It seems like so many problems of crossovers is due to the fact that the high/low passes are slopes, so the two drivers have some overlapping signals near the crossover, and this creates all sorts of problems like lobing, interference, phase shifts, cone resonance/breakup, tweeter protection, time alignment problem, etc etc etc. I thought we have to deal with this because of inherent limitations of analog components only able to roll off signals instead of brick walling them.
With the availability of DSP's, why can't crossovers be made with a brick wall? I'm not talking about 100dB/oct slopes, but complete brick wall. For example, if you have a 2000Hz crossover, all <2000Hz content goes to the woofer. If a signal is 2001Hz, it goes completely to the tweeter. No roll off slopes or anything. If you ask someone with no audio background on how to separate signals, that's exactly what they'd do, because it just makes intuitive sense. It also eliminates every single problem listed above with crossovers.
So why isn't this done?
With the availability of DSP's, why can't crossovers be made with a brick wall? I'm not talking about 100dB/oct slopes, but complete brick wall. For example, if you have a 2000Hz crossover, all <2000Hz content goes to the woofer. If a signal is 2001Hz, it goes completely to the tweeter. No roll off slopes or anything. If you ask someone with no audio background on how to separate signals, that's exactly what they'd do, because it just makes intuitive sense. It also eliminates every single problem listed above with crossovers.
So why isn't this done?
Last edited:
first of all it can be done.
but the result would be one huge horrid mess.
with no overlapping the source of the sound if it would excees the crossover point would jump from one place to an other.
it has been done in the past, if you don't mind i won't digg up the articles.
the end result was a system that guaranteed headaches.
your brain knows this would be not natural, and processing this kind of information simply made the listeners head ahce. for real.
within minutes.
not good.
so its not done as its useless.
so to make things sound good, you want drivers that overlap a lot, and use a 1st order filter. you want it to be an active one.
to avoid sufferring with impedance mismatch, and you want it to be digital so you can sort out delay correction without a tonnnnnnnnn of circuitry.
allso every mechanical resonating system has some ring down, a brick wall separator insted of a CROSS-OVER (ya, the name sortha suggest something if you are smart. like.. it ain't separating, but ->tricky part<- crossing over from one to an other/others <-end of tricky part-> ) would result in a bloody mess of beheaded children and cute fluffy toys. just to stay conservative 😀
but the result would be one huge horrid mess.
with no overlapping the source of the sound if it would excees the crossover point would jump from one place to an other.
it has been done in the past, if you don't mind i won't digg up the articles.
the end result was a system that guaranteed headaches.
your brain knows this would be not natural, and processing this kind of information simply made the listeners head ahce. for real.
within minutes.
not good.
so its not done as its useless.
so to make things sound good, you want drivers that overlap a lot, and use a 1st order filter. you want it to be an active one.
to avoid sufferring with impedance mismatch, and you want it to be digital so you can sort out delay correction without a tonnnnnnnnn of circuitry.
allso every mechanical resonating system has some ring down, a brick wall separator insted of a CROSS-OVER (ya, the name sortha suggest something if you are smart. like.. it ain't separating, but ->tricky part<- crossing over from one to an other/others <-end of tricky part-> ) would result in a bloody mess of beheaded children and cute fluffy toys. just to stay conservative 😀
i'm using 300db/octave linear phase for years and it's pretty much considered a ''brick wall''
i mean, 50db drop in 1/6 octave... that's kind of brutal, isnt it ?
i mean, 50db drop in 1/6 octave... that's kind of brutal, isnt it ?
This isn't that way because it's not natural. The history of lifeforms
on planet Earth teaches us that changes are gradual and only in that
sense the progress exists.
I don't mean any offense by this but these are somewhat naive questions that hint at your level of understanding in certain key subject areas. Answering this question is like trying to answer the question "why after more than 100 years of manufacturing refinement they cannot make cars that get over 50 miles per US gallon and produce only clean emissions"? (tenuous VW reference here). How do you explain that without invoking chemistry, physics, and economics? Even then, the answer is not a simple one.
To understand the answers to your questions about crossover slopes one needs to have a rather advanced level of knowledge and proficiency in mathematics, electronics, and acoustics. Mathematics is needed to understand how the "brick wall" filter can or cannot be realized, the different ways that filters can be formulated, and how the phase behavior is related (or not) to amplitude behavior. Electronics is needed to understand what can actually be accomplished using analog and digital circuits and where there are real limitations. For DSP, you would need mathematics and computer science to understand how DSP signal processing (e.g. filtering) algorithms are formulated and implemented. Acoustics is needed to understand why a brick wall filter may not be pleasant to listen to. When you have the appropriate tools and knowledge you can explore the reasons behind the question and decide for yourself what the "answer" shall be. I personally find the process of learning to be very rewarding and I strongly suggest that you consider that road yourself. Doing so can result in a stimulating, lifelong pursuit.
See: Give a man a fish and you feed him for a day; teach a man to fish and you feed him for a lifetime.
If that's not your style then you can always take a "get er dunn" approach (e.g. just try something and see if you like it) and don't worry about the "why".
Last edited:
Absolutes and engineering design don't mix (is that an absolute statement?)
Infinite slope means infinite time delay. I'd be happy to sell you a crossover like that, but on the understanding that when you input your signal into it, it will NEVER get to the other side to any of the drivers!
If you don't want to deal with the math, think of it this way: how long would you have to look a a sine wave to know that it is at 2000Hz and not at 2000+(infinitessimally more) ? A crossover would have to monitor it forever, to determine that too!
More practically, you could have a slope that is very high, though, as mentioned above (it would either have a lot of group delay variation or absolute delay, though). If done with coaxially arranged drivers like concentric or synergy horn, it can be done without location or abrupt polar pattern problems even. But why, what's the payoff?
Infinite slope means infinite time delay. I'd be happy to sell you a crossover like that, but on the understanding that when you input your signal into it, it will NEVER get to the other side to any of the drivers!
If you don't want to deal with the math, think of it this way: how long would you have to look a a sine wave to know that it is at 2000Hz and not at 2000+(infinitessimally more) ? A crossover would have to monitor it forever, to determine that too!
More practically, you could have a slope that is very high, though, as mentioned above (it would either have a lot of group delay variation or absolute delay, though). If done with coaxially arranged drivers like concentric or synergy horn, it can be done without location or abrupt polar pattern problems even. But why, what's the payoff?
If you prefer the mechanical world I can ask a similar question- Why can't we have a single driver that's large enough to produce good bass, but with very high speed and bandwidth, so it can reach 20 kHz. That would solve the crossover problem by eliminating it.
I have a lot of respect for you and your work in audio, but I must strongly respectfully disagree.
Your analogy is implying my question is something like "why can't we have 5'' drivers with 5'' of xmax with <1% distortion". However, what I'm asking is questioning the *fundamental* way we do things in audio, and why we are doing it this way when it has so many problems. So a closer, but not a perfect analogy would be someone asking 50 years ago "why are we dealing with combustion engines and all the problems with poor efficiency, torque, environmental problems, and all the little parts that wear and tear and fail and are unreliable, when we can just stick an electric motor and battery to power the car with way more efficiency, power, and no moving parts to break?"
Do you agree that the existence of a crossover that rolls off frequencies instead of brick walling them is the source of dozens of speaker design problems in audio such as phase shift and lobing errors?
So if so many problems arise due to a non brick wall crossover, why is it so naive to question the reason for using a literal Pandora's box?
I'm rather appalled by several immature responses in this thread to a man genuinely curious and wanting to learn. I expect better from this knowledgeable community.
"Give a man a fish and you feed him for a day; teach a man to fish and you feed him for a lifetime". Would you (and Arty) kindly post links on where I can educate myself on this topic?
Thank you. That's the kind of intelligent response I'm looking for. Explains perfectly why this can't be done. Now I have to research a bit on understanding the bolded part
OK - I am no expert - but have been reading around this for a while due to the release of the minidsp OpenDRC - DA8.
As I understand it several papers were published almost 10 years ago looking at the implementation of digital crossovers.
If IIR DSP crossovers are used, they essentially mimic analogue active crossovers in that an LR4 IIR DSP crossover will induce phase shift. It seems generally accepted that anything higher than LR8 or so, induces enough phase shift to be audible and therefore people tend to feel better sound is achieved with lower order slopes (I use BW3 on my DSP based crossover...)
However, FIR DSP filters allow phase to be independent of crossover slope, so you can easily have a "300db/octave linear phase" filter which has 'near brick wall' properties AND flat phase.
BUT there is no free lunch, the very high order FIR filters have two problems:
1 - latency (as mentioned above) - which is fine for home playback, but a real problem for live performances, recording studios etc
2 - Pre-ringing - again there have been papers published on the audibility of this - and it was generally felt that as long as the slopes were kept under 512dB/octave, the pre-ringing was not audible.
These linear phase very high order filters - 'brick wall' filters, if you like, have negligible lobing, and from the reading I have done appear to have little drawbacks if you are already using DSP.
Here is one of the papers I have recently read published in 2008 on the audibility of such high order linear phase crossovers -
Perceptual Study of Loudspeaker Crossover Filters. Henri Korhola
http://lib.tkk.fi/Dipl/2008/urn011933.pdf
There is bound to be resistance to this type of technology, just like the augments between passive vs active crossovers, DSP vs analogue etc, etc.
But for all the reading I have done, I am struggling to find a drawback to them in a properly designed speaker.
For instance I wouldn't choose two drivers with very different dispersion characteristics and cross them with a 'brick wall filter' - the power response would have no blending and a very obvious 'step'.
Also I wouldn't have distantly sited drivers - (which is much more possible with a brick-wall filter as there is no lobing caused by the C-C distance as there would be with an IIR filter) as the sound source would appear to 'jump' from one driver to another.
However, when I design my speakers, I always try to keep C-C distance minimized and dispersion of drivers matched at crossover points anyway, so I am still struggling to see the problem with very high order linear phase crossovers......
In short - I will likely soon by buying a miniDSP OpenDRC - DA8 to play with.
As I understand it several papers were published almost 10 years ago looking at the implementation of digital crossovers.
If IIR DSP crossovers are used, they essentially mimic analogue active crossovers in that an LR4 IIR DSP crossover will induce phase shift. It seems generally accepted that anything higher than LR8 or so, induces enough phase shift to be audible and therefore people tend to feel better sound is achieved with lower order slopes (I use BW3 on my DSP based crossover...)
However, FIR DSP filters allow phase to be independent of crossover slope, so you can easily have a "300db/octave linear phase" filter which has 'near brick wall' properties AND flat phase.
BUT there is no free lunch, the very high order FIR filters have two problems:
1 - latency (as mentioned above) - which is fine for home playback, but a real problem for live performances, recording studios etc
2 - Pre-ringing - again there have been papers published on the audibility of this - and it was generally felt that as long as the slopes were kept under 512dB/octave, the pre-ringing was not audible.
These linear phase very high order filters - 'brick wall' filters, if you like, have negligible lobing, and from the reading I have done appear to have little drawbacks if you are already using DSP.
Here is one of the papers I have recently read published in 2008 on the audibility of such high order linear phase crossovers -
Perceptual Study of Loudspeaker Crossover Filters. Henri Korhola
http://lib.tkk.fi/Dipl/2008/urn011933.pdf
There is bound to be resistance to this type of technology, just like the augments between passive vs active crossovers, DSP vs analogue etc, etc.
But for all the reading I have done, I am struggling to find a drawback to them in a properly designed speaker.
For instance I wouldn't choose two drivers with very different dispersion characteristics and cross them with a 'brick wall filter' - the power response would have no blending and a very obvious 'step'.
Also I wouldn't have distantly sited drivers - (which is much more possible with a brick-wall filter as there is no lobing caused by the C-C distance as there would be with an IIR filter) as the sound source would appear to 'jump' from one driver to another.
However, when I design my speakers, I always try to keep C-C distance minimized and dispersion of drivers matched at crossover points anyway, so I am still struggling to see the problem with very high order linear phase crossovers......
In short - I will likely soon by buying a miniDSP OpenDRC - DA8 to play with.
That's not entirely correct, in general. An FIR filter can independently vary the amplitude and phase responses.
This is a question that requires education in electrical engineering to get a grasp on, it's hard to explain without the background. The mathematics, physics and system theory that created analog and digital filters do not have a way to brick wall the signal as far as I know. Maybe start reading a book on signal processing, causality and Fourier transform to get started. If you come up with the mathematical solution to your question you might become very well off🙂
Sent from my iPhone using Tapatalk
Sent from my iPhone using Tapatalk
These are also some excellent articles on the matter:
AV: FIR-ward Thinking: Examining Finite Impulse Response Filtering In Sound Reinforcement Systems - Pro Sound Web
AV: A Useful Tool: Creating & Applying FIR Filters - Pro Sound Web
And a short quote from the above:
Proof In The Polars:
I spaced two loudspeakers 1 wavelength at 1 kHz (about 1.1 feet), shown in Figure 8.
Figure 8 – A horizontal polar was measured on the two devices. They were offset 1 wavelength at 1 kHz.
This configuration should produce a polar that looks like a shamrock on St. Patrick’s day, since the loudspeakers will be 180 degrees out-of-phase at 1 kHz at several angles around the polar, and in-phase at other angles. I used this configuration to evaluate the polar response at the crossover frequency of 1 kHz.
On the left in Figure 9, we see the polar response at crossover for the Linkwitz-Riley crossover network. The response lobes, because both transducers are “talking” in the crossover region. Since they are offset physically, lobing is unavoidable. As expected, the response resembles a four-leaf clover.
On the right is the polar response at crossover for the FIR crossover network. The lobing is eliminated because the offset transducers are not overlapping in frequency.
The transducers in multi-way loudspeakers have always been interdependent, and lobing has always been a big issue. The use of linear phase brickwall filters can eliminate lobing by allowing the passbands to behave independently. Given the axial transfer function and polar response, it is impossible to determine that this is a two-way system with offset transducers – pretty amazing.
AV: FIR-ward Thinking: Examining Finite Impulse Response Filtering In Sound Reinforcement Systems - Pro Sound Web
AV: A Useful Tool: Creating & Applying FIR Filters - Pro Sound Web
And a short quote from the above:
Proof In The Polars:
I spaced two loudspeakers 1 wavelength at 1 kHz (about 1.1 feet), shown in Figure 8.
Figure 8 – A horizontal polar was measured on the two devices. They were offset 1 wavelength at 1 kHz.
This configuration should produce a polar that looks like a shamrock on St. Patrick’s day, since the loudspeakers will be 180 degrees out-of-phase at 1 kHz at several angles around the polar, and in-phase at other angles. I used this configuration to evaluate the polar response at the crossover frequency of 1 kHz.
On the left in Figure 9, we see the polar response at crossover for the Linkwitz-Riley crossover network. The response lobes, because both transducers are “talking” in the crossover region. Since they are offset physically, lobing is unavoidable. As expected, the response resembles a four-leaf clover.
On the right is the polar response at crossover for the FIR crossover network. The lobing is eliminated because the offset transducers are not overlapping in frequency.
The transducers in multi-way loudspeakers have always been interdependent, and lobing has always been a big issue. The use of linear phase brickwall filters can eliminate lobing by allowing the passbands to behave independently. Given the axial transfer function and polar response, it is impossible to determine that this is a two-way system with offset transducers – pretty amazing.
its..quite possible to divide a digital sound in a way that would produce a brick wall filter, but it would not be called a filter to be honest. but a separator.
it would simply introduce far more problems than it would solve.
in theory even digital speakers are an option..
but it wold be a bad option anyways.
you would have problems with delay, with source forming.
you would not solve every lobing problem if you had a filter like this, as many lobing problems come from the box it self. so no crossover can help that.
impedance mismatch issues.. well you could solve it, but it can allready be solved without generating more issues.
to sum it up it would be pointless and would be bleeding from many wounds.
you should read up on how and why filters work.
your best bet is DSP, if you are on a cheaper side of thing pack up some opamps so you can make your own active filters.
its a lot of fun anyways.
it would simply introduce far more problems than it would solve.
in theory even digital speakers are an option..
but it wold be a bad option anyways.
you would have problems with delay, with source forming.
you would not solve every lobing problem if you had a filter like this, as many lobing problems come from the box it self. so no crossover can help that.
impedance mismatch issues.. well you could solve it, but it can allready be solved without generating more issues.
to sum it up it would be pointless and would be bleeding from many wounds.
you should read up on how and why filters work.
your best bet is DSP, if you are on a cheaper side of thing pack up some opamps so you can make your own active filters.
its a lot of fun anyways.
I 100% disagree - but that is the fun with DIY audio! Some people love full range drivers, some tubes, some class A, pick your poison!
It would be worth you reading some of the articles I have linked to though.
Lots of excellent, well funded and very polite answers above , you can´t feel hurt simply because they run against a preconceived idea you have.
FWIW I might have found the
FWIW I might have found the
bit self-explanatory of the situation, but since you could get angry, I won´t even mention it.
If I may steer the OP in a direction, I think it's better to ask, "at what order do we practically get a brickwall?" I mean a quick look at a 6th order Chebyshev pushes the ripples in the stopband down tremendously, and thus makes for a rather sharp transition.
Nice link: http://www.analog.com/media/en/technical-documentation/dsp-book/dsp_book_Ch20.pdf
(Then you can ask how important that really is)
Nice link: http://www.analog.com/media/en/technical-documentation/dsp-book/dsp_book_Ch20.pdf
(Then you can ask how important that really is)
Last edited:
It is simple to create a perfect brick wall filter. Get wav file of a complete song, take FFT, set all frequency values below crossover frequency to zero, inverse FFT and that is the time signal for the tweeter, set all frequency values above crossover frequency to zero, inverse FFT and that is the time signal for the woofer. Note the high latency, the horrible step change in directivity between the drivers, the potential for serious pre-ringing issues, etc... But it is a simple test to perform if you want to check how good an idea it may be.
- Status
- Not open for further replies.