If there’s no significant response anomalies , why would you care?I am not a fan of coaxes mounted on bridges or posts, it’s impossible to prevent tweeter radiation from reflecting backwards off the cone.
Same effect as called SBIR with speaker positioning context. Sound from transducer propagates to all directions (whose wavelength is > transducer size), reflects towards listener from the boundary behind. This makes comb filter with direct sound. Path length for the reflection is a round trip so comb filter starts quite low in frequency.
If dome tweeter was 1" away from the cone, the reflection has 2" extra path length compared to direct sound and first interference null of the comb filter is at 4" wavelength (2" half wavelength), so about 3,4kHz. The comb filter fades away with increasing directivity, and in case of 1" dome directivity increases about 1" wavelength so about 14kHz. Basically the whole tweeter bandwidth is plagued by the comb filter.
Same effect is responsible for the ~1300Hz dip in the Beyma coax sim Gaga posted bit earlier in the thread. 1300Hz is ~26cm long so 13cm extra path length for a reflection does it. Gaga thought it is due to diffraction, but diffraction has the backwave phase reversed so needs to be delayed full wavelength, 26cm, for 1300Hz dip which in this case and that would be outside the driver.
edit. 50cm baffle, or edge 25cm from driver makes interference like this, first dip at ~1300Hz. So, the beyma tweeter dip is also the SBIR.

Difference of the Beyma coax to the dome coax is that the waveguide controls directivity lower in frequency so the comb filter is mainly on the low end of it's bandwidth. That's why they make the waveguide beam, or it likely beams for other reasons, and it's good thing to avoid this SBIR effect in the response, except beaming isn't that desirable either, in many use cases.
If dome tweeter was 1" away from the cone, the reflection has 2" extra path length compared to direct sound and first interference null of the comb filter is at 4" wavelength (2" half wavelength), so about 3,4kHz. The comb filter fades away with increasing directivity, and in case of 1" dome directivity increases about 1" wavelength so about 14kHz. Basically the whole tweeter bandwidth is plagued by the comb filter.
Same effect is responsible for the ~1300Hz dip in the Beyma coax sim Gaga posted bit earlier in the thread. 1300Hz is ~26cm long so 13cm extra path length for a reflection does it. Gaga thought it is due to diffraction, but diffraction has the backwave phase reversed so needs to be delayed full wavelength, 26cm, for 1300Hz dip which in this case and that would be outside the driver.
edit. 50cm baffle, or edge 25cm from driver makes interference like this, first dip at ~1300Hz. So, the beyma tweeter dip is also the SBIR.

Difference of the Beyma coax to the dome coax is that the waveguide controls directivity lower in frequency so the comb filter is mainly on the low end of it's bandwidth. That's why they make the waveguide beam, or it likely beams for other reasons, and it's good thing to avoid this SBIR effect in the response, except beaming isn't that desirable either, in many use cases.
Last edited:
The only few to not suffer from this issues are from Fulcrum Acoustics but it requires extensive FIR to tame the artifacts.
Yeah perhaps it can be juggled so that it's not big of an issue. If it fixes some issue that is audibly worse, then it's worth it. Like uniform sound to all of the 100 people in the audience, no one gets perfect sound, but also no one gets the worst, but all get pretty nice. For home use, perhaps use coax where the tweeter doesn't protrude from the woofer to avoid this stuff.
It worked pretty well on the Presonus Scepter ( which used a licence from D.Gunness-Fulcrum owner) which is a nearfield monitor.
I must say i was very sceptical when i first listened to it because i was biased ( i don't like the idea to have a protubing horn over a direct radiating woofer). 10 second listening convinced me, they are on par with the big Tannoy i liked this much.
Fulcrum acoustics offer a wide range of this kind of coax and they claim that thanks to the fir treatment they all sound the same and are swappable from 8" to 15". I never heard them but the Scepter sure sounded damn right to me so i would tend to believe Fulcrum claims.
I must say i was very sceptical when i first listened to it because i was biased ( i don't like the idea to have a protubing horn over a direct radiating woofer). 10 second listening convinced me, they are on par with the big Tannoy i liked this much.
Fulcrum acoustics offer a wide range of this kind of coax and they claim that thanks to the fir treatment they all sound the same and are swappable from 8" to 15". I never heard them but the Scepter sure sounded damn right to me so i would tend to believe Fulcrum claims.
Gunnes sonic focus / temporal EQ, the paper is very interesting. Unfortunately DIY scene has not yet dissected what's going on with his filters. Several years ago @mark100 had a thread where this came up and there was some attempt with the stuff here on diyaudio 😀 mine was limited trying to measure, but lack of knowledge got me nowhere. I think mark100 got into FIR then, not sure he has done any of this sonic focus stuff since.
edit. here is the thread
https://www.diyaudio.com/community/threads/measuring-horn-reflections-back-into-the-throat.337806/
edit. here is the thread
https://www.diyaudio.com/community/threads/measuring-horn-reflections-back-into-the-throat.337806/
Last edited:
Well I have an idea to be tested. Point tweeter up and fire into a 45deg deflector coaxial with the dustcap, so as to align acoustic centers.
Inspiration came from reflecting telescope optics.
Inspiration came from reflecting telescope optics.
Hi, the problem is always the same: two physical objects (transducers) outputting same wavelength (at crossover) and they both are in close proximity of each other. If either of them needs to be efficient at the output, and have some control over the wavelength, it's by necessity disturbing the other. The sound inevitably interacts not just for the transducer that emits it but also the structure of the other near by. The woofer interacts with the tweeter sound, and tweeter with the woofer sound around xo frequency. If the tweeter is front of the woofer reflection and diffraction happens, and while these might not be that problematic it would be ideal to avoid any reflections and diffraction issues altogether to get acoustic response to some specified bracket, within +-3db for example. This leaves the option to have the tweeter and the woofer same physical object, basically tweeter recessed into the woofer somehow, like some coaxials have it, or woofer recessed into the tweeter, MEH basically.Well I have an idea to be tested. Point tweeter up and fire into a 45deg deflector coaxial with the dustcap, so as to align acoustic centers.
Inspiration came from reflecting telescope optics.
Same issue is with any multiway speakers transducers sharing a baffle: woofer is likely at tweeter proximity affecting the tweeter response being there physically, and vice versa. This cannot be eliminated as we live in a physical world, but can be reduced by making them each physically separate and have them deal with their own structure diffraction. Now they would still interact, affect each other acoustic response, but the interference is more somewhere else than on the listening window. Or, coax or meh, just deal with the diffraction and reflections, try to minimize them as they are secondary sounds making interference with direct sound.
Last edited:
I think mark100 got into FIR then, not sure he has done any of this sonic focus stuff since.
Mark regularly ask about a way to 'reverse' an impulse response so i'm pretty sure he still haven't found a way to apply this. 😉
From some video this seems to be a large part of temporal eq process, inverse impulse response ( probably an average from multiple location recording) and apply it(them) to the driver(s).
Some 'room correction' software offer this kind of thing ( open DRC iirc, the one Ronald used with his line array) as there is function to automatically compensate for anomalies in step response.
Without proper knowledge of how FIR actually works I think it's like making secondary sound that comes after the actual output. Not so much reqular EQ where frequency response is corrected live, but the initial sound is left intact and the filter adds a delayed sound that cancels some of the reflections. So, what one has to do is to measure, or calculate, a reflection and then make the inverse impulse to that. Initial sound (impulse) intact, but a secondary sound (impulses) that come after it are counteracted, noise cancelling of sorts. Perhaps noise cancelling algorithms could give some hint how it's done?
We had idea to make quick impulse with the transducer and then measure with the transducer itself how it behaves after the impulse and cancel that. But I don't know how to actually do that properly and even if I did I have no idea how to implement FIR filter to cancel it 🙂 I think mark100 asked about this from Gunnes and got back a smile so perhaps it could work 🙂 Gunnes paper hints they use some math, and some measurements to make such filter, but I don't know how they actually do it. Some FIR pro might be able to explain this stuff better and perhaps figure out how to do it, how to measure and how to implement filter for it.
We had idea to make quick impulse with the transducer and then measure with the transducer itself how it behaves after the impulse and cancel that. But I don't know how to actually do that properly and even if I did I have no idea how to implement FIR filter to cancel it 🙂 I think mark100 asked about this from Gunnes and got back a smile so perhaps it could work 🙂 Gunnes paper hints they use some math, and some measurements to make such filter, but I don't know how they actually do it. Some FIR pro might be able to explain this stuff better and perhaps figure out how to do it, how to measure and how to implement filter for it.
Last edited:
Well I have an idea to be tested. Point tweeter up and fire into a 45deg deflector coaxial with the dustcap, so as to align acoustic centers.
Inspiration came from reflecting telescope optics.
It's an idea but i fear there is other issue at play: how would you manage directivity behavior? It have to be (at least partially) the same as the woofer driver in area surrounding xover, or you have a 'step' in directivity. This is not something rare amongst coax 'Tannoy' style.
In fact most of them have this issue to some degree, exception being latest Kef and some Tannoy. It's one of the compromise in the principle we have to live with.
Sometime i wonder if it's not one of the reason for which i prefer large diameter coax ( with their inherent highest dynamic capability).
For sure even if this is better managed with Kef driver i own and they are (almost) 'time aligned' by design they definitely lack in membrane area for my prefererences ( low sensibility and lack of dynamic if used in low mid).
So, what one has to do is to measure, or calculate, a reflection and then make the inverse impulse to that. Initial sound (impulse) intact, but a secondary sound (impulses) that come after it are counteracted, noise cancelling of sorts. Perhaps noise cancelling algorithms could give some hint how it's done?
...
We had idea to make quick impulse with the transducer and then measure with the transducer itself how it behaves after the impulse and cancel that. But I don't know how to actually do that properly and even if I did I have no idea how to implement FIR filter to cancel it
I think you are on right track and this is how it's performed in openDRC ( it 'realign' step response) . But i think there is something more to manage: the back of the horn profile: if it's shape offer something 'linear' in it's response or the way it interact then it would be a lot easier to approach than something non-linear.
I think mark100 asked about this from Gunnes and got back a smile so perhaps it could work 🙂
If i were Mr. Gunness and had to protect something i earn money from i would answer a smile to any question asked whatever close or far from the principle. 😉
Gunnes paper hints they use some math, and some measurements to make such filter, but I don't know how they actually do it. Some FIR pro might be able to explain this stuff better and perhaps figure out how to do it, how to measure and how to implement filter for it.
How ant what to measure is the real issue imho. Math behind, there is a bunch of geniuses able to apply and figure way to compensate for anything. But define what to compensate and gather measurements might be the issue imho.
Yep, how to measure a thing without having other things with the mix?🙂 Perhaps one could use impedance tube or something ,to measure compression driver phase plug stuff, or just the math to compensate for transmission to the phase plug gaps, series of ~50% transmission with slot spacing delay as per Gunnes paper. Woofer cone resonance, how to measure that with a mic? infinite baffle perhaps, and then knowledge at which frequency and delay the correctable issues could be, some processing and perhaps it becomes visible in the measurement. But such stuff reflects to voice coil as well I think, so perhaps just measuring voltage and current on the wires could be enough? Mouth reflection back to throat, well, I guess not all sound goes to throat but reflects earlier in the profile back out so not a thing for mic to measure, or perhaps it is. Here the stuff that actually comes through the throat can be measured with the transducer again. All of which comes back inside move the diaphragm could be counteracted with it, quite perfectly I think.
Perhaps measurements aren't that hard, but how to process them? and how to make the suitable FIR then? I hope someone solves this, it would be nice addition to DIYer toolkit. Although, some system designs overcome this stuff, like good waveguides have no reflection back to the device. Compression driver phase plug issues would be nice to overcome, but perhaps even this is not a problem with modern small good compression drivers as they show quite exceptional performance in this regards. Well, woofer cone surround resonances then? why not. Perhaps there is some to solve.
edit. I think this stuff in compression driver measurements is the transmission stuff in the phase plug, only part of sound enters a slot and rest continues past it exiting the next one, until few times most of the sound got through. On long wavelengths this is not noticeable, but when wavelength gets short delay gets greater compared to wavelegth and we see this stuff, typically worse on bigger diaphragm drivers, and quite good with the small ones:

screenshot from https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-7767369
Perhaps measurements aren't that hard, but how to process them? and how to make the suitable FIR then? I hope someone solves this, it would be nice addition to DIYer toolkit. Although, some system designs overcome this stuff, like good waveguides have no reflection back to the device. Compression driver phase plug issues would be nice to overcome, but perhaps even this is not a problem with modern small good compression drivers as they show quite exceptional performance in this regards. Well, woofer cone surround resonances then? why not. Perhaps there is some to solve.
edit. I think this stuff in compression driver measurements is the transmission stuff in the phase plug, only part of sound enters a slot and rest continues past it exiting the next one, until few times most of the sound got through. On long wavelengths this is not noticeable, but when wavelength gets short delay gets greater compared to wavelegth and we see this stuff, typically worse on bigger diaphragm drivers, and quite good with the small ones:

screenshot from https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-7767369
Last edited:
If the tweeter is front of the woofer reflection and diffraction happens
Thank you both for detailed replies. I'm a telescope nut but this idea only came to me a few days ago (after LXing "Metals" shown in the fullrange gallery) so it's surely half-baked. I have pretty limited coaxial experience: (1) co-axing a 35mm car ceramic tweeter ringed by plast-tape-tulip-WG in front of wideband 15"; (2) passive-XOing Tannoy Precision 8 (1st-order HPF/2nd-order LPF); and passive-XOing KEF LS50 Meta 1793 (~2nd-order HPF/3rd-order LPF). For me they were challenges. I think the three issues reflection, diffraction, and directivity could all be tweaked/massaged away, perhaps more easily/parsimoniously than before, given the additional freedoms of this crazy idea. For example: use of baffles (as in optics) or non-flat deflector surface; free-positioning-and-angling of tweeter/deflector to align acoustic centers.how would you manage directivity behavior? It have to be (at least partially) the same as the woofer driver in area surrounding xover, or you have a 'step' in directivity.
Hi, yeah it is an interesting idea, but can you get the last reflector coaxial to woofer? if not, then the reflector as the "final" sound source for treble is not coincident with the woofer and same interference issues ensue as with normal stacked tweeter and woofer, that are not coincident, no? You also need to control beamwidth to quite narrow beam in order to reflect it and not have sound leak beside it. This would already happen without reflectors with tweeter outputing sound through the woofer magnet, as most coaxials do.
edit. I'm not able to imagine a working solution, but perhaps you have it figured out? There is no way o have it better than traditional stacked system or any coaxials, unless serious directivity mismatch between woofer and tweeter (reflector) sound, which isn't good for at least some applications, where power response matters.
edit. I'm not able to imagine a working solution, but perhaps you have it figured out? There is no way o have it better than traditional stacked system or any coaxials, unless serious directivity mismatch between woofer and tweeter (reflector) sound, which isn't good for at least some applications, where power response matters.
Something like a Fresnel lens ( used in LightHouse to focus light beam) could be effective to manage directivity and coupled to delay could solve acoustic center and directivity but it would be a large diffraction source too if a membrane is colocated in rear ( to reflect sound you need an object 3x the size of the wavelength of interest, eg at 1khz it means circa 1m ).
Beyma had some 15" with diffraction slot (15xa38 or something) but it souded somewhat diffuse to me in highs ( a character that was already described by others about coax highs).
Maybe you are into something idk. Could you sketch what ou have in mind?
Beyma had some 15" with diffraction slot (15xa38 or something) but it souded somewhat diffuse to me in highs ( a character that was already described by others about coax highs).
Maybe you are into something idk. Could you sketch what ou have in mind?
Yeah, laser light is about single wavelength and relatively easy to control as about single sized physical object can do it. Visible light in general is quite easy to reflect due to very short wavelength and relatively small objects required to produce and reflect. As you say, sound has huge wavelength in comparison, and we cannot limit it to one wavelength as we need the whole audible spectrum for music so the lensing would become quite big. Problem is the crossover region where same wavelength is sourced from multiple physical objects, which inevitably need to be similar in size to get similar control for the overlapping bandwidth, and by definition cannot colocate, or get to 1/4wl spacing because they need to be bigger than that. If the objects do not have to have any control over the overlapping bandwidth, they can be much smaller than wavelength at crossover frequency, a multiway speaker for home use could be such (SPL capability is limited by the size).
So it's the wavelength and physical objects we deal with, and there is no way around it, other than utilize knowledge what is audible and what is not, and try to make the swap, the crossover, where it's not so audible and with side effects that are less audible. On some application it might help to diver issues from listening window to some other direction, while on some application it might be benefit to try and have issues uniformly compromised to any direction.
Traditional speaker could have quite flawless response toward listening window, compared to coaxial with some of the issues like the SBIR thing with tweeter, but have less than ideal towards some off-axis angles (and power response) where there is lobing at crossover. So, perhaps the coaxial is just fine solution to many situations, like for dancing or for really near field listening where movement can change listening axis many degrees. Small coax with dome in the apex for home use might be just the ticket if listening spot varies. For more SPL capability perhaps MEH works better, as the 3D configuration of drivers inside the device are basically projected into 2D sound source (the mouth) which enables big transducers be close together and enable point source, which could cater hundreds of people with pretty much similar sound for all. Maybe the traditional stacked drivers configuration is just fine and no need to lust for coaxial system.
Of course some issues and compromises on all of these, but what ever works best for one application then it will, while something else is better for some other application. Perhaps many configurations are fine for many applications, it's just the critical listener and hifi tinkerer who are interested and try to push it to extreme 😀 and the pro audio companies.
And yeah, single fullrange driver can do it, except when it truly is fullrange it beams and has no SPL capability. Fine for some applications, while not so great for another. Conversely, if small fullrange driver has enough SPL for application then go for it, no need to complicate further. Need more bass? need to address interaction with room somehow? then start looking how to get these and some kind of a multiway it is. There is not much magic in any of this, my posts are just simple logic with sound wavelength, which is the fundamental that defines the whole game with sound.
So it's the wavelength and physical objects we deal with, and there is no way around it, other than utilize knowledge what is audible and what is not, and try to make the swap, the crossover, where it's not so audible and with side effects that are less audible. On some application it might help to diver issues from listening window to some other direction, while on some application it might be benefit to try and have issues uniformly compromised to any direction.
Traditional speaker could have quite flawless response toward listening window, compared to coaxial with some of the issues like the SBIR thing with tweeter, but have less than ideal towards some off-axis angles (and power response) where there is lobing at crossover. So, perhaps the coaxial is just fine solution to many situations, like for dancing or for really near field listening where movement can change listening axis many degrees. Small coax with dome in the apex for home use might be just the ticket if listening spot varies. For more SPL capability perhaps MEH works better, as the 3D configuration of drivers inside the device are basically projected into 2D sound source (the mouth) which enables big transducers be close together and enable point source, which could cater hundreds of people with pretty much similar sound for all. Maybe the traditional stacked drivers configuration is just fine and no need to lust for coaxial system.
Of course some issues and compromises on all of these, but what ever works best for one application then it will, while something else is better for some other application. Perhaps many configurations are fine for many applications, it's just the critical listener and hifi tinkerer who are interested and try to push it to extreme 😀 and the pro audio companies.
And yeah, single fullrange driver can do it, except when it truly is fullrange it beams and has no SPL capability. Fine for some applications, while not so great for another. Conversely, if small fullrange driver has enough SPL for application then go for it, no need to complicate further. Need more bass? need to address interaction with room somehow? then start looking how to get these and some kind of a multiway it is. There is not much magic in any of this, my posts are just simple logic with sound wavelength, which is the fundamental that defines the whole game with sound.
Last edited:
- Home
- Loudspeakers
- Multi-Way
- what is the point of expensive coaxials with ragged response?