I have come a long way building loudspeakers, first with traditional passive crossovers and then switched to most advanced custom digital crossovers using Analog Sharc DSP processors. Also I have always used upper market drivers like higher Scan Speak or Seas lines. Based on this experience and experimenting with different designs and DSP I believe the best concept to build an ultimate quality loudspeaker is using a coaxial driver or possibly a fullrange plus tweeter. The key justification is our ears are much more sensitive to timing dimension than to the frequency reponse. This is getting more and more recognition recently in hi-fi, see e.g. MQA technology.
Coaxials or fullrange are much better suited to time dimension fidelity but a question I am not sure about is are there drivers on the market that offer comparable overall speaker parameter quality as the best classical drivers ? Or is it more a niche segment where no serious R&D is happening ?
I am not interested in opinions how coaxials/fullranges "sound" since it is way too much dependent on the individual appplication and conditions. It is more about their technical parameters, technology advancement and overall technical potential. I am not too much worried about problems in frequency response either since I can correct them using DSP . Although unfortunately any DSP also distorts the timing since otherwise it could not work. So what I need is decent FR and best possible impulse response in possibly 150Hz - 15kHz range delivered from a point source over this range.
The ones that I have looked so far were Seas Prestige coaxials and Jordan Eikona fullrange.
What are your opinions or experiences with coaxials/fullranges ?
Coaxials or fullrange are much better suited to time dimension fidelity but a question I am not sure about is are there drivers on the market that offer comparable overall speaker parameter quality as the best classical drivers ? Or is it more a niche segment where no serious R&D is happening ?
I am not interested in opinions how coaxials/fullranges "sound" since it is way too much dependent on the individual appplication and conditions. It is more about their technical parameters, technology advancement and overall technical potential. I am not too much worried about problems in frequency response either since I can correct them using DSP . Although unfortunately any DSP also distorts the timing since otherwise it could not work. So what I need is decent FR and best possible impulse response in possibly 150Hz - 15kHz range delivered from a point source over this range.
The ones that I have looked so far were Seas Prestige coaxials and Jordan Eikona fullrange.
What are your opinions or experiences with coaxials/fullranges ?
Hi-Fi coaxial speakers of the past were often large and worked well, Tannoys for example. Today large speakers are out of fashion, smaller speakers are easier to make "fullrange" for a number of reasons. Coaxials are more expensive to manufacture than a fullrange. Of course there is still the timing and crossover issues to deal with with a coaxial speaker. I'm very happy with my Jordan Eikonas.
Last edited:
What exactly does DSP system distort in time domain of what passive system will not distort, in your opinion?
In fact, DSP can correct any time domain-related distortion which cannot be corrected in passive system in any way...
If you looked inside of Seas transducers, didn't you see Excel C18EN002 model? This is only one coaxial driver freely accessible for DIY which sounds right to my ears. In fact, it sounds exceptional. It behaves well not only in time domain, but also in terms of distortion and dispersion, which is exceptional. It is bloody expensive, unfortunately.
In fact, DSP can correct any time domain-related distortion which cannot be corrected in passive system in any way...
If you looked inside of Seas transducers, didn't you see Excel C18EN002 model? This is only one coaxial driver freely accessible for DIY which sounds right to my ears. In fact, it sounds exceptional. It behaves well not only in time domain, but also in terms of distortion and dispersion, which is exceptional. It is bloody expensive, unfortunately.
That is not my opinion that the minimum-phase DSP I just meant has any particular time domain distortions compared to analog.
Having said that however, DSP that is not realisable in analog like FIR filtering brings such distortions.
Yes, surely. And that is what I am doing and that's one of the key reasons why coax designs would be so much superior for me. This all-pass time domain correction is *very* sensitive to spatial differences.
What information is available on the time domain parameters of this driver ? - in the datasheet I see only the FR.
I am not sure if I got it right - do you mean that coaxial speaker help deal with timing issues or that they introduce some themselves ?
Last edited:
I've done a fair bit of work with coaxial speakers, and a fair bit of work with Unity horns. I currently have Unity horns in my car, coaxials in my living room (Kef), coaxials in my garage (Kef), D'Appolitos in my office, and conventional speakers in my TV room (Vandersteen.)
Here's my take on it:
Coaxials are nice, Unity horns are better. My Kefs sound fairly similar to a conventional two-way, but the sound is a little more consistent when you move around the room. The high frequencies are a little dull, even though they don't measure that way.
The Unity horns have a 'magic' that's pretty tough to beat. The articulation and imaging is without peer (at least among my set of speakers.)
Synergy Eggstravaganza
These coaxials came pretty close.
I think the project linked above, and Unity horns, work better for the following reason:
In a loudspeaker, there is an ideal combination of the following three parameters:
3) spacing
2) crossover frequency
1) crossover slope
This isn't a subjective opinion; we can objectively see that when we get the spacing, the crossover frequency and the crossover slope correct, the response 'snaps into place.'
So here's one of the reasons that Unities and coaxials with Radiation Boundary Integrators work so nice: They allow us to get the spacing EXACTLY correct. For instance, with a 5" coaxial, the midrange sound is radiating from a ring that's nearly four inches in diameter. In a Unity or an RBI, the midrange sound is radiating from a ring that's about 3/4" in diameter. So the spacing is more exact. The difference between the two might seem like splitting hairs, but it's not. At a crossover frequency of 2000hz, the sound waves are 17cm long at the crossover point. This means that a pathlength difference of just 4.25cm, or 1.67", will make a difference in the phase, amplitude and impulse response. So getting the spacing *exactly* right yields audible benefits.
Objectively, this improves the polar response. You can measure it, you can see it.
Subjectively, I find that it improves the articulation and imaging.
Here's my take on it:
Coaxials are nice, Unity horns are better. My Kefs sound fairly similar to a conventional two-way, but the sound is a little more consistent when you move around the room. The high frequencies are a little dull, even though they don't measure that way.
The Unity horns have a 'magic' that's pretty tough to beat. The articulation and imaging is without peer (at least among my set of speakers.)
Synergy Eggstravaganza
These coaxials came pretty close.
I think the project linked above, and Unity horns, work better for the following reason:
In a loudspeaker, there is an ideal combination of the following three parameters:
3) spacing
2) crossover frequency
1) crossover slope
This isn't a subjective opinion; we can objectively see that when we get the spacing, the crossover frequency and the crossover slope correct, the response 'snaps into place.'
So here's one of the reasons that Unities and coaxials with Radiation Boundary Integrators work so nice: They allow us to get the spacing EXACTLY correct. For instance, with a 5" coaxial, the midrange sound is radiating from a ring that's nearly four inches in diameter. In a Unity or an RBI, the midrange sound is radiating from a ring that's about 3/4" in diameter. So the spacing is more exact. The difference between the two might seem like splitting hairs, but it's not. At a crossover frequency of 2000hz, the sound waves are 17cm long at the crossover point. This means that a pathlength difference of just 4.25cm, or 1.67", will make a difference in the phase, amplitude and impulse response. So getting the spacing *exactly* right yields audible benefits.
Objectively, this improves the polar response. You can measure it, you can see it.
Subjectively, I find that it improves the articulation and imaging.
Last edited:
Do you know any more expensive commercial loudspeakers that use coaxials and if so what driver models ?
I also looked at Excel C18EN002 - it is quite an expensive driver and its frequency response looks pretty bad, especially tweeter's high irregularities. At first look it seems without a DSP equalization it should hardly be acceptable in upper hi-fi products. Also there is a pretty big difference in sensitivity between the two drivers. So I am wondering if DIY-ers and speaker manufacturers have used these drivers and what results (e.g. in terms of Freq resp.) they've achieved ?
I also looked at Excel C18EN002 - it is quite an expensive driver and its frequency response looks pretty bad, especially tweeter's high irregularities. At first look it seems without a DSP equalization it should hardly be acceptable in upper hi-fi products. Also there is a pretty big difference in sensitivity between the two drivers. So I am wondering if DIY-ers and speaker manufacturers have used these drivers and what results (e.g. in terms of Freq resp.) they've achieved ?
Pawel, as you have already noted, the Seas co-axial units are very decent although of somewhat compromised HF output which btw does not necessarily mean it could not be remedied sufficiently to not be objectionable. When there is absolutely no space available for a separate tweeter, co-axials are a treasure, very much better than wide band drivers.
Why not? Most manufacturers provide replacement drivers which are often competitively priced compared to DIY driver prices. Go to a hi-fi shop that sells a relevant complete speaker and ask. You will likely have to measure to get the driver parameters but many do this anyway.
The KEF coaxials in the Q and R series are perhaps the obvious examples of good value, high performance coaxials. A few years ago the demand for replacement drivers was sufficiently high in the US KEF stepped in to limit the numbers being sold. A small commercial speaker manufacturer was even using replacement KEF coaxial drivers in one their models.
Interesting that you say coaxials are a nice, but stil backup-ish solution. I'm very interested in the opinions and experiences on the forum as I think (and have quite a bit of theory and data) the really proper top hi-fi loudspeaker quality is actually not achievable without a possibly single point source over the range above 200Hz.
If you look to what measures, actually a bit ridiculous, many manunfacturers go to improve time alignment - e.g. Wilson Audio Alexx or JM Lab Utopia.
I also well remember as I had an opportunity to talk to Christopher Cabasse who presented his flagship Cabasse La Sphere (with DSP) at Audio Show in Poland. I'd read a bit about this design before and knew they used lots of Sharc processors and DSP. But clearly Sphere have more unusual design solutions so I asked him to pick what in his opinion is the most impactful and beneficial solution there. He said coincident drivers.
These are all top shelf, five-digit speakers and they still are very far from the timing target that, as I understand, coaxials are much closer to. So I'm a bit confused why coaxials are nevertheless too small a fraction of this market ? Maybe you can provide some hints ? I am about to start a no-compromise, pretty costly project with coaxials and I want to make sure I am not overlooking sth important here.
Last edited:
I should have been a bit more precise. I rather mean commercial products in open distribution, in particular I thought about Seas Excel Prestige line - have they been used in any commercial system ?
And I just looked at the SB coaxials line as hinted and I like (from the specs) these drivers more than Excels while they are quite inexpensive.
BTW, the thing that I miss a lot is impulse response of speakers, in particular midwoofers. It's a real pity hardly anyone publishes this data although it is much easier to actually measure than FR stretched to lower bass. Do you know of sites where impulse reponse data for drivers can be found in larger quantities ?
Last edited: