure in room response with MMM method at listenning position for each ways in order to define PEQ et Filters needed to match acoustically the electrical targeted filter (Saying LR24). Then I start with delay corresponding to physical distances and I'm adjusting it till a MMM control measurement show that I've reach the max SPL level at Fc. (Or an inverse show a deapest null)
Quick question.... It may be silly, maybe not. What if a Butterworth electrical filter gets you even closer to a nice Linkwits Riley acoustic target, would the outcome be the typical Linkwits Riley summation and phase shift (I think I'm saying that right)? Or, is using a different electrical filter type to achieve an LR4 going to cause issue at the crossover for phase? I have the ability to measure both, but I'm still learning and don't want to go down a rabbit hole if there is an easy answer.
Hey Tickle, Yes, I believe you're right. Whatever you do, as long as it hits the magnitude and phase targets that you want, you'll get the expected summation result.
Im confused as to what the purpose of this exercise is. Is the goal that once you are satisified with this simulated model, you mash a button and it spits out a schematic? You then built an actual prototype circuit, to then listen to the result? With all of those special parametric tweaks youve applied, I can imagine these are not trivial circuits...
I didn't get a transition from an amplitude to a power... You imported an acoustical data and continued to work with electrical amplitude graphs. But.. When -3dB changed to -6dB pls?
Using only a -6dB/oct LPF on a subwoofer just to fullfill the shape of the LR 36dB/oct works, but then you feed quite much out of band signal into the Voice Coil of the woofer, and that results in a higher temperature / less usefull power handling. I would rather recommend to use/ add a higher order LPF (additionally) at a higher frequency, so it does not disturb the Amplitude Frequency plot, and compensate with EQ/APF/ Delay to still match the Slope target.
Hi Nathan, will you be abel to do a Video how to use Live Avarage in Smaart V8 for alligning? The expected scenarios are: 1. Subs in a Line with Main L and Main R 2. Conventional L & R Sub configeration Can you please explain it step by step how you wwould to it in those two scenarios? Thanks a lot in Advance
Hi Priyan, I cover this in detail in these two workshops: 1. Intro to the Phase Graph - www.sounddesignlive.com/intro-phase-graph 2. Phase Alignment Science Academy - www.sounddesignlive.com/phase-alignment-science-academy/
Hey Priyan, I have updated the article and attempted to answer your question there. Let me know what questions come up: www.sounddesignlive.com/stop-recommending-linkwitz-riley-filters/
Very good recommendation for dealing with filters. I develop passive loudspeakers in a similar way. I measure the voltage frequency response of passive crossovers with CLIO (Audiomatica) and compare them with the acoustic results. But you should know exactly what properties filters have, otherwise it won't help much. Greetz from Switzerland
can't you just apply the necessery crossover and than correct the acoustical response to the target with eq ? this way you can change slopes and cross filters freely
no matter what speaker or room or cross over you and amp you use .because no speaker, amp ,cross over and Room are the same .its pointless to try to make the perfect Cross over .....just try to get it flat to a degree even with some dips and bumps as long as the avg is flat ..then eq it using a ref mic and listing position ..and use the most precise instrument YOU have your EARS
Yep, I just tried an active crossover with the Linkwitz filters, and it killed the sound coming out of my Maggies. Flat, lifeless, but the crossover worked. I went back to my non-Linkwitz pro audio crossover and we were back in business. I'm not a measurement guy, not with audio gear, for this very reason. So many folks go by measurements, just silly in my book, because in my experience, this is the result, crap sound.
Hi Nathan, love your content and I get heaps out of it myself. I’d love abit more information on time and phase aligning higher frequency drivers because for some reason I struggle a lot more to get my Bi amplified JBL am6212 aligned well when compared to aligning subs to mid drivers. Thanks
Hey Dane, I have updated the article and attempted to answer your question there. Let me know what questions come up: www.sounddesignlive.com/stop-recommending-linkwitz-riley-filters/
![5 ideas to improve Main+Sub phase alignment with Smaart® [GSwSST35]](/img/n.gif)
What is your favorite app to practice with filters at home?
REW... there's just no justification to spend hundreds or thoursands of $/€ if you're only setting up your own equipment.
Xsim & Rew
virtuixcad and boxsim
Boxsim and CLIO12
do I need polarity inversion as mandatory? in 2order
I have a couple of observations.
The first is that there are advantages of particular filter shapes, there is no mention of the advantages or disadvantages of a Likwitz-Riley filter with regards to directivity or summation. Typically Linkwitz-Riley is recommended for the scenarios it gets recommended for due to the effect it has on the polar response through the crossover when the source of the lower frequencies are from below the source of the higher frequencies.
If we only look at phase and magnitude graphs from one position it’s easy to think that this is all that matters.
I now use the Linkwitz transform to produce high pass filters for speaker drivers. This works for woofers, midrange and tweeter equally well. A second order Linkwitz transform allows you to perfectly compensate for the natural speaker highpass Q and F3 and effectively move it to the desired second order highpass frequency and Q that you desire for your crossover. If you then want a fourth order crossover, simply cascade another 2nd order high pass filter to achieve that or any other result. No guess work or tweeking required. You simply measure the existing response (F3 and Q) and compute the filter that actually compensate the driver perfectly with the Linkwitz transform.
Question: When you say 'polarity' inversion... I assume that means to only one of the two drivers. Is the result the same if you invert the polarity of the woofer, as if you had inverted the polarity of the high frequency driver? Or is there an industry standard that one always inverts? Can you explain?
Polarity inversion is applied to one of the two outputs you plan to combine. No industry standard. My personal preference is to invert whatever output will have the least effect on the rest of the system.
I can only speak from a car audio perspective...
All relevant persons with any knowledge (like Andy Wehmeyer from Audiofrog) always talk about acoustical alignments when refering to LR24db filters.
They always say use whatever filter you need to get the slope in the crossover to match the acoustical LR24db alignment as close as you can, or, linearize the driver in question as flat as possible at least one octave below/above (depending if talking about Lowpass or Highpass filter) the desired crossover point and then use an electrical LR24db filter.
Great! And that's been working well for you?
@@nathanlively Both works equally well in a car audio environment. When you have a perfect linear acoustical response an electrical 24db LR filter gives you an acoustical 24db LR filter.
But there may be reasons to choose other electrical filter slopes and do the EQ afterwards. Maybe you need too much boost in the electrical domain to get a linear response, especially when you're at the stretch of the passband and in the natural roll off of the speaker f.e.
Therefor you always measure first and look how "bad" the acoustical response is and then simulate different electrical slopes and the necessary EQ work in REW.
Putting this into WinISD f.e. can show you if you exceed the Xmax of the driver in question or not, gives you the estimated power needed in the passband, shows you an estimated group delay graph... Based on all this you can make decisions about electrical filters to suit your acoustical environment.
Excellent video. You have quite a knack for really getting the information across without over complicating the message. That really helped me have my ah-ha moment by rethinking all of my past experiences in tuning car audio. Thank you.
Glad it was helpful!
I think your title is a little misleading… it should read stop recommending generic filter types…
And then “use whatever filter type you need to make the crossover fit the crossover curve type and slope you require”
Thanks Adam. Great point. The message is really that there is no one filter that will work all of the time under all circumstances. How about:
1. For every season, there is a filter.
2. There is no one right filter.
3. The best filter achieves the desired result.
Superb! I should had seen this before I tried out DSP. Some DSP software recommends all LR4 filters and equalization. Looks great on graph, but sounds horrible to the ears.
Glad it was helpful!
This happens when you compare apples with lemons
Hello Nathan, do you think this method is applicable for 'in room' measurements at listenning position (As I saw that the measurements you are using are quasi-anechoid) ? Or it's more a PA way of working ?
Hey Julien, I definitely prefer near anechoic for this work, but the method will work with any data. The only thing that changes is my confidence level. :)
That being said, can you not start by taking a measure very close to the speaker to remove as much of the room as possible?
Hey Julien, I have updated the article and attempted to answer your question there. Let me know what questions come up: www.sounddesignlive.com/stop-recommending-linkwitz-riley-filters/
In fact I'm asking myself if it's relevant to use anechoic data for in-room alignement of sub + main in a Small room, because for bass till Fs we know that the room dominate ...
Actually, I'm testing some way of doing that very pragmatically: I measure in room response with MMM method at listenning position for each ways in order to define PEQ et Filters needed to match acoustically the electrical targeted filter (Saying LR24).
Then I start with delay corresponding to physical distances and I'm adjusting it till a MMM control measurement show that I've reach the max SPL level at Fc. (Or an inverse show a deapest null)
Quick question.... It may be silly, maybe not. What if a Butterworth electrical filter gets you even closer to a nice Linkwits Riley acoustic target, would the outcome be the typical Linkwits Riley summation and phase shift (I think I'm saying that right)? Or, is using a different electrical filter type to achieve an LR4 going to cause issue at the crossover for phase? I have the ability to measure both, but I'm still learning and don't want to go down a rabbit hole if there is an easy answer.
Hey Tickle, Yes, I believe you're right. Whatever you do, as long as it hits the magnitude and phase targets that you want, you'll get the expected summation result.
Im confused as to what the purpose of this exercise is. Is the goal that once you are satisified with this simulated model, you mash a button and it spits out a schematic? You then built an actual prototype circuit, to then listen to the result? With all of those special parametric tweaks youve applied, I can imagine these are not trivial circuits...
Hey Greg, these are the standard tools available in an output processor.
anyone else notice how the high frequency suffered at 4:25 ?
What’s the software you used called?
Crosslite+?
I didn't get a transition from an amplitude to a power... You imported an acoustical data and continued to work with electrical amplitude graphs. But.. When -3dB changed to -6dB pls?
Thanks Nathan :)
Using only a -6dB/oct LPF on a subwoofer just to fullfill the shape of the LR 36dB/oct works, but then you feed quite much out of band signal into the Voice Coil of the woofer, and that results in a higher temperature / less usefull power handling. I would rather recommend to use/ add a higher order LPF (additionally) at a higher frequency, so it does not disturb the Amplitude Frequency plot, and compensate with EQ/APF/ Delay to still match the Slope target.
Hey, do you know of any moderately priced servo driven subs for home audio? I might want to get one.Thanks!😊
Hey M, unfortunately I don’t know much about home theater. In this case, I don’t even know what a servo driven sub is.
Hi Nathan,
will you be abel to do a Video how to use Live Avarage in Smaart V8 for alligning?
The expected scenarios are:
1. Subs in a Line with Main L and Main R
2. Conventional L & R Sub configeration
Can you please explain it step by step how you wwould to it in those two scenarios?
Thanks a lot in Advance
Hi Priyan, I cover this in detail in these two workshops:
1. Intro to the Phase Graph - www.sounddesignlive.com/intro-phase-graph
2. Phase Alignment Science Academy
- www.sounddesignlive.com/phase-alignment-science-academy/
Hey Priyan, I have updated the article and attempted to answer your question there. Let me know what questions come up: www.sounddesignlive.com/stop-recommending-linkwitz-riley-filters/
Very good recommendation for dealing with filters. I develop passive loudspeakers in a similar way. I measure the voltage frequency response of passive crossovers with CLIO (Audiomatica) and compare them with the acoustic results. But you should know exactly what properties filters have, otherwise it won't help much. Greetz from Switzerland
can't you just apply the necessery crossover and than correct the acoustical response to the target with eq ?
this way you can change slopes and cross filters freely
Thank's! So how to get my model into my dcx2496 crossover?
Hi aman533! I'm...confused. I'm not familiar with the dcx2496. Does it include models? Of what?
no matter what speaker or room or cross over you and amp you use .because no speaker, amp ,cross over and Room are the same .its pointless to try to make the perfect Cross over .....just try to get it flat to a degree even with some dips and bumps as long as the avg is flat ..then eq it using a ref mic and listing position ..and use the most precise instrument YOU have your EARS
It worked for me!
LR filter
very helpful thank you Nathan !
Yep, I just tried an active crossover with the Linkwitz filters, and it killed the sound coming out of my Maggies. Flat, lifeless, but the crossover worked. I went back to my non-Linkwitz pro audio crossover and we were back in business. I'm not a measurement guy, not with audio gear, for this very reason. So many folks go by measurements, just silly in my book, because in my experience, this is the result, crap sound.
Hi Nathan, love your content and I get heaps out of it myself. I’d love abit more information on time and phase aligning higher frequency drivers because for some reason I struggle a lot more to get my Bi amplified JBL am6212 aligned well when compared to aligning subs to mid drivers. Thanks
Hey Dane, tell me more. What are some of your specific questions or challenges in this area? Can you send me your measurements and some photos?
Hey Dane, I have updated the article and attempted to answer your question there. Let me know what questions come up: www.sounddesignlive.com/stop-recommending-linkwitz-riley-filters/
It was useful, thank you!
You're welcome!
Oh interesting!
Amazing
Excel
What?
Amazing