Here is how to take this philosophy into practice: if one thinks crossover as mere task and that optimal is always within reach, the design space opens up and there is higher chance to optimize a playback system for a particular situation. Think you have ideal transducers on your hands, any size and number of, and optimal crossover at the end what ever system design you came up with, now what should I design, what's important?
So, how to think crossovers depends on what you are doing: if system design then think the crossover is optimal for what ever you have there, allocate resources to important things like how to make the system measure as you want and assume crossover will work out fine. When it's time to actually implement a crossover, then think what is optimal and how to actually make it happen. One might do several and compare. And there is feedback loop within the whole process of course, experience.
If one has limited the system design to something particular, like pre-existing equipment / parts, the design space narrows considerably, and situation is not ideal and has to be navigated backwards. This likely makes suboptimal end result in absolute terms, although likely works out just fine for what it is. There is always some practical limitations, like budget.
By backwards I mean from equipment perspective working towards perception, when ideally one starts with perception and works out the equipment.
Using ideal drivers and optimal crossovers as design option do not make them necessary or force to some particular solution. It's just design freedom. Like, after you have worked out the system that would work, you'd start to work on details, like can I build it, what it looks like, what the crossover is reqyired to accomplish, and so on. In the end one might end up with small two way speaker with passive crossover, or something else completely, like big two way, or more ways and DSP and passive xo and what ever isnrequired to meet the goal you saw clearly. Or at least work toward the goal, if it's too expensive then minimize the costs by building only psrt ofnit and rest later, or what ever feels reasonable for you with your project and how to reach a goal you set for it to make it a successfull project.
If one is set to build particular speaker, like two way speaker using trendy tweeter and trendy woofer, and make it work with existing equipment and particular budget it's likely a passive filter then, the project is set in stone, and not much of this high level stuff I wrote applies, and thats it. It's just philosophy, everyone can do what they want, hopefully it's fun as it's a hobby and it should always be fun time, whatever that is for each 🙂
So, how to think crossovers depends on what you are doing: if system design then think the crossover is optimal for what ever you have there, allocate resources to important things like how to make the system measure as you want and assume crossover will work out fine. When it's time to actually implement a crossover, then think what is optimal and how to actually make it happen. One might do several and compare. And there is feedback loop within the whole process of course, experience.
If one has limited the system design to something particular, like pre-existing equipment / parts, the design space narrows considerably, and situation is not ideal and has to be navigated backwards. This likely makes suboptimal end result in absolute terms, although likely works out just fine for what it is. There is always some practical limitations, like budget.
By backwards I mean from equipment perspective working towards perception, when ideally one starts with perception and works out the equipment.
Using ideal drivers and optimal crossovers as design option do not make them necessary or force to some particular solution. It's just design freedom. Like, after you have worked out the system that would work, you'd start to work on details, like can I build it, what it looks like, what the crossover is reqyired to accomplish, and so on. In the end one might end up with small two way speaker with passive crossover, or something else completely, like big two way, or more ways and DSP and passive xo and what ever isnrequired to meet the goal you saw clearly. Or at least work toward the goal, if it's too expensive then minimize the costs by building only psrt ofnit and rest later, or what ever feels reasonable for you with your project and how to reach a goal you set for it to make it a successfull project.
If one is set to build particular speaker, like two way speaker using trendy tweeter and trendy woofer, and make it work with existing equipment and particular budget it's likely a passive filter then, the project is set in stone, and not much of this high level stuff I wrote applies, and thats it. It's just philosophy, everyone can do what they want, hopefully it's fun as it's a hobby and it should always be fun time, whatever that is for each 🙂
Last edited:
Here is a paper by Scott Hinson where he discusses various the various attributes that need consideration in a crossover design, that might add some enlightenment: https://drive.google.com/file/d/11r...jG-VcdvEJU4nkuvMtw_aem_qjX63cqwc9nvVdGQxj-viA
Hi A4eaudio, I'll try to answer your questions and explain my viewpoints.
This leads me to a decision on a speaker type...conventional box, line array, MEH, open-baffle, etc.
Whatever design is chosen, I have to decide how many driver sections will be used to achieve the SPL goal and bass extension, and how their radiation patterns will merge smoothly.
Then sims help produce plans for a proto-type. (But sims help less than paying basic attention to driver c2c distances, baffle/horn widths, and displacement capabilities of driver sections at their lowest frequency range of operation.)
And then build the proto-type, which is always work, but mainly labor, and not so much as mental work. So not hard.
Through completion of the proto, I haven't given crossovers 30 seconds of thought, other than knowing the approximate xover points based on the acoustic design.
So what's been hardest till now? 🙂
Well. assuming I don't have a measurement rig /spinorama in my back pocket, I'll first need to build one. For what size speaker? so that it carry the weight, and how big can the turntable be before it becomes a diffraction problem on its own?
Then, once the spinorama is built, which driver section should be the key section, having its point-of-rotation directly over the center of the turntable?
How do I handle the inevitable timing differences between the key section and other driver sections as I rotate and take measurements?
Even bigger question...at what distance do I measure? What is the minimum measurement distance for onset of the acoustic far-field?
Can I do that distance indoors, in a reflection free enough environment, or do I need to move outdoors. If indoors, can I live with the kludge of gating to reduce reflections' influence, or the inaccuracy of nearfield acoustic meging techniques? If I go outdoors, can I live with the noise and wind in my measurements? Should I do ground plane measurements, or flown as high off the ground as possible?
What will equate to what I hear?
We can read about sound forever, we can sim it forever .......
I've never yet heard a sim make music, until I build a speaker trying to match the sim.
And then once built, how do I fix what I don't like or question? Other than take measurements, that I believe I can act on.
So in the end it's back to measurements.
But hey, I really think there's been a swerve into sims vs measurements that wasn't intended...I certainly didn't mean to go there...
Topic in my mind is more about the comparative difficulty of crossovers vs measurements.
Frankly, and I know this will draw criticism....the reason crossovers are seen as difficult is because of one word...PASSIVE.
What an unnecessary complicated path imho, and worse leaves attainable SQ on the table, for all but the simplest of speaker designs.
(Flame suit on. 🙂
Active with the right software tools can generate near perfect complementary crossovers in about 5 minutes or less, and can separate fixed time alignment (using precise digital delays) from mag and phase alignment. Either through IIR or FIR.
All based on measurements..so that tuning work is final, and only dependent of the quality of the measurements.
Just my proven method...how can I not see measurements as more difficult work than crossovers??
For me, a DIY speaker starts with what size room is it for, what radiation pattern do I want from it, and then how loud and how low do I want it to go.
This leads me to a decision on a speaker type...conventional box, line array, MEH, open-baffle, etc.
Whatever design is chosen, I have to decide how many driver sections will be used to achieve the SPL goal and bass extension, and how their radiation patterns will merge smoothly.
Then sims help produce plans for a proto-type. (But sims help less than paying basic attention to driver c2c distances, baffle/horn widths, and displacement capabilities of driver sections at their lowest frequency range of operation.)
And then build the proto-type, which is always work, but mainly labor, and not so much as mental work. So not hard.
Through completion of the proto, I haven't given crossovers 30 seconds of thought, other than knowing the approximate xover points based on the acoustic design.
So what's been hardest till now? 🙂
Well. assuming I don't have a measurement rig /spinorama in my back pocket, I'll first need to build one. For what size speaker? so that it carry the weight, and how big can the turntable be before it becomes a diffraction problem on its own?
Then, once the spinorama is built, which driver section should be the key section, having its point-of-rotation directly over the center of the turntable?
How do I handle the inevitable timing differences between the key section and other driver sections as I rotate and take measurements?
Even bigger question...at what distance do I measure? What is the minimum measurement distance for onset of the acoustic far-field?
Can I do that distance indoors, in a reflection free enough environment, or do I need to move outdoors. If indoors, can I live with the kludge of gating to reduce reflections' influence, or the inaccuracy of nearfield acoustic meging techniques? If I go outdoors, can I live with the noise and wind in my measurements? Should I do ground plane measurements, or flown as high off the ground as possible?
Hey, why don't my measurements look as pretty as the sims I ran ??????? Which is the final reality?
What will equate to what I hear?
We can read about sound forever, we can sim it forever .......
I've never yet heard a sim make music, until I build a speaker trying to match the sim.
And then once built, how do I fix what I don't like or question? Other than take measurements, that I believe I can act on.
So in the end it's back to measurements.
But hey, I really think there's been a swerve into sims vs measurements that wasn't intended...I certainly didn't mean to go there...
Topic in my mind is more about the comparative difficulty of crossovers vs measurements.
Frankly, and I know this will draw criticism....the reason crossovers are seen as difficult is because of one word...PASSIVE.
What an unnecessary complicated path imho, and worse leaves attainable SQ on the table, for all but the simplest of speaker designs.
(Flame suit on. 🙂
Active with the right software tools can generate near perfect complementary crossovers in about 5 minutes or less, and can separate fixed time alignment (using precise digital delays) from mag and phase alignment. Either through IIR or FIR.
All based on measurements..so that tuning work is final, and only dependent of the quality of the measurements.
Just my proven method...how can I not see measurements as more difficult work than crossovers??
Yikes...what a great paper to show what a pain passive crossovers are !
Passives do not leave unattainable sound quality on the proverbial table. Horse biscuits! If that were the case, I would be using active filters all of the time. Guess what- I'm not.
Just because active xovers are easier to dial in and tweak for the ones with experience with active and inexperienced with passive does not mean active xovers automatically sound better.
I have never walked in to hear an active system and been elated, sworn off passives, and thought they were the best I'd ever heard. I have thought that of passive systems.
I have heard transfer function matched active and passive systems, and preferred the passive one.
Just because active xovers are easier to dial in and tweak for the ones with experience with active and inexperienced with passive does not mean active xovers automatically sound better.
I have never walked in to hear an active system and been elated, sworn off passives, and thought they were the best I'd ever heard. I have thought that of passive systems.
I have heard transfer function matched active and passive systems, and preferred the passive one.
Thanks everyone, so much food for thought!
So would we say the categories of crossover are::
Analogue passive
Analogue active
Digital (DSP) active
Has anyone here done a blind test between all the crossover types on identical speakers, all setup with the same XO frequencies and slopes, and been able to point out which is which?
So would we say the categories of crossover are::
Analogue passive
Analogue active
Digital (DSP) active
Has anyone here done a blind test between all the crossover types on identical speakers, all setup with the same XO frequencies and slopes, and been able to point out which is which?
In addition to IIR digital crossovers equivalent to analog crossovers, DSP FIR crossovers are also another category.
FIR crossovers can flatten phase response, something not easily achievable in IIR crossovers, and may be audible.
Passive crossovers placed between the amplifier and speaker are designed around each driver's specific impedance profiles.
At high power levels, voice coils heat up and their impedance rises, which also changes the crossover/driver frequency response. Different program material requires different power demands for each driver, which can result in dynamic program dependent frequency response changes.
The response of crossovers placed ahead of amplification does not change with voice coil temperature, so at high power levels there is considerably less change in the speaker's response.
As a result, it can be fairly easy to hear the difference between active crossovers and passive at high levels, but little to no difference at low volumes, assuming the frequency and phase response is matched.
Art
Doesn't matter, it is what the builder prefers and outcome goal is the same.
You will get a wide range of opinions and most of them will be biased towards
each individuals preferences. Many times based on assumptions and wild guessing from beginners.
And real world facts from experienced. Keeping in mind experienced people still imagine
and make up audio theories and spend a lot of time convincing others they work.
As suggested why ask, discover for yourself.
Software sim has a learning curve and better to start sooner than never.
It wont lie , make up things and have bias like others opinions
You will see how sound actually behaves, and learn from technical and real world
visual representation. Once you learn to accurately model a driver in Full space.
To understand the Half space response of the baffle.
Summing the drivers with active or passive filters is the same process.
You wont be destroying sound. Each driver has a ideal bandwidth.
Depending on size, eventually off axis response falls when wavelengths are smaller than the driver.
You simply filter out what it cant do. And move along to the next smallest driver to reproduce
smaller wavelengths. AKA the higher the frequency the smaller the wavelength.
Last edited: