Yes, you've raised many sound caveats but I think equally applicable to traditonal multi-way and coaxial speakers. If my 15" ~94dB-sensitive wideband flat to ~4.5khz were listened to from truly afar, my unobtainium dipole tweeter Naturelle placed on top would have sufficed. But inside ~2m the situation called for coax. This was done during the pandemic before I started aligning acoustic centers. I was convinced the plast-tape ad hoc tulip-waveguide raised the 92dB-sensitive ceramic-dome tweeter by a couple dB and changed dispersion/directivity so that musical instrument tonality was fine reasonably far off-axis (at a listening distance of 1-2m -- high SPL unbearable) and comb-filtering was not observed (heard). Now I want to try aligning acoustic centers, not just (trivially) phase at XO. The dustcap on the 15" is pretty large so there's room to work (without nuking it).
Consider a thought experiment: if XO is 10khz (wavelength 3.4cm) and tweeter wave bounces off a 10cm-minor-axis soft-edged deflector co-axial with the cone, time-aligned. Then lower the XO and baffle-step kicks in -- but the HPF curve is changed too and compensates somewhat.
Consider a thought experiment: if XO is 10khz (wavelength 3.4cm) and tweeter wave bounces off a 10cm-minor-axis soft-edged deflector co-axial with the cone, time-aligned. Then lower the XO and baffle-step kicks in -- but the HPF curve is changed too and compensates somewhat.
Last edited:
Hi guys, fun conversation ......I'll through a few random thoughts, replies/comments.
Re coaxials in general. I like the idea of a waveguide in front of the woofer, like Fulcrum uses. I think this is better than using the woofer cone as part of the wave guide like many coaxes do.
I think the trick with a waveguide in front of the woofer, is having a low enough xover frequency for the high end of the woofers response to be able to bend around the horn without too much anomaly.
tmuikku, yeah ..i really agree with your take of an unenclosed tweeter on a bridge ...bad idea, imo.
Speaking of Fulcrum. When I had the good fortune to ask Dave G if my experiments with using one section of a coax CD as a mic, and the other as the driving element....for the purpose of isolating and measuring reflections back into the horn to build some Temporal EQ...he smiled while saying "that will work".
So for a while I concentrated on trying to refine that technique and building a FIR file solely for that task. Which wasn't hard and worked fine.
But, there was still the basic task of flattening the CD's mag and phase on whatever horn for which I had built a return reflection-nulling FIR.
And it all ends up going into one FIR file. I found there was little if any difference in building the final FIR file in stages of simply doing it all at once by taking a normal measurement.
Return horn reflections, just like basic response ripples/anomalies, just like diffraction, are all part of the impulse response stretched out wherever they occur in time.
I may be a bit off, but I view TEQ as a super technique idea, but more basic and easy to apply in terms of how to correct with FIR, than I first thought.
It's worth noting that the FIR files Fulcrum provides for a variety of processors, often also correct frequencies below the CD/horn section.
krivium, another yeah...I'm constantly asking about inverting impulse responses. Not "how to invert", however. That's the easy well known part.
The questions for me always come down to "what to" invert. As in what are the deviations from flat response that are truly minimum phase and time invariant,...... and stay so both on and off axis.
Ime/imo, the big problem with impulse inversion and FIR, is that it is blind in terms of what to correct. It doesn't give a damn...it just corrects everything in the impulse.
A valid measurement is everything. Same holds true for all forms of tuning, passive, active IIR, of FIR....huh?
It's just FIR is so much more powerful, fully correcting, than the others, that it becomes a two edged sword, cutting the wielder if used improperly.
Heck, I'm still a student of quasi-anechoic speaker tuning. Room correction via FIR (which simply some specified form of impulse inversion) is over my head.
Re being a student. I'm on such a processing experimentation kick! Sprung for a prosound measuring, processing simulation, and FIR generator all in one program called Crosslite+. It's so fast to use, the other day just for kicks, I raced to see how fast I could tune a 4-way (syn10) from total scratch.
Starting with no raw measurements, and then building FIR files with EQ's and crossovers for each way then adding delays and levels. 22 minutes !!
Also, I found that FirDesigner has a nifty auto-IIR generator, that matches raw response to target.
Like REW it also has choices for export to a specific manufacturer, although directly as biquads rather than PEQ nomenclature.
Q-sys is supported and allows 1 - 256 biquads per channel.......approaching the overcorrection damage capability of FIR Lol.
Set fractional octave correction, set frequency range, set number of IIR filters to generate, set max gain per filter, and hit generate. Save coefficients, import into q-sys IIR Custom filter block. Done. 1-2 minutes for complete IIR correction post measurement.
I've been playing with both IIR preconditioning to lower the tap-count using FIR filters, as well as full out IIR processing.
Active processing tools and DSP hardware are crazy good today, huh? Honestly, I think mastering active processing bears more sonic fruit than sweating over any other particular DIY speaker aspects, ....other than basic acoustic design along with the general driver selections for it.
ps..sorry for the length......hope i didn't ramble too much
Re coaxials in general. I like the idea of a waveguide in front of the woofer, like Fulcrum uses. I think this is better than using the woofer cone as part of the wave guide like many coaxes do.
I think the trick with a waveguide in front of the woofer, is having a low enough xover frequency for the high end of the woofers response to be able to bend around the horn without too much anomaly.
tmuikku, yeah ..i really agree with your take of an unenclosed tweeter on a bridge ...bad idea, imo.
Speaking of Fulcrum. When I had the good fortune to ask Dave G if my experiments with using one section of a coax CD as a mic, and the other as the driving element....for the purpose of isolating and measuring reflections back into the horn to build some Temporal EQ...he smiled while saying "that will work".
So for a while I concentrated on trying to refine that technique and building a FIR file solely for that task. Which wasn't hard and worked fine.
But, there was still the basic task of flattening the CD's mag and phase on whatever horn for which I had built a return reflection-nulling FIR.
And it all ends up going into one FIR file. I found there was little if any difference in building the final FIR file in stages of simply doing it all at once by taking a normal measurement.
Return horn reflections, just like basic response ripples/anomalies, just like diffraction, are all part of the impulse response stretched out wherever they occur in time.
I may be a bit off, but I view TEQ as a super technique idea, but more basic and easy to apply in terms of how to correct with FIR, than I first thought.
It's worth noting that the FIR files Fulcrum provides for a variety of processors, often also correct frequencies below the CD/horn section.
krivium, another yeah...I'm constantly asking about inverting impulse responses. Not "how to invert", however. That's the easy well known part.
The questions for me always come down to "what to" invert. As in what are the deviations from flat response that are truly minimum phase and time invariant,...... and stay so both on and off axis.
Ime/imo, the big problem with impulse inversion and FIR, is that it is blind in terms of what to correct. It doesn't give a damn...it just corrects everything in the impulse.
A valid measurement is everything. Same holds true for all forms of tuning, passive, active IIR, of FIR....huh?
It's just FIR is so much more powerful, fully correcting, than the others, that it becomes a two edged sword, cutting the wielder if used improperly.
Heck, I'm still a student of quasi-anechoic speaker tuning. Room correction via FIR (which simply some specified form of impulse inversion) is over my head.
Re being a student. I'm on such a processing experimentation kick! Sprung for a prosound measuring, processing simulation, and FIR generator all in one program called Crosslite+. It's so fast to use, the other day just for kicks, I raced to see how fast I could tune a 4-way (syn10) from total scratch.
Starting with no raw measurements, and then building FIR files with EQ's and crossovers for each way then adding delays and levels. 22 minutes !!
Also, I found that FirDesigner has a nifty auto-IIR generator, that matches raw response to target.
Like REW it also has choices for export to a specific manufacturer, although directly as biquads rather than PEQ nomenclature.
Q-sys is supported and allows 1 - 256 biquads per channel.......approaching the overcorrection damage capability of FIR Lol.
Set fractional octave correction, set frequency range, set number of IIR filters to generate, set max gain per filter, and hit generate. Save coefficients, import into q-sys IIR Custom filter block. Done. 1-2 minutes for complete IIR correction post measurement.
I've been playing with both IIR preconditioning to lower the tap-count using FIR filters, as well as full out IIR processing.
Active processing tools and DSP hardware are crazy good today, huh? Honestly, I think mastering active processing bears more sonic fruit than sweating over any other particular DIY speaker aspects, ....other than basic acoustic design along with the general driver selections for it.
ps..sorry for the length......hope i didn't ramble too much
I dug out the 15" EchoTech big-basin (actually more than meets the eye) and realized I could just bounce the tweeter off of the dustcap, so I gave it a try. Whipped up a series 1st-order XO around 4khz (tweeter reverse polarity for convenience or subconscious rationale I'm not sure), and simply moved the ceramic dome around, pointed toward dustcap, while playing test tones 1k-12khz or high violin music. Got feedback very quickly -- high frequency does bounce, and this way I got pretty even response to 11.5khz (limit of my hearing confidence) and even dispersion to ~25deg on the side not in-line with the tweeter. I fitted a steamer rack and also tried flat disk over the dustcap etc. For this quick test listening distance up to ~1.2m (standing with head lowered looking sideways). The tweeter was a couple dB less sensitive than the widerange but acceptable once time- and phase-aligned; before, tweeter facing front, I had made a plast-tape whizzer for it. Sounded great. Next holiday I might try comparing the two versions.
Of course, if I kept the 15" up-firing I'd just point the tweeter toward me (LX).
Last edited:
p.s. series 1st-order ~4khz 0.3mH 4.7uF now wired in-phase
Three-way comp: "coax" unaligned acoustic centers vs bounce-aligned vs LX-aligned
https://www.diyaudio.com/community/threads/full-range-speaker-photo-gallery.65061/post-7783353
Three-way comp: "coax" unaligned acoustic centers vs bounce-aligned vs LX-aligned
https://www.diyaudio.com/community/threads/full-range-speaker-photo-gallery.65061/post-7783353
Last edited:
Right, please see above tweeter backwards. (In optics the analogue is called an "off-axis unobstructed Cassegrain telescope", more or less.) Surely I cannot be the first crazy person to try something as simple as turning over the bridge. Everyone, find me precedents please.
Near-field the time and phase bounce-aligned off-coaxial is insanely palpably holographically realistic; in nested washbasins response falls off gradually below 230hz, transcient perfect ultra-fi above. Wide dispersion above 10khz thanks to convex dustcap (and custom deflector could be designed and mounted). I could hear the twang in plucked strings I hadn't heard before.
p.s. I think the KEF LS50 Meta 1793 coaxial has massive cross-contamination between tweeter/phase-plug/cone, by design to be sure, but still required asymmetric high-order XO to tame (3rd-order LPF 0.46mH/6.8uF-0.28mH, 2nd-order HPF 2.5uF/0.47mH; a real challenge for me, reported in the How good are our DIY thread).
https://www.diyaudio.com/community/threads/full-range-speaker-photo-gallery.65061/post-7769008
Last edited:
How about attaching the tweeter to the cone, suspended say 3mm above the cone and pointed towards the cone, so the space between cone and tweeter forms a horn aperture?
If the sides of the tweeter curved back, the space between the dust cap and the tweeter would act more like a horn than a reflector. I don’t think response would be great on the woofer’s axis, but if the woofer were tilted 45 degrees it might be just about right.
The tweeter needs to be low mass and attached to the cone with spacers.
If the sides of the tweeter curved back, the space between the dust cap and the tweeter would act more like a horn than a reflector. I don’t think response would be great on the woofer’s axis, but if the woofer were tilted 45 degrees it might be just about right.
The tweeter needs to be low mass and attached to the cone with spacers.
The gap needs to be static hence attach to dustcap with spacers? Added weight might eat too much efficiency of the woofer? at least sets special requirements for dust cap rigidity. Perhaps fine if the tweeter structure was designed as part of the cone. If there was no dustcap you'd have more control of it all. Both these cases one could just flip the tweeter and it'd be more traditional coax without issues with objects on the way reflecting and diffracting.
Whats the main issue with coaxials, what is it that needs fixing? response around xo? but that is fixed with just using a fullrange driver, and trade off is less than ideal radiation pattern of the treble. Add tweeter to fix the treble and now the problem is back around the crossover. So just use either scheme which suits better your application, and make it as good as you can. There is no escaping this because sound wavelength is very short on highs and very long on lows.
This and the off-axis source are variations of same scheme solving one downside but another follows, which is due to static objects and varying wavelength. Until physics change it's just matter of choosing which arrangement suits best ones application/interest.
It would be really cool if you guys make original design work, but I'm afraid there is trade-offs just like with any variation of full audible bandwidth speaker. Which doesn't mean it couldn't be better than all the variations we have so far, so I wish you best of luck!🙂
ps. Easy start trying to tacle the problem is take vituixcad, or BEM simulator, and start simulating with ideal drivers that have no physical issues like surrounds or motors, just ideal response and any physical shape, just to simplify things. Task is to make flawless response by any means. Can you do it? for sure, just connect wire from source to driver when you first open vituixcad and it's point source without any physical size. Or BEM sim and make 5mm sized sphere, or big wavefuide with small source at the throat. But, there is no means to implement these into reality to cover full bandwidth. For example calculate how much SPL 5mm diameter sphere can make at any frequency? The sound source would need to be made out of nothing, have no physical bounds to it. Since this kind of technology isn't available all we can do is juggle around with various schemes we do have, physical objects making sound.
Whats the main issue with coaxials, what is it that needs fixing? response around xo? but that is fixed with just using a fullrange driver, and trade off is less than ideal radiation pattern of the treble. Add tweeter to fix the treble and now the problem is back around the crossover. So just use either scheme which suits better your application, and make it as good as you can. There is no escaping this because sound wavelength is very short on highs and very long on lows.
This and the off-axis source are variations of same scheme solving one downside but another follows, which is due to static objects and varying wavelength. Until physics change it's just matter of choosing which arrangement suits best ones application/interest.
It would be really cool if you guys make original design work, but I'm afraid there is trade-offs just like with any variation of full audible bandwidth speaker. Which doesn't mean it couldn't be better than all the variations we have so far, so I wish you best of luck!🙂
ps. Easy start trying to tacle the problem is take vituixcad, or BEM simulator, and start simulating with ideal drivers that have no physical issues like surrounds or motors, just ideal response and any physical shape, just to simplify things. Task is to make flawless response by any means. Can you do it? for sure, just connect wire from source to driver when you first open vituixcad and it's point source without any physical size. Or BEM sim and make 5mm sized sphere, or big wavefuide with small source at the throat. But, there is no means to implement these into reality to cover full bandwidth. For example calculate how much SPL 5mm diameter sphere can make at any frequency? The sound source would need to be made out of nothing, have no physical bounds to it. Since this kind of technology isn't available all we can do is juggle around with various schemes we do have, physical objects making sound.
Last edited:
Coffee break philosophy:
Yeah also something that is associated two transducers at very close proximity affecting each others response. Same stuff plagues single transducer as well: because cone area is static the excursion must vary with wavelength (frequency) to maintain flat SPL. Difference of how much more cone moves for long wavelength to short gets bigger the wider the bandwidth, double per every octave assuming ideal omni source.
So, reproducing a low frequency with same SPL as high frequency makes the high frequency modulate. If one wishes to maintain same excursion and SPL for both long and short wavelength to reduce modulation, the radiating area must change instead, and this cannot happen ideally but in steps as in coaxial, small driver in middle of bigger one. Idea with fullrange drivers is the same, somewhat controlled breakup so that highs are reproduced with small Sd, while rest of the cone not participating, except perhaps being bit out of control and cancelling out some to various directions. And since there is no full control of the breakup, the response ends up quite ragged, one would measure varying response by make and model. If this stuff is most important for some project, perhaps consider making a multiway speaker stacked, arranged so that each of the drivers has least amount of bandwidth and they are not affecting each others response too much, being further apart and not at immediate acoustic environment for each other.
All this is very academic, obviously all problems can't be fixed as some of them always exist due to scale of wavelengths on audible spectrum.
This is interesting stuff to think and imagine though, so I think it eventually leads for better loudspeakers, what ever that is for each person and application. For example some magnetostats or electrostats can sound quite fabulous, but if you need top and bottom octave for it it's back to crossovers and physical objects being each others way. Now, instead of agonizing over things one is forced to think what really is important, and then take a design path that aligns and try to make it well implemented. It's about knowing what needs to be done and then deal with the trade-offs, so that one can be happy with it in the end.
More detail: I think that thinking wavelength really helps spot issues and problems quite far down in the line and ties together all of this speakers and acoustics stuff providing fundamental understanding of the compromises involved with anything. Sound wavelength is the fundamental. As one now starts to design a speaker, it's quite easy to know the trade-offs for any design decision by knitting all the information one has learned, and wavelength works as glue for all of it, helps to connect various detailed knowledge together. Now, if one has project goals straight and is equipped with this knowledge it's easy to take the trade-offs (compromises) that align with the goals, just by reflecting the trade-offs against a priority list. For example, if one want's to design a speaker whose most important aspect is response within +-1db from 20Hz-20kHz, then it's obvious job for DSP, or very flat drivers are required, and the acoustic design, the structure, must be so that there is no secondary sound sources that would make interference with the direct sound making the response vary more than a dB or so. Basically one cannot make a shoe box speaker, because trade-off is edge diffraction issues, which does not align with the goal of +-1db response. Except, if the shoe boxes were per driver, fine for some particular narrow bandwidth. Got expensive drivers, expensive crossover and bad voltage sensitivity, lot's of huge boxes with big problems to build? well, yeah, but these weren't number one on the priority list so suck it, that's the deal, you took the trade-offs that were aligned with the priority list so it's a successful project nevertheless, no need to agonize over it.
Or just change the plan, perhaps use DSP instead of passives because decision to make it passive was not limitation from physics, but from the person wanting something. Some other project might have more important to be cheap and ease of built than +-1db response, then by no means put a fullrange driver on a shoebox, bam, cheap fast and fun but response isn't +-1db. Just suck it in and call it a day.
Both could have beautiful sound, depending what's your expectations, how you setup, whats the environment and so on 🙂 Problem is with the person itself, who agonizes the fact that system is never ideal and fails to understand the application. Figure out suitable goal and make a working plan that aligns with goal and then stick to it, know and accept the trade-offs. Just know what you are dealing with and make best out of it acknowledging there is physics behind that eventually dictates this stuff.
edit. Well, that went quite far into philosphy. What I was thinking initially was the topic title, and why frequency response is ragged? Take ideal driver and it's flat, at least to some observation angle. Now take a realistic driver, a physical object, and it is flat only some small bandwidth due to many reasons. Basically either the source is not ideal anymore, or there is some secondary sound from elsewhere that makes the response vary by interfereing with direct sound. Likely there is both at once. Now, just reduce the source varying by making a multiway, then reduce the interference by eliminating secondary sounds as well as possible, meaning diffraction backwave and reflections. That's about it, not too difficult?🙂
Yeah also something that is associated two transducers at very close proximity affecting each others response. Same stuff plagues single transducer as well: because cone area is static the excursion must vary with wavelength (frequency) to maintain flat SPL. Difference of how much more cone moves for long wavelength to short gets bigger the wider the bandwidth, double per every octave assuming ideal omni source.
So, reproducing a low frequency with same SPL as high frequency makes the high frequency modulate. If one wishes to maintain same excursion and SPL for both long and short wavelength to reduce modulation, the radiating area must change instead, and this cannot happen ideally but in steps as in coaxial, small driver in middle of bigger one. Idea with fullrange drivers is the same, somewhat controlled breakup so that highs are reproduced with small Sd, while rest of the cone not participating, except perhaps being bit out of control and cancelling out some to various directions. And since there is no full control of the breakup, the response ends up quite ragged, one would measure varying response by make and model. If this stuff is most important for some project, perhaps consider making a multiway speaker stacked, arranged so that each of the drivers has least amount of bandwidth and they are not affecting each others response too much, being further apart and not at immediate acoustic environment for each other.
All this is very academic, obviously all problems can't be fixed as some of them always exist due to scale of wavelengths on audible spectrum.
This is interesting stuff to think and imagine though, so I think it eventually leads for better loudspeakers, what ever that is for each person and application. For example some magnetostats or electrostats can sound quite fabulous, but if you need top and bottom octave for it it's back to crossovers and physical objects being each others way. Now, instead of agonizing over things one is forced to think what really is important, and then take a design path that aligns and try to make it well implemented. It's about knowing what needs to be done and then deal with the trade-offs, so that one can be happy with it in the end.
More detail: I think that thinking wavelength really helps spot issues and problems quite far down in the line and ties together all of this speakers and acoustics stuff providing fundamental understanding of the compromises involved with anything. Sound wavelength is the fundamental. As one now starts to design a speaker, it's quite easy to know the trade-offs for any design decision by knitting all the information one has learned, and wavelength works as glue for all of it, helps to connect various detailed knowledge together. Now, if one has project goals straight and is equipped with this knowledge it's easy to take the trade-offs (compromises) that align with the goals, just by reflecting the trade-offs against a priority list. For example, if one want's to design a speaker whose most important aspect is response within +-1db from 20Hz-20kHz, then it's obvious job for DSP, or very flat drivers are required, and the acoustic design, the structure, must be so that there is no secondary sound sources that would make interference with the direct sound making the response vary more than a dB or so. Basically one cannot make a shoe box speaker, because trade-off is edge diffraction issues, which does not align with the goal of +-1db response. Except, if the shoe boxes were per driver, fine for some particular narrow bandwidth. Got expensive drivers, expensive crossover and bad voltage sensitivity, lot's of huge boxes with big problems to build? well, yeah, but these weren't number one on the priority list so suck it, that's the deal, you took the trade-offs that were aligned with the priority list so it's a successful project nevertheless, no need to agonize over it.
Or just change the plan, perhaps use DSP instead of passives because decision to make it passive was not limitation from physics, but from the person wanting something. Some other project might have more important to be cheap and ease of built than +-1db response, then by no means put a fullrange driver on a shoebox, bam, cheap fast and fun but response isn't +-1db. Just suck it in and call it a day.
Both could have beautiful sound, depending what's your expectations, how you setup, whats the environment and so on 🙂 Problem is with the person itself, who agonizes the fact that system is never ideal and fails to understand the application. Figure out suitable goal and make a working plan that aligns with goal and then stick to it, know and accept the trade-offs. Just know what you are dealing with and make best out of it acknowledging there is physics behind that eventually dictates this stuff.
edit. Well, that went quite far into philosphy. What I was thinking initially was the topic title, and why frequency response is ragged? Take ideal driver and it's flat, at least to some observation angle. Now take a realistic driver, a physical object, and it is flat only some small bandwidth due to many reasons. Basically either the source is not ideal anymore, or there is some secondary sound from elsewhere that makes the response vary by interfereing with direct sound. Likely there is both at once. Now, just reduce the source varying by making a multiway, then reduce the interference by eliminating secondary sounds as well as possible, meaning diffraction backwave and reflections. That's about it, not too difficult?🙂
Last edited:
Thanks I'll play with it next chance I get, though I've already simulated 45deg by listening from nearly all angles "6DF" head position & direction, and also placement of tweeter nearly 6DF. But I'm clueless horn-wise, never used one except whizzers.
@tmuikku This thread started with a premise that coaxial speakers exist for a reason (or many). Ragged response (if true) does not invalidate that; but brings out cost-benefit analysis of solutions and compromises. Having a large number of discrete sound sources segregated by frequency band is not really part of a coaxial discussion.
What I have done, is a feasibility demonstration of near-field 15" coaxial-after-the-fact, drivers time-aligned and XO phase-aligned, by bouncing tweeter off of the convex dustcap. I so far (yesterday ~8 hours) judge the 15" near-field mono sound to be enveloping, immersive, and highly realistic (i.e. "to die for"). I haven't heard the Tannoy 15 or another large coaxial so can't make a comparison. Possibly I'll tire of the front-row live music experience (which was my aim), or wish to upgrade components (these four drivers cost me ~$100), but the important thing to me was simply making something new -- even if it was just an idea everyone had missed, or dismissed.
I would be totally unsurprised if no one has ever done this before; there are lots of effective and seemingly obvious ideas that haven't been thought of yet. I would like to see some measurements of it, especially with the woofer playing at high excursion simultaneously, to see how much it modulates the tweeter.
Apart from the FM modulation issue we would still have the other issue which the topology should actually solve: We would still have interference between the sound that is reflected off the cone and the one that directly bends around the tweeter.
Regards
Charles
Regards
Charles
Seas just released another coaxial with a ragged tweeter response. Probably not gonna be cheap.
Metamodal™ TPCD coaxial E0121-04/06 C16NX001/X
It looks worse than their larger excel driver E0060-08/06 C18EN002/A.
But maybe the developers have thought of something, or the tweeter sounds better than the measurements suggest.
Metamodal™ TPCD coaxial E0121-04/06 C16NX001/X
It looks worse than their larger excel driver E0060-08/06 C18EN002/A.
But maybe the developers have thought of something, or the tweeter sounds better than the measurements suggest.
Quick reply while at work. There are at least seven or eight issues, that each of us may rank quite differently.
o time-alignment of acoustic centers (this is the primary reason to bounce tweeter off of dustcap)
o phase-alignment at crossover region (now simpler, even series 1st-order)
o on-axis frequency response
o off-axis frequency response (helped by convex dustcap, good integration with cone and wide dispersion even above 10khz)
o effectively point-source (less destructive or chaotic interference from center-to-center distance and time/phase incoherence; especially necessary for near-field tweeter/midbass, less needed far-field)
o intermodulation (depends on a lot of things including size/driver/cab/damping/band/distance/SPL and how tweeter/deflector are implemented -- can be mitigated if need arises)
o doppler
o diffraction and unwanted reflection (size difference matters; can be mitigated with felt; tweeter off-axis wall-side is unobstructing toward wide listening area or height)
Yes, my specific application stacked the cards for transcient response fidelity: well-chosen large wide-band and tiny car tweeter; high bass roll-off from well-damped aperiodic; near-field listening hence low SPL, and so on. I think many forum members can readily replicate and expand the experiment -- please do.
o time-alignment of acoustic centers (this is the primary reason to bounce tweeter off of dustcap)
o phase-alignment at crossover region (now simpler, even series 1st-order)
o on-axis frequency response
o off-axis frequency response (helped by convex dustcap, good integration with cone and wide dispersion even above 10khz)
o effectively point-source (less destructive or chaotic interference from center-to-center distance and time/phase incoherence; especially necessary for near-field tweeter/midbass, less needed far-field)
o intermodulation (depends on a lot of things including size/driver/cab/damping/band/distance/SPL and how tweeter/deflector are implemented -- can be mitigated if need arises)
o doppler
o diffraction and unwanted reflection (size difference matters; can be mitigated with felt; tweeter off-axis wall-side is unobstructing toward wide listening area or height)
Yes, my specific application stacked the cards for transcient response fidelity: well-chosen large wide-band and tiny car tweeter; high bass roll-off from well-damped aperiodic; near-field listening hence low SPL, and so on. I think many forum members can readily replicate and expand the experiment -- please do.
Last edited:
Its surprising that turning the tweeter around and using the woofer cone as a reflector instead of as a waveguide works so well. Had you not disclosed it publicly, that - non-obviousness - would be the basis for a patent, with subclaims around optimizations based on the shapes of the various surfaces.
Its so surprising I find myself reasoning could one not do better by opening up that dust cap and putting a tweeter there or a shaped sound path back through the magnet to a compression driver. That has already been done with varying results, some quite good and most fairly expensive. The reverse bridge has the potential at least to be relatively inexpensive, especially for DIYers reaching into their stockpile of set-aside drivers.
But we haven't seen any measurements, just a subjective assessment. It might be that this approach is yet another that has HF raggedness yet still sounds quite good.
Its so surprising I find myself reasoning could one not do better by opening up that dust cap and putting a tweeter there or a shaped sound path back through the magnet to a compression driver. That has already been done with varying results, some quite good and most fairly expensive. The reverse bridge has the potential at least to be relatively inexpensive, especially for DIYers reaching into their stockpile of set-aside drivers.
But we haven't seen any measurements, just a subjective assessment. It might be that this approach is yet another that has HF raggedness yet still sounds quite good.
Here’s the thing…..both coax’s and fullrange drivers work very well in the near field…..better than most multi ways for the obvious…..the phase alignment combined with the acoustically small point source radiation is as ideal as sound reproduction gets. Here’s the other thing……near field listening for enjoyment other than work (mix/master/other) just plain ol isn’t much fun…….how a speaker dynamically interacts and acoustically couples to the room is where the icing from the cake lives……in the mid field….allowing the embedded stereo processing the opportunity to do its thing…..maximize spacial cues, soundstage depth, height and separation. While a larger format coax can still do this, it wouldn’t be my first choice…..not when you consider the options given the distance for better, individual and separate drivers to do what they can do.
It all comes down to the listener and how we enjoy music. As an event, seated comfortably in the same spot vertically and horizontally within the stereo triangle is how I do it…….its kick back and relax time…..let the body and the ear/brain mechanism disconnect from each other for a while so the ear/brain can fully enjoy the alone time.
Given the above, it should come as no surprise that the limited vertical response of a planar, amt or ribbon tweeter doesn’t matter at all to me, and these light, fast and well controlled driver elements do high frequency better than domes or CDs any day of the week when used properly.…..I’d challenge anyone to a blind test of a true ribbon from 3.5khz on up to not be preferred over any other high frequency transducer. Of course, YMMV and everyone’s acoustic space may not hold up for anything but near field listening……….to those folks I say electrostatic headphones……they’ll beat any near field speaker every single minute of a day.
It all comes down to the listener and how we enjoy music. As an event, seated comfortably in the same spot vertically and horizontally within the stereo triangle is how I do it…….its kick back and relax time…..let the body and the ear/brain mechanism disconnect from each other for a while so the ear/brain can fully enjoy the alone time.
Given the above, it should come as no surprise that the limited vertical response of a planar, amt or ribbon tweeter doesn’t matter at all to me, and these light, fast and well controlled driver elements do high frequency better than domes or CDs any day of the week when used properly.…..I’d challenge anyone to a blind test of a true ribbon from 3.5khz on up to not be preferred over any other high frequency transducer. Of course, YMMV and everyone’s acoustic space may not hold up for anything but near field listening……….to those folks I say electrostatic headphones……they’ll beat any near field speaker every single minute of a day.
Last edited:
Yeah, I don't get that Seas response or any of their offerings. They tout they are "ultra-high performance and exceptional sound quality."
But geez, all of the uglyness when they should be taking note of what some of the real top of the line concentric coaxs are employing. (KEF, Tannoy) Not sticking a fancy cone/dome on the same old antiquated design.
But geez, all of the uglyness when they should be taking note of what some of the real top of the line concentric coaxs are employing. (KEF, Tannoy) Not sticking a fancy cone/dome on the same old antiquated design.
Thanks. I've given up patenting and will probably let lapse the last of my search engine patents (first few were employer-paid, the rest cost >$100K each to prosecute and maintain). Examiners as a rule consider anything they can understand to be "obvious", despite both demonstrated utility and nonexistence in the marketplace.
Oooh 15" coherent near-field isn't like a common fullrange/coaxial -- enveloping and pore-penetratingly immersive, to exaggerate a bit. I have had way too many normal speakers and headphones and projects (despite starting diy only since Covid). In fact I just did a diyaudio tally -- 19 or 20 experiments over ten months in the fullrange photo gallery (last ten pages), and maybe another dozen scattered reports; started only a few threads including one on headphone stereo sound. Many of them are what I call CTY "Closer To You" for bringing the musicians closer -- front-row -- and forced by city-living to sit and listen closer. (I actually live at work, a small game workshop mavr3d.com Martial Arts VR3D.)
Last edited:
nd535, yeah it's interesting to see how wchang reflector system is implemented. Particularly interesting is directivity of the tweeter, how to get the treble toward the cone only. If it radiates directly to ear it's not coax system anymore as there is non-coincident sound source. If direct sound gets ear first before the reflection, then it's even worse in this regard than coax. This would always happen if tweeter is between woofer and listener and "leaks" (diffracts) any sound around. So directivity of the tweeter is what is complicated here, because to aim the treble toward the cone the tweeter (structure) needs to be big, waveguide or array or something, another 15" sized object perhaps, to control sound down to few kiloHertz. Perhaps it doesn't matter and makes better sound still by flooding the treble everywhere, less direct treble and more diffuse treble?
So, if it sounds good regardless then why not, it doesn't matter what the spec is. I just don't see how it could work any better than tweeter through / at the cone in technical sense. What makes coax response ragged is because there is two sound sources trying to squeeze into same physical space, and neither being ideal around crossover wavelength, because of the other being there. Same here, physical objects being there making response ragged, so it's just variation of the same theme / problem with similar issues. If it sounds better, then cool.
So, if it sounds good regardless then why not, it doesn't matter what the spec is. I just don't see how it could work any better than tweeter through / at the cone in technical sense. What makes coax response ragged is because there is two sound sources trying to squeeze into same physical space, and neither being ideal around crossover wavelength, because of the other being there. Same here, physical objects being there making response ragged, so it's just variation of the same theme / problem with similar issues. If it sounds better, then cool.
Last edited:
I wasn't suggesting you should have patented it. I have an equally cynical view of the whole thing. If you aren't in a position to maintain and defend a patent, you are better off not patenting it and instead keeping the technique secret. Provisional patent applications can be filed quite cheaply but if you can't follow up, you've just given the technology away. OTOH, better to put it in the public domain as you did than let a large company claim it with their own patent based on first to file.