It is known that combining signals that are of different phase is not a problem. It changes the level of the combined response. When you get close to 0 degrees difference, some change makes almost no difference to the level. When you approach 180 degrees, the null comes in much faster but we never use it like that.
You ask for documentation but this is an advanced topic that needs much other work done before. May I suggest if you are interested then you start learning about three dimensional crossovers.
You ask for documentation but this is an advanced topic that needs much other work done before. May I suggest if you are interested then you start learning about three dimensional crossovers.
MrHIfi, three dimensional crossover AllenB mentions = The acoustic output of the system!
Making crossover with just on axis measurements ignores 99% of the speaker output (to other directions). While on axis response is important it is not what the response is at listening spot in a typical room few meters away from the speaker.
There is literally xillion ways to make the on axis response nice if you look at the on axis response only, in crossover simulator. if you had measured the response to all directions and looked at that instead you'd get much more complete view on what the speaker response is to all directions and estimate way better what it is at the listening spot in a room few meters away! Suddenly you notice there is not xillion ways to make the xo but only a handful. You are tuning the output of the whole system since the output is 3D you cant tune individual directions alone (in electrical domain, in the xo).
Trust me, using the on axis data only you literally cannot see what the response actually is in the room. it is like being lost in the dark woods, any direction will look like the way to home! Popup your google maps on the smartphone and you see exactly where you are at and which is the right direction. Clear as day, a trivial task. Voodoo in crossovers is because of this. There literally is no voodoo after you have the whole picture of the system output (to all directions) in front off your eyes in the simulator.
Hope it helps. Read VituixCAD measurement manual.
Making crossover with just on axis measurements ignores 99% of the speaker output (to other directions). While on axis response is important it is not what the response is at listening spot in a typical room few meters away from the speaker.
There is literally xillion ways to make the on axis response nice if you look at the on axis response only, in crossover simulator. if you had measured the response to all directions and looked at that instead you'd get much more complete view on what the speaker response is to all directions and estimate way better what it is at the listening spot in a room few meters away! Suddenly you notice there is not xillion ways to make the xo but only a handful. You are tuning the output of the whole system since the output is 3D you cant tune individual directions alone (in electrical domain, in the xo).
Trust me, using the on axis data only you literally cannot see what the response actually is in the room. it is like being lost in the dark woods, any direction will look like the way to home! Popup your google maps on the smartphone and you see exactly where you are at and which is the right direction. Clear as day, a trivial task. Voodoo in crossovers is because of this. There literally is no voodoo after you have the whole picture of the system output (to all directions) in front off your eyes in the simulator.
Hope it helps. Read VituixCAD measurement manual.
Last edited:
@MrHifiTunes I should have been more accurate and said there isn't much documentation available saying it isn't necessary.. however if you look at the Butterworth family of filters they have a 90 degree phase difference at the crossover.
Yes I fully understand those things. Like you said before If it doenst match 100% it will effect the combined output. Also off axis response will have influence on it so at listing position the perception will or will not be influenced.
What I try to diggest is :
deep null is only at XO point. Drivers have overlap where phase isn't optimum. Hence off axis will have their contribution? If one use different cross over types.(LR / Butterworth or Bessel) what is the influence of those topologies. Is one better then the other in maintaining max output on and off axis? What are the benefits/downside to use one or an other?
Some practical question on VituixCad.
I have traced IEC baffle FRD and ZMA files. Can I use those to sim the XO? Or should I first make 4pi with VituixCad diffraction tool and use those files.
Indeed. But I was referring to the flat response and power response. I was of the understanding that you only get a flat response with LR filter, if you chose 4' order. This is why I asked about the order of the filter.
All it takes for a flat response is that the two add up at the crossover to the same level as their passband. Can you think of a reason why that isn't possible for odd orders? I can get a flat response for any order, and for Butterworth or LinkwitzRiley.
However then Linkwitz Riley has a power dip and Butterworth has flat power.
Do you have concerns about one of these statements more than the other, or is it both?
However then Linkwitz Riley has a power dip and Butterworth has flat power.
Do you have concerns about one of these statements more than the other, or is it both?
It's the power dip that I can't understand. Because I have been running LR4 here for many years, and a power dip to me, is a dip in the off-axis response, when the on-axis is flat. But then I read the following, and found that maybe I need to re-read everything again - even though my system plays great as is:
https://techtalk.parts-express.com/...142-uniform-power-response-best-way-to-get-it
https://techtalk.parts-express.com/...142-uniform-power-response-best-way-to-get-it
LR xo has dip in vertical off-axis response, unlike (a well done) asymmetrical or BW. This is because of timing change (phase sync).
Lateral dip can come from increasing directivity of low-passed driver, but timing remains same as on-axis. Asymmetric xo can compensate this.
(This applies to traditional vertical WMT constructions)
Power response is combination of lateral and vertical off-axis responses, weighted on horizontal.
Lateral dip can come from increasing directivity of low-passed driver, but timing remains same as on-axis. Asymmetric xo can compensate this.
(This applies to traditional vertical WMT constructions)
Power response is combination of lateral and vertical off-axis responses, weighted on horizontal.
Looking back over the many projects I've done I've never made a symmetrical XO, the exception being when using the Behringer EXO units.
So having the same slopes and order on both drivers is a non-issue to me as is off-axis as I only listen on-axis.
Other peoples rooms might however make a much bigger difference than my poor listening area and most of my speakers sound better in the open air.
So having the same slopes and order on both drivers is a non-issue to me as is off-axis as I only listen on-axis.
Other peoples rooms might however make a much bigger difference than my poor listening area and most of my speakers sound better in the open air.
All the sound comes back to you one way or another regardless of which axis you listen on. The more similar a reflection is to the direct sound the easier it is for your brain to recognise it as a reflection. Outside or with absorption at reflections points stops a lot to most of the sound coming back leaving you in a stronger direct sound field which is probably one reason why you prefer it.
Flat on axis and smooth off axis is what the good Dr Toole would recommend.
This is only true if you have your speakers in a box, while this is what we mainly build it isn't true in all scenarios. Also it ignores the fact that we hear better at certain frequencies related to screaming children and a balanced sound reproduction might want a slightly different emphasis; that is perhaps more bass to counter our poor hearing at bass frequencies.
I'm not saying you are wrong; simply it's not that simple
I'm not saying you are wrong; simply it's not that simple
This is not the reason @digitalthor was asking about. The reason Butterworth has theoretically flat power and axial response at the same time is due to keeping the phase difference as mentioned at the beginning of this part of the conversation.
I wasn't sure I needed to highlight this but it looks as though this factor became lost in the conversation.
It's true for all speakers but the directivity pattern of them will dictate where and when the sound returns from, box dipole or something else. The determination in the brain of whether something is a reflection or contributes to timbre modification is quite different and which frequencies are perceived more loudly is separate.
The general preference for a gentle down sloping in room response coincides nicely with a flat direct sound and rising directivity. The smoother the off axis the smoother the in room response will be.
I re-read my messages and can see, that it's me who wrote before I was thinking 
I chose the LR filters because I read the following - and found it the best suited for my active speaker project - with focus on smooth flat response and minimum phase shift:
https://www.ranecommercial.com/legacy/note160.html
It seems like, when I chose a LR4 filter in my DSP - it only works properly, when I use my drives within their linear pass band. So if I selected a driver that already start to roll off, before or during the crossover region. Then I need to choose another electrical slope, to reach the desired acoustical slope.
This is also the reason why I usually choose drivers that have a wider smooth bandwidth, rather than trying to make them "stretch" . At the moment it's 2 x SB23NRX, MW13TX and Seas DXT. Considering 2 x WO24 to have the best and most smooth drivers to work with. Even with a DSP, I find less EQ to better. I strive to achieve the best result with minimum hassle. Ironing a driver flat with a DSP, rarely sounds the best.
I chose the LR filters because I read the following - and found it the best suited for my active speaker project - with focus on smooth flat response and minimum phase shift:https://www.ranecommercial.com/legacy/note160.html
It seems like, when I chose a LR4 filter in my DSP - it only works properly, when I use my drives within their linear pass band. So if I selected a driver that already start to roll off, before or during the crossover region. Then I need to choose another electrical slope, to reach the desired acoustical slope.
This is also the reason why I usually choose drivers that have a wider smooth bandwidth, rather than trying to make them "stretch" . At the moment it's 2 x SB23NRX, MW13TX and Seas DXT. Considering 2 x WO24 to have the best and most smooth drivers to work with. Even with a DSP, I find less EQ to better. I strive to achieve the best result with minimum hassle. Ironing a driver flat with a DSP, rarely sounds the best.