I'm in the process of considering the design of a new 3 way active system with subwoofer. I want to use 24db/oct filters with major driver resonances catered for. Linkwitz has good info on this for one. Usually the cabinet for each driver, (leaving out the tweeter of course), is made larger than required to push the main resonance as low as possible and this affects the response curve below crossover at that point. He has filter designs which complement the resonance of driver/cabinet to give good controlled attenuation below crossover.
With the existing info and equipment for measuring driver/cabinet parameters nowadays I am curious as to the feasibility of designing the cabinet as a part of the crossover. With a sealed cabinet of accurate initial volume you can measure the driver resonant frequency and Q when mounted and use this as a 12db/oct crossover stage. You could then tailor a second 12db/oct electronic stage to match the cabinet in terms of resonant frequency and complement its Q to get an overall setup of your choice. Mid range and bass cabinets then become smaller as they are not trying to drive the resonant frequency down as low as possible and bottom end response becomes more linear and matched.
Like all "good ideas", it may be a complete non-starter for any number of obvious reasons of course. Has anyone tried this approach with any degree of success, or can they comment on the idea?
With the existing info and equipment for measuring driver/cabinet parameters nowadays I am curious as to the feasibility of designing the cabinet as a part of the crossover. With a sealed cabinet of accurate initial volume you can measure the driver resonant frequency and Q when mounted and use this as a 12db/oct crossover stage. You could then tailor a second 12db/oct electronic stage to match the cabinet in terms of resonant frequency and complement its Q to get an overall setup of your choice. Mid range and bass cabinets then become smaller as they are not trying to drive the resonant frequency down as low as possible and bottom end response becomes more linear and matched.
Like all "good ideas", it may be a complete non-starter for any number of obvious reasons of course. Has anyone tried this approach with any degree of success, or can they comment on the idea?
You're nuts. I say that with the best of intentions since mostly people with very imaginative imaginations ever get called nuts.... so it is meant as a kind of compliment.
Joking aside, it is unstable and unsatisfying to traffic in resonances which smoosh around and behave badly. Also, you are talking as if you could make a definitive design incorporating the raw, bleeding edges of speaker performance.
Sadly, all our cone subs are primitive in the sense that we obliged them to produce sound outside their principal design range (at least at the south end of their range)... through various tricks of resonance, rear wave boostering, and so on or, in the case of sealed boxes, planning to have the system resonance at a beneficial place and not too boomy.
Even though we are forced to live with primitive subs, it seems unsound to try to use these out-of-bounds ranges in any positive way, as you imaginatively suggest.
Technical footnote: the driver/mechanical factors that would broaden your Q enough to make the pass-band useful are factors that harm transient response.
Ben
Joking aside, it is unstable and unsatisfying to traffic in resonances which smoosh around and behave badly. Also, you are talking as if you could make a definitive design incorporating the raw, bleeding edges of speaker performance.
Sadly, all our cone subs are primitive in the sense that we obliged them to produce sound outside their principal design range (at least at the south end of their range)... through various tricks of resonance, rear wave boostering, and so on or, in the case of sealed boxes, planning to have the system resonance at a beneficial place and not too boomy.
Even though we are forced to live with primitive subs, it seems unsound to try to use these out-of-bounds ranges in any positive way, as you imaginatively suggest.
Technical footnote: the driver/mechanical factors that would broaden your Q enough to make the pass-band useful are factors that harm transient response.
Ben
Last edited:
Hi,
Its done all the time in some aspects of speaker design.
Two typical cases :
1) Incorporating driver roll offs into the required acoustic function.
See here Zaph|Audio - ZMV5 - MCM / Vifa 5" System for 4th
order acoustic using 2nd and 3rd order electrical functions.
2)Sealed satellites with (a) sub(s). Sat is tuned to Q=0.71
Butterworth at Fb. Active sub is rolled off 4th order L/R at
Fb and sats 2nd order BW at Fb = symmetrical 4th order L/R.
However a 3 way is not usually a sub/sat system and
for various reasons Fb of the midrange is often below
the x/o point. Often you can get away with a fairly high
Q at Fb for the mid in a small enclusure with the x/o.
I can't see how the idea makes bass cabinets any
smaller as they are not oversized in the first place.
rgds, sreten.
Its done all the time in some aspects of speaker design.
Two typical cases :
1) Incorporating driver roll offs into the required acoustic function.
See here Zaph|Audio - ZMV5 - MCM / Vifa 5" System for 4th
order acoustic using 2nd and 3rd order electrical functions.
2)Sealed satellites with (a) sub(s). Sat is tuned to Q=0.71
Butterworth at Fb. Active sub is rolled off 4th order L/R at
Fb and sats 2nd order BW at Fb = symmetrical 4th order L/R.
However a 3 way is not usually a sub/sat system and
for various reasons Fb of the midrange is often below
the x/o point. Often you can get away with a fairly high
Q at Fb for the mid in a small enclusure with the x/o.
I can't see how the idea makes bass cabinets any
smaller as they are not oversized in the first place.
rgds, sreten.
Last edited:
Yes, that is exactly what Mr. Linkwitz did in the first iteration of this design.
I've tried the same thing at home with my DIY loudspeakers and it works very well.
However, the only way to reduce the box volume is if you get a heavyly overdamped driver, so you can make a smaller box in order to get a higher Q... somewhere near 0.71
Last edited:
No "overdamped driver" for me because of the trade-off mentioned earlier.
Lots of things are done for legitimate reasons like saving money, parts count, partisan design politics, etc that are not inherently to improve sound.
If you are working at the ragged (and unstable) edge of your speaker's passband, you have to take shortcomings of your speaker into consideration in configuring the crossover, of course. But grudgingly. As much as I like OB dipole sound, I think that is a fair description of the way such systems are "holding on by their fingernails" to produce bass.
I don't think anybody has so far posted erroneous views. Just a matter of applying good judgment in developing the OP's concept and not unduly focusing on niche applications, however illuminating as colour-commentary these may be.
Ben
Hi,
TBH I have no idea what Ben is talking about and its applicability to general 3 way active design.
rgds, sreten.
TBH I have no idea what Ben is talking about and its applicability to general 3 way active design.
rgds, sreten.
Last edited:
To get back to the original question, I have had very good luck doing exactly what the OP describes. I use Kef 101 with a 12db/oct roll-off starting at 100 Hz, add another 12 db/oct at that point, and add subwoofers at 24 db/oct at 100 Hz. Done active in MPD. Very happy with results.
Not sure I have the analogy worked out to be clear enough for all, but consider the internal combustion engine. It has a "passband" limited by lugging at the south end and valve float at the north end. Sure, with 95 HP in my 500 lb bike, I can tootle around town all day at lugging rpm. But that isn't proud riding.
Just because on freq traces, the KEF 101 kind of poops out below 100 Hz, kind of like resembling 12 dB/8ave, doesn't mean it is behaving right anywhere in the vicinity of 100 Hz... even if measured kind of flat.
You can't just take complex systems and reduce them to freq curves* and then do some kind of magic algebra adding and subtracting dBs and think you've made great sound because your theoretical curves sort of add up to flat.
Ben
*I'm all for curves, I'm just saying that no single measure tells much of the story. You need distortion, dispersion, ringing.....
Just because on freq traces, the KEF 101 kind of poops out below 100 Hz, kind of like resembling 12 dB/8ave, doesn't mean it is behaving right anywhere in the vicinity of 100 Hz... even if measured kind of flat.
You can't just take complex systems and reduce them to freq curves* and then do some kind of magic algebra adding and subtracting dBs and think you've made great sound because your theoretical curves sort of add up to flat.
Ben
*I'm all for curves, I'm just saying that no single measure tells much of the story. You need distortion, dispersion, ringing.....
Last edited:
Ben, I don't really want to get into an arguement, but . . .
I have yet to hear a better combination in my living room than I have now,
The behavior of a theoretical piston, critically damped in cabinet does in fact match quite well with 12 db/octave butterworth slope and delay function, per Mr. Linkwitz various articles.
Please don't simply denigrate ideas and practices without testing and researching.
Yes, I could have better equipment. Most of us could. Life is a compromise. What I got was better than before, and it didn't cost a lot. Made me happy.
I have yet to hear a better combination in my living room than I have now,
The behavior of a theoretical piston, critically damped in cabinet does in fact match quite well with 12 db/octave butterworth slope and delay function, per Mr. Linkwitz various articles.
Please don't simply denigrate ideas and practices without testing and researching.
Yes, I could have better equipment. Most of us could. Life is a compromise. What I got was better than before, and it didn't cost a lot. Made me happy.
I guess you need a bit more understanding about "complex systems" as you call them. Yes, YOU CAN REDUCE them to frequency curves and get all the info you need from them. That's the way control systems works, that's the way electronic systems works, and of course.. that's the way loudspeaker system works.
Sure you can talk about love, emotion and that kind of messy things, but none of them defines the real behaviour of a loudspeaker 😉
I am delighted you conflate my recommendation about assessing distortion, dispersion, and transient response with love, emotion, and messy things. (My wife berates me for doing the exact same thing when I am deep into working with the HiFi.)
B.
Last edited:
Where can I buy one of those theoretical pistons?
Kidding aside, didn't Olson's textbook mention it.... at least 70 yrs ago?
But pointed remarks aside, it takes more than freq-curve algebra to make great sound.
If there is any merit to working/engaging with the untweaked behaviour of speakers, it is from avoiding upstream electronic clutter and cost. Endless examples of doing so with one common instance you find is using just a capacitor as a tweeter cross-over. But maybe dislike of electronic clutter is not a lot different than arguing about vinyl-vs-CD. Given the goodness of electronics and the crudeness of cone drivers, I'd opt for the clutter.
You feel my past 57 yrs of experimenting with speakers (starting with a Karlson-15 and including decades with motional feedback) just isn't enough cred to allow me to express an opinion here? But jplesset, I'm not sure I can make it to 58!
B.
Last edited:
Hi,
Try making great sound without using basic freq-curve principles.
You won't get very far in your endeavour without resorting to them.
FWIW regarding the B110 and a 100Hz x/o point active (+ box rolloff):
You can do it as outlined and it will work very well. The acoustic
result will be near symmetrical 4th order L/R x/o at 100Hz.
Is it the best option ? Probably not regarding maximum SPL.
When does it make the most sense ? When adding a small
subwoofer that goes deep but has limited output capability
and as such that x/o point is a good match between the
speakers capabilities and the subwoofers capabilities.
With bigger subs capable of more SPL you'd prefer two
x/o higher to the B110, so the B110 can do more SPL.
rgds, sreten.
I think some may be misunderstanding the details of what I was proposing here. (Figures quoted below are not intended to be real, just to give a feel for the idea).
Assume an active system with a mid range driver with a fundamental resonance at 50Hz and we are using 2x12db/oct Xover filters at 200Hz and Q of 0.71. Usually the mid cabinet would be designed to keep the combined cabinet/driver resonance as low as possible, all other factors being acceptable. Then the crossover is designed to give us that Q of 0.71 at 200Hz break point. This leads to an undefined unhandled physical resonance somewhere above 50Hz. That is less than 2 octaves away and will lead to response anomalies from just above that point and on down.
With Linkwitz (and other) type filters you can tune out this resonance but surely it would be nice to bring it into our design and use it if possible? And using it to remove the need for a second electronic stage is a plus too.
I was suggesting raising the cabinet/driver resonant point to the crossover frequency by matching the cabinet volume. We then have a reasonably well defined acoustic 12db/oct filter to start with. (At least I believed that this was a well defined 12db/oct until I posted this thread). Surely moving up in frequency like this is working further within the drivers capabilities? We then match the acoustic Q with a single complementary Q 12db/oct electronic filter at the same frequency to give an overall Q of our desired 0.71 . (Incidentally I have seen this method of cascading differing Qs to achieve the overall result proposed as being superior to equal Q cascading).
We can measure the resonant point and its Q quite easily nowadays, isn't this done all the time to fine tune and optimise our ported designs? (Well I do it with highly improved results both on scope screen and in listening so I know the method works!)
You are right, the process may not be accurate to precision levels but there is a wide production tolerance of parameters amongst drivers of the same model for even a static design, and we should be taking this out of the equation by measuring for our specific drivers to hand.
Strangely, someone mentioned doing this with KEF101s and the units I am proposing using are KEF. I have a number of pairs of B139s, both SP6171 and SP1044, which I wanted to put into large independent sealed cabinets for lower bass, (I deliberately didn't say sub-woofers as I live with restrictions imposed by size of room and living alongside others). These can be extended actively slightly at the bottom end to give really pleasing deep bass without being too demanding and transient response should be better than my current large (LARGE! and ugly) cylindrical ported units. I have a pair of B169 units of Fs 50Hz (brand new unused!) to put into a 3 way satellite unit for bass, crossing over to the B139s at around 75Hz. The same thing could be done with B110s, (or a modern replacement as I think these are a bit long in the tooth now), for mid above the B169s.
This would give a neat compact 3 way satellite unit as enclosures would be smaller than normal, with plenty of space within the stands for the electronics. Simulations show sensible starting points of 32litres for the B169 to give Fb=74.75Hz and F3=61.54Hz. I know in the real world this will not be exact of course, but it suggested a system worth trying.
Assume an active system with a mid range driver with a fundamental resonance at 50Hz and we are using 2x12db/oct Xover filters at 200Hz and Q of 0.71. Usually the mid cabinet would be designed to keep the combined cabinet/driver resonance as low as possible, all other factors being acceptable. Then the crossover is designed to give us that Q of 0.71 at 200Hz break point. This leads to an undefined unhandled physical resonance somewhere above 50Hz. That is less than 2 octaves away and will lead to response anomalies from just above that point and on down.
With Linkwitz (and other) type filters you can tune out this resonance but surely it would be nice to bring it into our design and use it if possible? And using it to remove the need for a second electronic stage is a plus too.
I was suggesting raising the cabinet/driver resonant point to the crossover frequency by matching the cabinet volume. We then have a reasonably well defined acoustic 12db/oct filter to start with. (At least I believed that this was a well defined 12db/oct until I posted this thread). Surely moving up in frequency like this is working further within the drivers capabilities? We then match the acoustic Q with a single complementary Q 12db/oct electronic filter at the same frequency to give an overall Q of our desired 0.71 . (Incidentally I have seen this method of cascading differing Qs to achieve the overall result proposed as being superior to equal Q cascading).
We can measure the resonant point and its Q quite easily nowadays, isn't this done all the time to fine tune and optimise our ported designs? (Well I do it with highly improved results both on scope screen and in listening so I know the method works!)
You are right, the process may not be accurate to precision levels but there is a wide production tolerance of parameters amongst drivers of the same model for even a static design, and we should be taking this out of the equation by measuring for our specific drivers to hand.
Strangely, someone mentioned doing this with KEF101s and the units I am proposing using are KEF. I have a number of pairs of B139s, both SP6171 and SP1044, which I wanted to put into large independent sealed cabinets for lower bass, (I deliberately didn't say sub-woofers as I live with restrictions imposed by size of room and living alongside others). These can be extended actively slightly at the bottom end to give really pleasing deep bass without being too demanding and transient response should be better than my current large (LARGE! and ugly) cylindrical ported units. I have a pair of B169 units of Fs 50Hz (brand new unused!) to put into a 3 way satellite unit for bass, crossing over to the B139s at around 75Hz. The same thing could be done with B110s, (or a modern replacement as I think these are a bit long in the tooth now), for mid above the B169s.
This would give a neat compact 3 way satellite unit as enclosures would be smaller than normal, with plenty of space within the stands for the electronics. Simulations show sensible starting points of 32litres for the B169 to give Fb=74.75Hz and F3=61.54Hz. I know in the real world this will not be exact of course, but it suggested a system worth trying.
Haha! Thanks Ezavalla, the Linkwitz article you posted at the start of this thread has been with me for many many years. I actually have the original Wireless World edition with it in still on my shelves, along with many other interesting ones over the years. That article started me thinking down this path so long ago in my mid-20s and has nagged me ever since. Now I have the time I can be determined to use some of what Linkwitz proposes.
AlbertB - that's an excellent explanation and my compliments again. It deserves worthy replies. And I'm certain you have the mature "grown-ups" judgment to make sense of the helpful, unhelpful, just-to-score-points, and plain un-nice posts in your thread.
Your post lays out the issues very nicely as far as it goes. But here are further issues. At the largest perspective, is the design procedure you are thinking about sensible.... disirregardless of the frequency curve magic (which seems to be the fixed-idea of others posting here)?
You distinguish "driver" behaviour from "speaker" behaviour. For the most part, we are looking at a "speaker" either way, box or not. However, it may be the right moment to introduce the concept of "air spring" as a more linear element of the system. To the extent that isobaric (or is it isothermal??) air behaviour (with sealed boxes, of course) promotes linearity, that's a good thing. But otherwise, not a distinction that matters. Boxed or not, we are talking about the irregular raw edge of performance that changes with the weather and the day they packed your driver at the factory. Sound like a solid foundation to build on?
Freq response is FAR from the whole story even of how loud your speaker sounds, not to mention all the other aspects of good performance! So that is no way to design. There's nothing wrong with your algebra. But that's not enough.
Ben
Your post lays out the issues very nicely as far as it goes. But here are further issues. At the largest perspective, is the design procedure you are thinking about sensible.... disirregardless of the frequency curve magic (which seems to be the fixed-idea of others posting here)?
You distinguish "driver" behaviour from "speaker" behaviour. For the most part, we are looking at a "speaker" either way, box or not. However, it may be the right moment to introduce the concept of "air spring" as a more linear element of the system. To the extent that isobaric (or is it isothermal??) air behaviour (with sealed boxes, of course) promotes linearity, that's a good thing. But otherwise, not a distinction that matters. Boxed or not, we are talking about the irregular raw edge of performance that changes with the weather and the day they packed your driver at the factory. Sound like a solid foundation to build on?
Freq response is FAR from the whole story even of how loud your speaker sounds, not to mention all the other aspects of good performance! So that is no way to design. There's nothing wrong with your algebra. But that's not enough.
Ben
Last edited:
That article produced the same "effect" in my thinking 😉
A couple of years ago I corrected the low frequency response of a pair of satellite loudspeaker by using the Linkwitz Transform in order to integrate them with the single subwoofer I have available, and I also applied the LT to it in order to get a bit more bass extension.
Now, I'm building the following iteration in order to eliminate the passive xover in the satellites and shape a better FR for the sub.
So, I encourage to apply Mr. Linkwitz's theories because they really works and it is like magic to see/measure the effects of the FR changes on the room and your ears.
Thanks for the courteous reply Ben even though you have strong reservations about the idea. I do know first hand how emotionally involving posting can get when you have strong opinions and deeper knowledge about a subject than the recipient, not a bad thing at all if kept in hand. (And that was meant as a general comment to also remind me to stay open minded, not aimed at yourself in any way). I really appreciate any input you can offer, all proposals need proponents and opponents to be at all meaningful.
I am aware of the relative ease in fiddling with methods of tailoring frequency response, over the years I have played with many attempts to improve my own systems in this way. And just as you say, it isn't always the route to overall happiness as other factors are often more important. I do work in engineering in electronic design with a mathematics background so in the general sense I understand many of the concepts relevant to our present discussion, but I am a long way from an expert in practical speaker design I appreciate.
My current speakers are based on the venerable KEFs I mentioned, B139/B110/T27. The B139s are in 100+litre cylindrical ported enclosures 40cm diameter and 90cm high including the other chambers. I made these in my garden as sheet steel and concrete sandwiches around wood ends, blended into narrow wood baffles just wide enough to take the B139s vertically aligned. Oh to be that young and mad again! They are remarkably dead as you could imagine. The idea was to achieve well stressed materials, avoid diffraction as much as possible, and allow the B139s to breathe naturally, ie in a large unrestrictive enclosure, while controlling system resonances the best I could.
It was a visionary experience to play with tuning those. Firstly, clearly seeing the dual resonance peak and being able to trim the port dimensions to position the port peak to balance the driver resonance was a blast. Then to dampen those peaks out usefully by trimming my bundles of McDonald straws was fun. And all in my living room with the most basic of equipment!
At the time the ultralight v massy enclosure materials debate was raging. I couldn't afford aircraft laminates so opted for the massy approach. Mind you I did make a formica/balsa laminate turntable shelf which took my weight and was absolutely excellent in use. I was also determined to see the benefits of a large cylinder as opposed to a square box. They're ugly but they work really well. Transient response is pretty good considering the design compromises and imaging is good too. But with the current improvements in electronics upstream I do now hanker to investigate whether there is something which may give me improved transient response too. Hence my investigating an all acoustic suspension setup trimmed electronically with the 4-way approach I described.
I still think it could be fun to set up the B169 as I outlined just to see what the results are, only in a simple test cabinet unless it should prove to work really well but I do take your words of caution and won't be so excited about the idea as to override the results.
Incidentally, I was amazed to find out that KEF, even though their days supplying to the DIY market are long over, were incredibly open and helpful when I needed TS parameter details for the B169s I picked up. I take my hat off to them!
I am aware of the relative ease in fiddling with methods of tailoring frequency response, over the years I have played with many attempts to improve my own systems in this way. And just as you say, it isn't always the route to overall happiness as other factors are often more important. I do work in engineering in electronic design with a mathematics background so in the general sense I understand many of the concepts relevant to our present discussion, but I am a long way from an expert in practical speaker design I appreciate.
My current speakers are based on the venerable KEFs I mentioned, B139/B110/T27. The B139s are in 100+litre cylindrical ported enclosures 40cm diameter and 90cm high including the other chambers. I made these in my garden as sheet steel and concrete sandwiches around wood ends, blended into narrow wood baffles just wide enough to take the B139s vertically aligned. Oh to be that young and mad again! They are remarkably dead as you could imagine. The idea was to achieve well stressed materials, avoid diffraction as much as possible, and allow the B139s to breathe naturally, ie in a large unrestrictive enclosure, while controlling system resonances the best I could.
It was a visionary experience to play with tuning those. Firstly, clearly seeing the dual resonance peak and being able to trim the port dimensions to position the port peak to balance the driver resonance was a blast. Then to dampen those peaks out usefully by trimming my bundles of McDonald straws was fun. And all in my living room with the most basic of equipment!
At the time the ultralight v massy enclosure materials debate was raging. I couldn't afford aircraft laminates so opted for the massy approach. Mind you I did make a formica/balsa laminate turntable shelf which took my weight and was absolutely excellent in use. I was also determined to see the benefits of a large cylinder as opposed to a square box. They're ugly but they work really well. Transient response is pretty good considering the design compromises and imaging is good too. But with the current improvements in electronics upstream I do now hanker to investigate whether there is something which may give me improved transient response too. Hence my investigating an all acoustic suspension setup trimmed electronically with the 4-way approach I described.
I still think it could be fun to set up the B169 as I outlined just to see what the results are, only in a simple test cabinet unless it should prove to work really well but I do take your words of caution and won't be so excited about the idea as to override the results.
Incidentally, I was amazed to find out that KEF, even though their days supplying to the DIY market are long over, were incredibly open and helpful when I needed TS parameter details for the B169s I picked up. I take my hat off to them!
- Status
- Not open for further replies.