Thanks for direction. This is good progress.
Have a reference for the numbers in the table above ? I find some publications with different numbers, for example >1ms is reported audible in 300Hz to 1KHz range.
Group delay and phase delay - Wikipedia [Blauert, J.; Laws, P. (May 1978), "Group Delay Distortions in Electroacoustical Systems", Journal of the Acoustical Society of America,]
Thanks!
I'll move one step away from group delay. It's too simple (assuming reasonable XO slopes).
Back to the dilemma: phase jump or phase slope, which is the lesser evil. Probably it depends on drivers (before the enclosure and the room).
It's st work full of downgrades and I get depressed.
The link to recent posts regarding time phase coherent has some interesting, educational and inspiring reading.
I think I'll better start to save money for buying the ideal speaker 1-way 10Hz to 40KHz with no inductance or capacitance, no resonances of any kind, perfect 3D sources... to listen them summer evenings in the garden (aka. no room modes, but is a bit noisy). Also wonder if I'll have enough money left to buy all MQA files that I like. But I have the wine prepared and it will be old enough !
I'll move one step away from group delay. It's too simple (assuming reasonable XO slopes).
Back to the dilemma: phase jump or phase slope, which is the lesser evil. Probably it depends on drivers (before the enclosure and the room).
It's st work full of downgrades and I get depressed.
The link to recent posts regarding time phase coherent has some interesting, educational and inspiring reading.
I think I'll better start to save money for buying the ideal speaker 1-way 10Hz to 40KHz with no inductance or capacitance, no resonances of any kind, perfect 3D sources... to listen them summer evenings in the garden (aka. no room modes, but is a bit noisy). Also wonder if I'll have enough money left to buy all MQA files that I like. But I have the wine prepared and it will be old enough !
Last edited:
I recommend reading this paper on listening tests conducted with model crossovers. AES Convention Papers Forum >> A Virtual Loudspeaker Model to Enable Real-Time Listening Tests in Examine the Audibility of High-Order Crossover Networks Cochenour and Rich conducted blind listening tests for low and high order crossovers. Modeling all in software and listening to synthesized speakers allows for direct comparison of different crossovers quickly and easily.
Oh and by the way. The impulse response of any IIR filter can be sampled to produce the coefficients of the FIR that will then produce exactly the same impulse response within the length of the filter. With sufficient length, that will be within the limits of audibility. There are some very good free college courses available on digital filtering. So all this fiction the OP has put forth about the imperfections of FIR filters appears to be just a lack of understanding of the fundamentals of digital filtering. Every digital IIR filter by its nature contains a FIR filter along with feedback coefficients that give it the infinite response.
For the moment I might have found a solution to what I want: the BO phase link form the 80'. What do you think of it, any opinion or interpretation of details and subtle inferences on components, differences with normal or other systems?
EE-wise: I simulated it, understood it and it seems tractable for DYI too.
Anybody can tell if recent BO follows the same phase-link principles in the background?
Is there anything better (or equal) ever since but with different technology ?
P.S. remember that answer should not contain DSP as mainfront technology for loudspeakers and XO. It may contain it for room correction and stuff like this as a separate module, but not for the basic operation of XO.
EE-wise: I simulated it, understood it and it seems tractable for DYI too.
Anybody can tell if recent BO follows the same phase-link principles in the background?
Is there anything better (or equal) ever since but with different technology ?
P.S. remember that answer should not contain DSP as mainfront technology for loudspeakers and XO. It may contain it for room correction and stuff like this as a separate module, but not for the basic operation of XO.
Hi, check out this rather recent thread, B&O Uniphase was mentioned there, is it same as the phase link you mention? An exercise in converting a speaker to time-phase coherent
I got feeling from the thread that technology used, speaker phase, and many other things are just a part of a speaker system while some parts are more important than others. For example limiting driver operating range related distortions is far more important function in a crossover than maintaining small system phase rotation, a better trade off when designing a system. So, for example first order filters don't make much sense to me with current speaker driver technology we have. Of course opinions and taste vary
These are interesting subjects, but nothing beats good old prototyping and testing and finding out with own ears what matters ( just remember any change in the system sounds always good, proper double blind tests required to find a real difference ). Anyway, always have fun!
I got feeling from the thread that technology used, speaker phase, and many other things are just a part of a speaker system while some parts are more important than others. For example limiting driver operating range related distortions is far more important function in a crossover than maintaining small system phase rotation, a better trade off when designing a system. So, for example first order filters don't make much sense to me with current speaker driver technology we have. Of course opinions and taste vary
These are interesting subjects, but nothing beats good old prototyping and testing and finding out with own ears what matters ( just remember any change in the system sounds always good, proper double blind tests required to find a real difference ). Anyway, always have fun!
Hi Charles,
Let me underline again that I do not build around what components I have. The DIY project that I contemplate is open even to home-build my own drivers, membranes, sicke etc. This open approach apply even to magnets, albeit with less desire (I will build around existing magnets currently offered, of any shape and strengths this does not matter if need be, but I will not order magnetic materials of a wished given shape and then sinter it).
So, the question of driver topology is still open in my approach, until the XO question is answered. One candidate is identified (the phase link technology). I wait feedbacks from you guys. Good is if and when the feedbacks are positive, but most useful will be those negative, specifically if there are any known problems with this XO approach.
Btw, I am waiting for any other hints: how good this "BO Phase Link" in relation with other systems, what are the problems and advantages, etc.
Then, once I decide for one (or maybe two; if available) "better" active XOs, THEN I will work out how to split the bands vs. drivers specs vs. cross check with driver topology (questions like "what topology is best suitable for phase link?", etc), vs. dispersion, vs. shape and materials of enclosure, vs. weight, vs. power, vs. distortion, vs. room, vs. weight, vs. WAF etc.
Example of reasoning: in my understanding "BO phase link" = needs a single very good driver, not 2 or 3. Electro-acoustically the Link driver must be fully capable to fight and correct both LO and HI drivers. Must find one that is not too weak for playing with LO and not too powerful for playing with HI.
Both the LO and the Link drivers must have the same acoustical-suspension hence share the same enclosure (hence Domes are out of the list). No Link driver-resonance in the link region until -20dB (so it can fight electrically easy and in phase). Must be careful of acoustic resonances inside and outside. Hence admittedly it needs also a good topology too (maybe different than what BO used) and also baffle and enclosure details must "be supportive" for the performances (point source, etc).
How much improvement vs. BO phase link? Care to share some precise examples (=products with specs and critics//users descriptions)? In this moment I can use anything for inspiration.
Thanks!
jan
Last edited:
Hi, check out this rather recent thread, B&O Uniphase was mentioned there, is it same as the phase link you mention? An exercise in converting a speaker to time-phase coherent
I got feeling from the thread that technology used, speaker phase, and many other things are just a part of a speaker system while some parts are more important than others. For example limiting driver operating range related distortions is far more important function in a crossover than maintaining small system phase rotation, a better trade off when designing a system. So, for example first order filters don't make much sense to me with current speaker driver technology we have. Of course opinions and taste vary
These are interesting subjects, but nothing beats good old prototyping and testing and finding out with own ears what matters ( just remember any change in the system sounds always good, proper double blind tests required to find a real difference ). Anyway, always have fun!
I'll go for trying my luck. First design and engineer, then build and see what happens. Having DIY fun
The distortion of drivers is clearly a very important point. As answered to Charles, if drivers not available I consider seriously home-building "everything". From scratch and raw materials (I do not consider to buy already made cones and spiders). Everything is in scope, except DSP for XO which is not even out of scope
Cheers,
jan
Hehe, sounds like fun indeed I've found out that, in general, best approach for the average DIYJoe like myself for best Completed DIY speaker to enjoy at home, regarding waf, wallet, room related constraints and personal perversions is to: decide some drivers/general concept, make as large enclosure/horn/baffle as possible/needed, figure out some options/pitfalls and test them with a prototype or two. Measure the box properly, design proper crossover with VituixCAD or similar software that allows to tune the crossover properly, listen the prototype. Figure out if there is something to be bettered in the enclosure, if not start building a final version. Tweak the crossover both listening for weeks and simulating, is it really meeting the specs, especially the sound, set in the beginning? Now there is a complete speaker, and only a year or two passed by
Anyway, best thing to do is build, simulate, listen tweak. While attending shows or visiting friends to have somekind of perspective of the goals, what the system should sound like.
There is a possibility that the selected topology was wrong, but thats only lesson learned, sell the stuff and start fresh. Such fun!
I've found out that, in general, best approach for the average DIYJoe like myself for best Completed DIY speaker to enjoy at home, regarding waf, wallet, room related constraints and personal perversions is to: decide some drivers/general concept, make as large enclosure/horn/baffle as possible/needed, figure out some options/pitfalls and test them with a prototype or two. Measure the box properly, design proper crossover with VituixCAD or similar software that allows to tune the crossover properly, listen the prototype. Figure out if there is something to be bettered in the enclosure, if not start building a final version. Tweak the crossover both listening for weeks and simulating, is it really meeting the specs, especially the sound, set in the beginning? Now there is a complete speaker, and only a year or two passed by Anyway, best thing to do is build, simulate, listen tweak. While attending shows or visiting friends to have somekind of perspective of the goals, what the system should sound like.
There is a possibility that the selected topology was wrong, but thats only lesson learned, sell the stuff and start fresh. Such fun!
If it would not be DIY - that non sense activity at home characterized by much limited resources and time - then I agree that is 100% best practice to do as many tests as needed or as time allows.
Unfortunately my day activity does not comes with "free" (trashable) speakers prototypes, production or repairs of many manufacturers (waf would also not agree to store anything like this). I also do not destroy any functional product I have (again waf rules would be to trash them out instantly).
What I think//hope I can do is to go the route concept-design-build for each and every component if needed. Then you can understand how initial study is very important: it does not worth the time if I could buy them. So they must be really really good, hot and impeccable in every aspect. Something of the kind that my kids would want take with them and take good care of it when they leave home.
I do not think I'd sell anything DIY. Better trashing them directly to bin.
Unfortunately my day activity does not comes with "free" (trashable) speakers prototypes, production or repairs of many manufacturers (waf would also not agree to store anything like this). I also do not destroy any functional product I have (again waf rules would be to trash them out instantly).
What I think//hope I can do is to go the route concept-design-build for each and every component if needed. Then you can understand how initial study is very important: it does not worth the time if I could buy them. So they must be really really good, hot and impeccable in every aspect. Something of the kind that my kids would want take with them and take good care of it when they leave home.
I do not think I'd sell anything DIY. Better trashing them directly to bin.
I wonder what you guys will say about these analytic results and analysis.
Simulation of B&O Phase link for 2-nd order active XO shows a need for more than exceptional "Mid" driver. Aside from Mid requirements to span 3+ decades, it has to provide the function to sum the 90° dephase to either LO or Hi drivers. I will let the problem of FR 3 decades aside, focusing here only on the phase-link task of the Mid driver and its design implications.
If I assume a 3m listening position to equal a minimum 10 air wavelengths (to consider signals "far-field") this computes as 1000Hz.
I assume that this frequency is set to be the point where Hi has 12dB attenuation (as in plot attached). It means that above this frequency the Mid and Hi drivers will be summed by my ears. Known physics and standard design flow for baffle should be enough to try to match the drivers details.
Below 1000Hz the ear position is assumed implicitly in near-field. I consider it would be optimal to sum the LO + Mid acoustic signals in the loudspeaker by acoustic suspension, or sligthly better an isobaric, if no "dual-voice coil per membrane" has been yet invented or realized.
But the Mid in the original 120.2 B&O was made with dome technology, so not possible to function with acoustic suspension. Maybe I am wrong, but for now I prefer to assume that the XO center in 120.2 was quite lowish, in the 300Hz region and the optimal listening distance was 6m (the latter is easy to to by setting 2x bigger price than already expensive technology of LS, hence insure a high probability that a large listening room is available).
The latter Thiel 2.7 apparently uses also Phase Link, but his Mid Phase-Link (if the XO is indeed made so) can do some degree of acoustic suspension with each of LO and HI drivers. I would probably assume here a XO center of 1000Hz and back to 3m listening position, as apparently was indeed the case studied by critics.
Now, creating the dome Mid from B&O is very hard to do, and the coaxial Mid+Hi from Thiel even harder.
First option always exists in theory, to find and buy a functional Mid replacement for these BO or Thiel models, if any available with a good amount if luck.
The second option is to assume different topology and hence drivers too. Good full range drivers exist and can be placed concentric to a coaxial LO+HI setup.
Third option is to study what can offer the 4th order active XO phase linear can offer and maybe we can use more standard drivers (not FR3 decades) (next post).
Simulation of B&O Phase link for 2-nd order active XO shows a need for more than exceptional "Mid" driver. Aside from Mid requirements to span 3+ decades, it has to provide the function to sum the 90° dephase to either LO or Hi drivers. I will let the problem of FR 3 decades aside, focusing here only on the phase-link task of the Mid driver and its design implications.
If I assume a 3m listening position to equal a minimum 10 air wavelengths (to consider signals "far-field") this computes as 1000Hz.
I assume that this frequency is set to be the point where Hi has 12dB attenuation (as in plot attached). It means that above this frequency the Mid and Hi drivers will be summed by my ears. Known physics and standard design flow for baffle should be enough to try to match the drivers details.
Below 1000Hz the ear position is assumed implicitly in near-field. I consider it would be optimal to sum the LO + Mid acoustic signals in the loudspeaker by acoustic suspension, or sligthly better an isobaric, if no "dual-voice coil per membrane" has been yet invented or realized.
But the Mid in the original 120.2 B&O was made with dome technology, so not possible to function with acoustic suspension. Maybe I am wrong, but for now I prefer to assume that the XO center in 120.2 was quite lowish, in the 300Hz region and the optimal listening distance was 6m (the latter is easy to to by setting 2x bigger price than already expensive technology of LS, hence insure a high probability that a large listening room is available).
The latter Thiel 2.7 apparently uses also Phase Link, but his Mid Phase-Link (if the XO is indeed made so) can do some degree of acoustic suspension with each of LO and HI drivers. I would probably assume here a XO center of 1000Hz and back to 3m listening position, as apparently was indeed the case studied by critics.
Now, creating the dome Mid from B&O is very hard to do, and the coaxial Mid+Hi from Thiel even harder.
First option always exists in theory, to find and buy a functional Mid replacement for these BO or Thiel models, if any available with a good amount if luck.
The second option is to assume different topology and hence drivers too. Good full range drivers exist and can be placed concentric to a coaxial LO+HI setup.
Third option is to study what can offer the 4th order active XO phase linear can offer and maybe we can use more standard drivers (not FR3 decades) (next post).
Attachments
A 4-th order phase linear active filter normally comes with minor lobing and also acceptable group delay.
My version was designed to goals to flatten the Mid response, to obtain perfect overlap and synchronicity of LO and Hi main drivers, and also important to use FR2 decade (or less) type of drivers. Another in-scope feature was to separate the Phase-Link drivers: the LO-Link driver should have roughly the same size and be able to better function in acoustic suspension with the main LO driver. Since phase linear XOs always come with symmetry, a Hi-Link driver will also be available. One can use electrical summing (one Hi driver) but I think that I'll prefer to keep the main HI separated, because a) it provides perfect phase overlap with the Lo-main driver, and b) it helps avoiding again Fr3 decade drivers.
See results in the plot below.
I see the following points as advantages too. One, the links now must fight only 135°(45°), not 90° as in the original BO 2nd order approach. Hence these two Link pairs can now function each with its proper technology for their frequencies (with acoustic suspension or isobaric for Lo-Link, and with almost identical diffraction and matching efficiency for the Hi-Link). Second, the market potentially offers us the suitable optimized drivers (W, MB, HB, M-T and T). Third, this concept admits easy possibility to adjust the XO to fit the optimal band and other particularities for each driver.
P.S. Going to higher orders to decrease the phase mismatch (minimize Link drivers fight) from 135° to 315° or 360° in 45° steps/2orders seems impractical to me. Noise and distortions of Opamps will start to dominate, lobing at Link regions will be exacerbated and group delay will defeat any HIFI definitions.
My version was designed to goals to flatten the Mid response, to obtain perfect overlap and synchronicity of LO and Hi main drivers, and also important to use FR2 decade (or less) type of drivers. Another in-scope feature was to separate the Phase-Link drivers: the LO-Link driver should have roughly the same size and be able to better function in acoustic suspension with the main LO driver. Since phase linear XOs always come with symmetry, a Hi-Link driver will also be available. One can use electrical summing (one Hi driver) but I think that I'll prefer to keep the main HI separated, because a) it provides perfect phase overlap with the Lo-main driver, and b) it helps avoiding again Fr3 decade drivers.
See results in the plot below.
I see the following points as advantages too. One, the links now must fight only 135°(45°), not 90° as in the original BO 2nd order approach. Hence these two Link pairs can now function each with its proper technology for their frequencies (with acoustic suspension or isobaric for Lo-Link, and with almost identical diffraction and matching efficiency for the Hi-Link). Second, the market potentially offers us the suitable optimized drivers (W, MB, HB, M-T and T). Third, this concept admits easy possibility to adjust the XO to fit the optimal band and other particularities for each driver.
P.S. Going to higher orders to decrease the phase mismatch (minimize Link drivers fight) from 135° to 315° or 360° in 45° steps/2orders seems impractical to me. Noise and distortions of Opamps will start to dominate, lobing at Link regions will be exacerbated and group delay will defeat any HIFI definitions.
Attachments
What do you think of the above results and points?
If drivers will be found (or made) and baffle design will be successful, will it improve the the 1-Link BO or Thiel solution?
What will be the current competition? Is there anything actually competing with the Link concept? Which one (examples)?
If drivers will be found (or made) and baffle design will be successful, will it improve the the 1-Link BO or Thiel solution?
What will be the current competition? Is there anything actually competing with the Link concept? Which one (examples)?
Will be expensive filterwise, you could measure some and try in a simulator. Textbook acoustic slopes might require lotsa tweaks to achieve. Looks like this relies heavily on multiple drivers doing their job exactly.
Ideally all drivers should be within 1/4 wavelenght c-c istance at crossover frequency, At ~1.5kHz that is something like 2"-3", so expect some lobing, and maybe waters the whole idea?
You could use some good 3.5" fullrange drivers for the mids perhaps and test it out, sounds like fun prototyping time ahead
For quick proto, buy for example four vifa tc9 fullrange drivers slap them to a box and try dsp crossover if you can hammer the phases to be good, then listen if it is worth the trouble. It is easier to get four identical drivers to play well there, tc9 has enough bass and good treble so you don't need bigger woofer or proper tweeter to test how that crossover region midrange would work and how the concept sounds like. There is a 69$ fourchannel dsp amp from some diyaudiomember now on sale, which i don't remember the name of unfortunately tc9 are like 15€ or something
Ideally all drivers should be within 1/4 wavelenght c-c istance at crossover frequency, At ~1.5kHz that is something like 2"-3", so expect some lobing, and maybe waters the whole idea?
You could use some good 3.5" fullrange drivers for the mids perhaps and test it out, sounds like fun prototyping time ahead
For quick proto, buy for example four vifa tc9 fullrange drivers slap them to a box and try dsp crossover if you can hammer the phases to be good, then listen if it is worth the trouble. It is easier to get four identical drivers to play well there, tc9 has enough bass and good treble so you don't need bigger woofer or proper tweeter to test how that crossover region midrange would work and how the concept sounds like. There is a 69$ fourchannel dsp amp from some diyaudiomember now on sale, which i don't remember the name of unfortunately
tc9 are like 15€ or something
Last edited:
Yes, indeed small cone drivers for LO-Links might be the solution here.
Myself I have been looking for 16x of 3" drivers (offers from 3$ up to 10$ each) last days but could not find one that I wanted to match acoustically a 12" LO-main driver (guitar type, to have some musical qualities up to around 2KHz). I guess this simply so because these are designed for cheapo systems.
Unfortunately the next ones of higher quality and better (=suitable) specs I found were 3.5" full range too... but "only" 200Eur each
Thanks for the sweet and better suggestion with Wifa drivers 3,5" for 21$ each and reasonable quality! Four of them should match a 7" LO bass-guitar speaker. I must check them, first on paper.
If no snow outside I won't be able to do any measurements in the garden until March or so. I need 15" of snow or high grass to measure. Now is pretty reflective. Anyway, I am still missing a calibrated microphone. Rest of setup has slowly arrived in the last months (spectrum and THD audio analyzer, high-impedance analyzer. The high-impedance amplifier suitable for loads up to 10mH (probably good for any heavy woofer up to 12" that I know of) I modded myself with good results: sufficient high output impedance and low distortions up to H10 (3dB well beyond the Lo-Link range).
Once I decide which dimension of LO and LO-Link drivers to pair for test, then I'll be able to look for a suitable closed enclosure for them for measurements.
Of course if you see in above analysis any other ideas worth of hints//notes//corrections//warnings from you please advise at any time.
Thanks!
Myself I have been looking for 16x of 3" drivers (offers from 3$ up to 10$ each) last days but could not find one that I wanted to match acoustically a 12" LO-main driver (guitar type, to have some musical qualities up to around 2KHz). I guess this simply so because these are designed for cheapo systems.
Unfortunately the next ones of higher quality and better (=suitable) specs I found were 3.5" full range too... but "only" 200Eur each
Thanks for the sweet and better suggestion with Wifa drivers 3,5" for 21$ each and reasonable quality! Four of them should match a 7" LO bass-guitar speaker. I must check them, first on paper.
If no snow outside I won't be able to do any measurements in the garden until March or so. I need 15" of snow or high grass to measure. Now is pretty reflective. Anyway, I am still missing a calibrated microphone. Rest of setup has slowly arrived in the last months (spectrum and THD audio analyzer, high-impedance analyzer. The high-impedance amplifier suitable for loads up to 10mH (probably good for any heavy woofer up to 12" that I know of) I modded myself with good results: sufficient high output impedance and low distortions up to H10 (3dB well beyond the Lo-Link range).
Once I decide which dimension of LO and LO-Link drivers to pair for test, then I'll be able to look for a suitable closed enclosure for them for measurements.
Of course if you see in above analysis any other ideas worth of hints//notes//corrections//warnings from you please advise at any time.
Thanks!
...or maybe shift the Lo_link XO frequency towards 500Hz and enable using a 12" woofer as main Lo driver and the 12" guitar driver being the Lo-Link in isobaric mode (guitar cone towards listener)?
... or maybe, also for 500Hz Lo-Link side of XO, use dual voice coil woofers from car industry as LO + LO-Link (plus some feedback circuit to adjust the Link and some distortion control functions too) ?
Can one throw an educated estimate which setup will sound better in this configuration... of course without telling to experimenting everything?
... or maybe, also for 500Hz Lo-Link side of XO, use dual voice coil woofers from car industry as LO + LO-Link (plus some feedback circuit to adjust the Link and some distortion control functions too) ?
Can one throw an educated estimate which setup will sound better in this configuration... of course without telling to experimenting everything?
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.