I'm messing around with DSP, and at the end of it I'll build some nice active three way speaker - digital from source to individual power amps.
Except - I'm wondering, with modern drive units if there is actually any benefit in going three way? Especially since DSP means you can shape the speaker response curve.
Part of the logic of going DSP is that you've lost the dodgy crossover stuff and hence you you can keep adding units at no quality cost.
But there is still the audio cost of the signal chopping from one speaker to another.
I've decided on SB Acoustics units. Their six inch bass mid - SB17NBAC35-8 - seems to be a better performer than their slightly smaller units. It's also cheap. I'm wondering if adding their 9.5in bass actually achieves anything useful.
Options are tweeter, 6in mid, 9.5 bass (the default choice).
Options are tweeter, 6in mid, 9.5 bass, single sub.
Tweeter, 2x 6in mid, single sub.
Tweeter, 6in mid, single sub.
Tastes are wide, but I listen to a lot of loud rock in a very big room.
Currently using EPOS m22s, and very happy with them. This is really about messing with DSP.
What's the current thinking on numbers of units when cost isn't a major concern?
Except - I'm wondering, with modern drive units if there is actually any benefit in going three way? Especially since DSP means you can shape the speaker response curve.
Part of the logic of going DSP is that you've lost the dodgy crossover stuff and hence you you can keep adding units at no quality cost.
But there is still the audio cost of the signal chopping from one speaker to another.
I've decided on SB Acoustics units. Their six inch bass mid - SB17NBAC35-8 - seems to be a better performer than their slightly smaller units. It's also cheap. I'm wondering if adding their 9.5in bass actually achieves anything useful.
Options are tweeter, 6in mid, 9.5 bass (the default choice).
Options are tweeter, 6in mid, 9.5 bass, single sub.
Tweeter, 2x 6in mid, single sub.
Tweeter, 6in mid, single sub.
Tastes are wide, but I listen to a lot of loud rock in a very big room.
Currently using EPOS m22s, and very happy with them. This is really about messing with DSP.
What's the current thinking on numbers of units when cost isn't a major concern?
Of course. Only three way. Let midrange do its job and deliver clear and quick sound, not blended with bass duties. 2 way is always a big compromise in terms of sound.
Hi,
three way system absolutely. You can do two way speakers if you add sub woofers making a three way system.
There are multiple reasons to do three way and the most fundamental is simply very different wavelengths for bass and treble. At very lows the wavelengths are size of the room, at very treble the wavelengths get smaller than the transducer. Then there is the mid range that falls in between that is kind of transition between the two extremes.
There are problems that just go away or are straight forward to deal with on a three way speaker like standing waves inside possible enclosure, noise from possible reflex port, panel resonances and electro mechanical properties of the transducers to name few.
As you divide the frequency band for different transducers you also divide the problems of each band into separate concerns which allows taking compromises that don't affect the particular band performance. For example when the bass and mid are handled by different transducers one is able to utilize two different size enclosures with possibly different alignments, bracing/damping, baffle shape and size. Both can have different set of compromises that benefit the sound of that particular band while the trade-offs can be pushed out of band, out of existence. On a two way speaker one cannot make a small box for the woofer because the bass suffers. If you make it big for the bass the mids suffer.
You can think it this way as well: Never compromise the top octave, there is nothing to fix it besides the transducer you plan to use for the treble (due to short wavelengths). If your hearing is good you need to use 1" or smaller transducer for optimal sound. This small transducers can't play lower than about 1kHz so at least two ways are needed. 1kHz wavelength is 10x shorter than 100Hz, which is about average livingroom dimensions, so maybe one transducer for the room sized waves and one for the mids between room and the tweeter band. Low bass can be compromised on the main speakers since one can always add more sub woofers (due to long wavelengths), hence two way + sub(s) is a practical solution. Anyway, a three way it is, minimum. You might find reasons to do even more ways.
Very simple stuff, the physics 😀 Good to figure out before even thinking about which brand transducers or crossover topology to use. I think only problem with three way is a bit more cost and complexity but the benefit is for audio quality. Have fun with your project!🙂
three way system absolutely. You can do two way speakers if you add sub woofers making a three way system.
There are multiple reasons to do three way and the most fundamental is simply very different wavelengths for bass and treble. At very lows the wavelengths are size of the room, at very treble the wavelengths get smaller than the transducer. Then there is the mid range that falls in between that is kind of transition between the two extremes.
There are problems that just go away or are straight forward to deal with on a three way speaker like standing waves inside possible enclosure, noise from possible reflex port, panel resonances and electro mechanical properties of the transducers to name few.
As you divide the frequency band for different transducers you also divide the problems of each band into separate concerns which allows taking compromises that don't affect the particular band performance. For example when the bass and mid are handled by different transducers one is able to utilize two different size enclosures with possibly different alignments, bracing/damping, baffle shape and size. Both can have different set of compromises that benefit the sound of that particular band while the trade-offs can be pushed out of band, out of existence. On a two way speaker one cannot make a small box for the woofer because the bass suffers. If you make it big for the bass the mids suffer.
You can think it this way as well: Never compromise the top octave, there is nothing to fix it besides the transducer you plan to use for the treble (due to short wavelengths). If your hearing is good you need to use 1" or smaller transducer for optimal sound. This small transducers can't play lower than about 1kHz so at least two ways are needed. 1kHz wavelength is 10x shorter than 100Hz, which is about average livingroom dimensions, so maybe one transducer for the room sized waves and one for the mids between room and the tweeter band. Low bass can be compromised on the main speakers since one can always add more sub woofers (due to long wavelengths), hence two way + sub(s) is a practical solution. Anyway, a three way it is, minimum. You might find reasons to do even more ways.
Very simple stuff, the physics 😀 Good to figure out before even thinking about which brand transducers or crossover topology to use. I think only problem with three way is a bit more cost and complexity but the benefit is for audio quality. Have fun with your project!🙂
Last edited:
Isn't the box size about controlling the speaker's response around the driver's resonance?
Yeah but if the bandwidth of that speaker(driver) in the speaker system extends to frequencies that have shorter wavelengths than the enclosure (much shorter than the driver resonance) you start to see wiggle in the impedance measurement, standing waves inside the enclosure affecting the performance. Yeah you can minimize these. Often it is a ported box, then the mids leak through the port. The port has its own resonance etc. Alright these can be dealt with as well. How about enclosure panels? Make them stiff with bracing and they still resonate at the pass band and affect the sound and you need damping to deal with it. How about excursion of the driver? raise the volume for nice SPL and the mid suffers the more the woofer excursion. Excursion of a transducer quadruples when you halve the frequency (simplified).
Alright, nothing too severe but these few examples alone are 5 aspects to improve on. You can skip most of it by not having an enclosure! But then there is no bass.. Deal with them all and sound should be better and there is more compromises to take. You can do it with two way but many "problems" don't exist or are lot easier to dealt with on a three way system. All due to wavelengths.
All loudspeakers are a bunch of compromises taken for various reasons. Audio quality is just one measure and then there is aesthetics, size constrains, the spouse and costs like time and money. Many things to compromise between and prioritize on each project. A 3 way system allows to push more of the trade-offs outside the audio quality, to cost, to complexity, out of band, and leave better audio quality behind. A two way compromises either lowest bass, the mids in between or the highest treble. If the bandwidth and/or SPL of the system can be relaxed a two way might be the right recipe, it is simpler and cheaper. Better yet a fullrange driver might be nice enough, no need for xo even! just hook the fullranger to an amp without enclosure and call it fine. What ever rocks the boat 😉
Alright, nothing too severe but these few examples alone are 5 aspects to improve on. You can skip most of it by not having an enclosure! But then there is no bass.. Deal with them all and sound should be better and there is more compromises to take. You can do it with two way but many "problems" don't exist or are lot easier to dealt with on a three way system. All due to wavelengths.
All loudspeakers are a bunch of compromises taken for various reasons. Audio quality is just one measure and then there is aesthetics, size constrains, the spouse and costs like time and money. Many things to compromise between and prioritize on each project. A 3 way system allows to push more of the trade-offs outside the audio quality, to cost, to complexity, out of band, and leave better audio quality behind. A two way compromises either lowest bass, the mids in between or the highest treble. If the bandwidth and/or SPL of the system can be relaxed a two way might be the right recipe, it is simpler and cheaper. Better yet a fullrange driver might be nice enough, no need for xo even! just hook the fullranger to an amp without enclosure and call it fine. What ever rocks the boat 😉
Last edited:
Depends how loud you need to go. Modern drivers can be very good.
I am not a fan of greater than quarter-wave driver spacing so as to preserve coherency and seemlessness, a 3-way would have XOs at 40-60 Hz and 250-450 Hz. A 2-way would lose the subwoofers.
Keep in mind XOs are “evil” and the lower you can push them the more they get out of the way.
dave
I am not a fan of greater than quarter-wave driver spacing so as to preserve coherency and seemlessness, a 3-way would have XOs at 40-60 Hz and 250-450 Hz. A 2-way would lose the subwoofers.
Keep in mind XOs are “evil” and the lower you can push them the more they get out of the way.
dave
The interesting thing about speakers is there are so many compromises.
All of these compromises have consequences.
The smaller the driver, the higher frequency it will play without beaming.
The smaller the driver, the lower max SPL it will play for a given frequency.
If you use a larger driver for high frequencies, it has to be designed in such a way that it has what amounts to a mechanical crossover.
You cannot fix Acoustic problems with DSP. Any correction you make fixes one spot but makes the other spots worse, as well as potentially the reverberant sound as well.
No perfect way to do this. Several ways that work.
HTH
Doug
All of these compromises have consequences.
The smaller the driver, the higher frequency it will play without beaming.
The smaller the driver, the lower max SPL it will play for a given frequency.
If you use a larger driver for high frequencies, it has to be designed in such a way that it has what amounts to a mechanical crossover.
You cannot fix Acoustic problems with DSP. Any correction you make fixes one spot but makes the other spots worse, as well as potentially the reverberant sound as well.
No perfect way to do this. Several ways that work.
HTH
Doug
acampm, you underestimate those that have come before. Old school crossovers were never a limitation.
This is a question that requires you to go back to basics. Forget the electrical filters and see what's happening before them.
This is a question that requires you to go back to basics. Forget the electrical filters and see what's happening before them.
Allen,
I must disagree. FIR processing is a big leap forwards, and allows things that simply cannot be done with passive filters.
My solution has been to use a passive filter to integrate the drivers, and correct for the filter's phase shift (and a few other things) using FIR. Keeps the setup simple (2-ch amp required), and sounds good.
In terms of 2-way vs 3-way, it's simply a matter of SPL and directivity.
All other things being equal, larger drivers are capable of more low-frequency SPL than smaller drivers. They will also show narrowing directivity at a lower frequency than a smaller driver.
A 6"/1" box is a pretty poor compromise (IMO) - the directivity plot is a mess, and a single 6" driver rarely produces adequate bass.
Moving to a 10"/4"/1" 3-way means a smoother directivity plot (the steps between the diaphragm sizes are smaller) and increased LF output.
I went in a different direction: pick a large enough bass driver to do the job, and then cross to a compression driver mounted to a constant-directivity horn.
It's still a 2-way design, but has the same benefits as the 3-way: plenty of bass, and a good directivity plot.
There's one problem, though, which I'm working to rectify: the upper bass response isn't great. They dig very deep into the bass, but the upper bass impact is missing. After some experimentation, I have realised the problem is floor (and/or ceiling) bounce, which results in a null where the "punch" should be.
My chosen solution is to add another pair of bass drivers, mounted close to the floor. They'll be run as a 2.5-way design.
A 3-way design with the LF unit mounted close to the floor would also mitigate this issue.
Chris
I must disagree. FIR processing is a big leap forwards, and allows things that simply cannot be done with passive filters.
My solution has been to use a passive filter to integrate the drivers, and correct for the filter's phase shift (and a few other things) using FIR. Keeps the setup simple (2-ch amp required), and sounds good.
In terms of 2-way vs 3-way, it's simply a matter of SPL and directivity.
All other things being equal, larger drivers are capable of more low-frequency SPL than smaller drivers. They will also show narrowing directivity at a lower frequency than a smaller driver.
A 6"/1" box is a pretty poor compromise (IMO) - the directivity plot is a mess, and a single 6" driver rarely produces adequate bass.
Moving to a 10"/4"/1" 3-way means a smoother directivity plot (the steps between the diaphragm sizes are smaller) and increased LF output.
I went in a different direction: pick a large enough bass driver to do the job, and then cross to a compression driver mounted to a constant-directivity horn.
It's still a 2-way design, but has the same benefits as the 3-way: plenty of bass, and a good directivity plot.
There's one problem, though, which I'm working to rectify: the upper bass response isn't great. They dig very deep into the bass, but the upper bass impact is missing. After some experimentation, I have realised the problem is floor (and/or ceiling) bounce, which results in a null where the "punch" should be.
My chosen solution is to add another pair of bass drivers, mounted close to the floor. They'll be run as a 2.5-way design.
A 3-way design with the LF unit mounted close to the floor would also mitigate this issue.
Chris
Look at a few distortion plots. Most drivers have a sharp increase in distortion below ~200Hz. You don't want that causing intermodulation in a mid driver. In a 3 way most drivers aren't being run as close to their limits as in a 2 way. So for me, 3 way it is.
As for driver compliment, my choice would be tweeter, 5" mid, 2x 9.5" woofers..... for wider dispersion and solid bass
As for driver compliment, my choice would be tweeter, 5" mid, 2x 9.5" woofers..... for wider dispersion and solid bass
We can do delay passively (not that all speakers need it), change phase alone, fine EQ and precise filtering. What else?
acampm wrote:
Allen wrote:
I agree with Allen, there is a lot of factors that need to be taken into account when designing a speaker, and a lot of cases where good judgment is required when the inevitable compromises must be made. Applying active DSP-based crossover filters can make things seem easier, but all the same issues are still there. In the last year I have designed both passive and active DSP systems.
read through this thread from a few months ago... it covers many of the same questions you are working through
Plans for an active 3 way speaker
Allen wrote:
I agree with Allen, there is a lot of factors that need to be taken into account when designing a speaker, and a lot of cases where good judgment is required when the inevitable compromises must be made. Applying active DSP-based crossover filters can make things seem easier, but all the same issues are still there. In the last year I have designed both passive and active DSP systems.
read through this thread from a few months ago... it covers many of the same questions you are working through
Plans for an active 3 way speaker
Last edited:
Yeah, with music's ~10 octaves [ignoring 16 Hz], then equal octave spreads is near enough 20-200 Hz, 200-2 kHz, 2-20 kHz, making driver selection easy and recommend an 8" 200-2 kHz for a sufficiently 'full' mid-range same as the pioneers concluded for vocals.
In my 3 way build hifijim quoted I went with a 1-5-10 TMW build with the SB ceramic drivers. Being able to have a dedicated midbass greatly eliminated the distortion below 200hz in the 5 inch mid. Once the veneers on I may post some pictures lol
Also LOVE those drivers. Extremely pleased overall. So thanks!
Also LOVE those drivers. Extremely pleased overall. So thanks!
I'm finding myself arguing against my own speaker but:
Crossovers are indeed evil things, most of them producing unwelcome phase changes, and passive ones are very hard to tweak. Designing speakers with a three way passive crossover is a nightmare, and you end up spending a fortune on changing bits. This doesn't apply in this case however because DSP puts me in charge of phase and changing the crossover behaviour is simply a matter of plugging in the USB cable.
No two units have quite the same sound and as you move unit to unit the effect is often quite.
The more units the more phasing issues.
Most crossovers produce a dip at the crossover point.
Big bass drivers mean wide baffles, which means poor imaging. it doesn't apply in this case though because each unit has a separate enclosure.
A three UNIT speaker can manage a higher SPL. But nobody said those units have to be different. You can use two mid/woofers.
Once upon a time the standard 3 way speaker used the B139, the 110 and the T27. You needed all three of those units to cover the range. But a modern bass/mid unit can cover pretty much the whole range of the 139 and 110, and a modern tweeter too can go significantly lower than a t27.
If I were building a modern passive single enclosure ported speaker I would undauntedly use one tweeter and two mids with a second order crossover.
But I'm not. I'm building an active DSP controlled speaker with separate sealed enclosures. Unless anyone cares to convince me otherwise I'll make it three way. I'm still not convinced three way is a good thing in general though, and I swore I'd never make another one...
Crossovers are indeed evil things, most of them producing unwelcome phase changes, and passive ones are very hard to tweak. Designing speakers with a three way passive crossover is a nightmare, and you end up spending a fortune on changing bits. This doesn't apply in this case however because DSP puts me in charge of phase and changing the crossover behaviour is simply a matter of plugging in the USB cable.
No two units have quite the same sound and as you move unit to unit the effect is often quite.
The more units the more phasing issues.
Most crossovers produce a dip at the crossover point.
Big bass drivers mean wide baffles, which means poor imaging. it doesn't apply in this case though because each unit has a separate enclosure.
A three UNIT speaker can manage a higher SPL. But nobody said those units have to be different. You can use two mid/woofers.
Once upon a time the standard 3 way speaker used the B139, the 110 and the T27. You needed all three of those units to cover the range. But a modern bass/mid unit can cover pretty much the whole range of the 139 and 110, and a modern tweeter too can go significantly lower than a t27.
If I were building a modern passive single enclosure ported speaker I would undauntedly use one tweeter and two mids with a second order crossover.
But I'm not. I'm building an active DSP controlled speaker with separate sealed enclosures. Unless anyone cares to convince me otherwise I'll make it three way. I'm still not convinced three way is a good thing in general though, and I swore I'd never make another one...
Last edited:
Unfortunately the unit I want to use is 6". I think it can cope.
This is, of course, the sensible reason for a 3 way system. But there are prices to be paid 🙂
Yes, interesting. I see that people have quite strong views on crossovers.
I've actually got two crossovers in my drawer, one active and one DSP. The plan is the system moves from DSP to Active when the turntable is in use, although that may change, I may find the DSP is so much better it's worth having an ADC in the signal path.