The following applies to a WE single unit full range system.
I absolutely agree with you. You should listen and use your own two good ears to evaluate any speakers that interest you. The proper taking of measurements, and their interpretaion, is skilled work and not all have the equipment or knowledge to pull it off. In the end, how it sounds to us is what matters.
If you are lucky enough to come across some WE 15a's at a hi fi show it will probably go like this: Even before you get to the room there will be a buzz: "Did you hear the Western Electrics? Like nothing I've ever heard before.". As you round the corner where they are, there will be a line outside the door. It will be worth the 20 minute wait and everyone will be eagerly anticipating a chance to hear them.
Once you finally squeeze into the room you will see the systems. Large, elegant, totally different than any of todays plastic junk. Before the demo an eloquent salesman stands up. Perhaps it is Joe Roberts. He will tell you with great passion and enthusiasm of the history of Western Electric, of their many great discoveries, their place in history, the pre-eminant engineers and scientists that worked there.
The system will be driven by some extremly expensive electronics. You will be told how it is a no holds barred system made with the most expensive components by the most dedicated ngineers and technicians. Vinyl of course, no digits here. Finally they drop the needle. It will sound like nothing you have heard before. It will be deficient in bass and treble but that won't seem to matter. The midrange will be glorious. The disk will be a female soprano or solo violin and it will have a focus of purpose and presence like you have never heard before. There will be a coherence and an almost palpable presence of the instrument in the room. Again, it will be unlike anything else at the show. All other speakers will sound "samey" and bland. Their added boom and tizz will be at odds with the music while this system's focus on the midrange will make the soul and melody of the music really pop.
After a pleasant half hour listening you will finally have to leave. Afterwards the memory of the sound will still haunt you. It was everything that you were told to expect. Nothing else at the show will be remotely close to the sound of the Western Electrics. What did they know that modern engineers are so ignorant of?
David S
Last edited:
Red Herring?..I was answering DavidL's direct muckraking question regarding Silbatone.
Exactly my point. Thanks. It is a sociocultural undertaking.
How much of what constituted scientific truth at any given point in history is socially constructed ? How much is agreed on by committee? How much rests in unspoken or even unreflected assumptions?
MOST of it.
Check your AES glossary for examples.
Not a non-sequitur. HF performace of TADs is often cited as PROGRESS. In fact it was done in this thread earlier. PS This phenomenon was mentioned in the JBL tech note on compression drivers linked by Speaker Dave immediately above my post, if you are looking for a smooth segue.
Yes, these are falsifiable statements and hence open to "science"--the taking empirical measurements part, anyway.
"Progress," on the other hand, gets slippery fast. I suggest spending more time in the social science department for that one.
FrankWW, you are welcome to come measure our WE systems. Seriously. If I had time, I'd do more of it. I posted what I have and links to what I can find.
But, seriously, I don't see the link between the graphs and the aesthetic experience. Occasionally I get a little hint of correlation but hardly enough that the graph is interchangeable with or able to stand in for the experience in any meaningful way.
Listening availability of 15A is a problem but this will not be solved by providing measurements here.
I really like the idea of making good stereo recordings of WE systems and I'll look into it. Been thinking about this for a while.
But in my mind, if not in that post, I do single certain actors out. So don't feel ignored, David L.
OTOH Speaker Dave is cool with me. I can see him processing info and trying to contribute from his background. He reminds me of the old hams who taught me electronics when I was a younger squirt, although I might be as old as Dave!
Speaker Dave strikes me as one of scientific bent, seeking to learn why.
My point with SD is that I think the clinging to contingent and faulty notions of progress is clouding his understanding, so I told him so and rib him about it. Didn't shake his faith yet, I fear.
Anyway, we are all literary characters here on the forums, so what does this have to do with personal attacks?
I'm not trying to "win friends" but friends will recognize me in these posts, because I am always like this.
I spent a lot of time in the audio professional world and many years in school studying this stuff. I'm sharing my findings which are based on lots and lots of field research and hands-on work in many aspects of audio, plus ten years of graduate level education in highly relevant fields of inquiry. Take it or leave it.
My point on "DIYforum science" is that much of the scientific thinking around here is based on rather flimsy models...got into that about 37 pages ago when DavidL started talking about predictability (control). This is pure rejected logical positivism. The Model T Ford jalopy of the intellectual raceway.
He attacked me first with this nonsense, trying to act educated, justified, outraged, and whatnot. Read a book, dude.
???
I'd offer, based on recent and close study of the texts, that Wittgenstein, who loved the cinema and probably heard a lot of London Western Electric systems, would say:
--We have no words for this.
--Even if we did have words, they would be words for talking about the 15A, which is a different game than listening to the 15A.
--You cant illuminate the inner workings of one game using the language of another game.
--Even if we did have words, there would be a gap between language and experience and we would be caught up in hermetic language games that do not reflect the world.
--Oh well, that's life.
That is a very defensible summary, in my reading. For more penetrating discussions on philosophy of science, write me direct.
There is really good stuff embedded in this endless thread and it does have to do with technical people trying to get to the bottom of things. Salut. Keep up the good work.
As I repeatedly put forward, we need smart tech guys to help figure out why an ancient relic like the 15A sounds so good. I'm not saying it is impossible to conceptualize part of it in technical language. If I had more measurements on hand, I would give them up. I think we need more.
What I would say is that it is indeed a challenge to our present systems of measurement, speaker design orientations, and even verbal description to fully come to terms with the 15A and other WE "Wide Range" implementations.
The sonic effect of big WE really is shockingly different, especially without prior exposure to large scale theater gear, but even then, it is different.
I think we, Silbatone and I, are immune to criticism on the empirical angle. What can be more empirically-oriented than dragging the 15As (and other WE gear) around the world at great expense so people can hear them? Come by next time and bring an RTA, laptop or whatever.
Just don't tell the people who were there and listened that they were wrong. I can empathize with the complaints as much as the praise from people who were there, because at least we share a frame of reference and experience.
But if you didn't hear it, there isn't much you can say about the sound. One can only imagine.
Go do your own research, you mindless objectivists. I did. Study the last couple hundred years work on objectivism while you are at it.
The part about "you all must be wrong because it is old" or, worse, the "I don't believe it, so you prove it or else you are wrong" schtick is lame.
(H'mm. I think they really mean empiricism not objectivism. but maybe not)
Yeah, I think these long threads can get tedious as hell. Guess I am just not a 24-7 forum type of guy. Since I'm here, I'm just speaking my mind, food for thought, if you can use it.
I originally came by because I am co-star in the movie that this interminable thread started out with. Where else am I a movie star but Multi Way subforum? Now is that a curse or a blessing?
Exactly my point. Thanks. It is a sociocultural undertaking.
How much of what constituted scientific truth at any given point in history is socially constructed ? How much is agreed on by committee? How much rests in unspoken or even unreflected assumptions?
MOST of it.
Check your AES glossary for examples.
Not a non-sequitur. HF performace of TADs is often cited as PROGRESS. In fact it was done in this thread earlier. PS This phenomenon was mentioned in the JBL tech note on compression drivers linked by Speaker Dave immediately above my post, if you are looking for a smooth segue.
Yes, these are falsifiable statements and hence open to "science"--the taking empirical measurements part, anyway.
"Progress," on the other hand, gets slippery fast. I suggest spending more time in the social science department for that one.
FrankWW, you are welcome to come measure our WE systems. Seriously. If I had time, I'd do more of it. I posted what I have and links to what I can find.
But, seriously, I don't see the link between the graphs and the aesthetic experience. Occasionally I get a little hint of correlation but hardly enough that the graph is interchangeable with or able to stand in for the experience in any meaningful way.
Listening availability of 15A is a problem but this will not be solved by providing measurements here.
I really like the idea of making good stereo recordings of WE systems and I'll look into it. Been thinking about this for a while.
But in my mind, if not in that post, I do single certain actors out. So don't feel ignored, David L.
OTOH Speaker Dave is cool with me. I can see him processing info and trying to contribute from his background. He reminds me of the old hams who taught me electronics when I was a younger squirt, although I might be as old as Dave!
Speaker Dave strikes me as one of scientific bent, seeking to learn why.
My point with SD is that I think the clinging to contingent and faulty notions of progress is clouding his understanding, so I told him so and rib him about it. Didn't shake his faith yet, I fear.
Anyway, we are all literary characters here on the forums, so what does this have to do with personal attacks?
I'm not trying to "win friends" but friends will recognize me in these posts, because I am always like this.
I spent a lot of time in the audio professional world and many years in school studying this stuff. I'm sharing my findings which are based on lots and lots of field research and hands-on work in many aspects of audio, plus ten years of graduate level education in highly relevant fields of inquiry. Take it or leave it.
My point on "DIYforum science" is that much of the scientific thinking around here is based on rather flimsy models...got into that about 37 pages ago when DavidL started talking about predictability (control). This is pure rejected logical positivism. The Model T Ford jalopy of the intellectual raceway.
He attacked me first with this nonsense, trying to act educated, justified, outraged, and whatnot. Read a book, dude.
???
I'd offer, based on recent and close study of the texts, that Wittgenstein, who loved the cinema and probably heard a lot of London Western Electric systems, would say:
--We have no words for this.
--Even if we did have words, they would be words for talking about the 15A, which is a different game than listening to the 15A.
--You cant illuminate the inner workings of one game using the language of another game.
--Even if we did have words, there would be a gap between language and experience and we would be caught up in hermetic language games that do not reflect the world.
--Oh well, that's life.
That is a very defensible summary, in my reading. For more penetrating discussions on philosophy of science, write me direct.
There is really good stuff embedded in this endless thread and it does have to do with technical people trying to get to the bottom of things. Salut. Keep up the good work.
As I repeatedly put forward, we need smart tech guys to help figure out why an ancient relic like the 15A sounds so good. I'm not saying it is impossible to conceptualize part of it in technical language. If I had more measurements on hand, I would give them up. I think we need more.
What I would say is that it is indeed a challenge to our present systems of measurement, speaker design orientations, and even verbal description to fully come to terms with the 15A and other WE "Wide Range" implementations.
The sonic effect of big WE really is shockingly different, especially without prior exposure to large scale theater gear, but even then, it is different.
I think we, Silbatone and I, are immune to criticism on the empirical angle. What can be more empirically-oriented than dragging the 15As (and other WE gear) around the world at great expense so people can hear them? Come by next time and bring an RTA, laptop or whatever.
Just don't tell the people who were there and listened that they were wrong. I can empathize with the complaints as much as the praise from people who were there, because at least we share a frame of reference and experience.
But if you didn't hear it, there isn't much you can say about the sound. One can only imagine.
Go do your own research, you mindless objectivists. I did. Study the last couple hundred years work on objectivism while you are at it.
The part about "you all must be wrong because it is old" or, worse, the "I don't believe it, so you prove it or else you are wrong" schtick is lame.
(H'mm. I think they really mean empiricism not objectivism. but maybe not)
Yeah, I think these long threads can get tedious as hell. Guess I am just not a 24-7 forum type of guy. Since I'm here, I'm just speaking my mind, food for thought, if you can use it.
I originally came by because I am co-star in the movie that this interminable thread started out with. Where else am I a movie star but Multi Way subforum? Now is that a curse or a blessing?
What makes you so sure they don't meet Dr. Toole's recommendations?
I say Quads do well on two aspects (the two that can't easily be fixed) and fall a bit short on one (that is easily fixed with extra equipment).
Let's do something notably lacking from the WE contingent, and study their measurements. I'll use Stereophile's measurements on the most recent iteration of Peter Walker's work*:
Toole Recommendation 1: smooth on-axis response. Quad? Check.
Toole Recommendation 2: smooth off-axis respons. Quad? Pretty good through the midrange, narrows above that. I'd call that a check.
Toole Recommendation 3: deep extended bass. Quad? Not so much. See nearfield FR, supra, and note that JA's mike technique exaggerates the upper bass, so there is some rise but not much and then a pretty steep fall.
That said, I could never live with Quads, even with big dipole subs under them. I just like full-scale presentations of Shostakovich, Mahler, and Radiohead too much. 🙂
*Quad hasn't done much new. The current speakers are basically the old speakers with more frame rigidity and a little more bass.
Last edited:
Comments are getting a bit too confrontational and personal, please take it down a couple of notches and stick to discussing ideas and not attacking each other.Point source is a matter of semantics.
The acoustic use of the word is a bit more.
A freely radiating point source in the acoustical sense is one who’s size is too small to impart directivity, it radiates omni directionally.
If presented with an impulsive signal, it only produces one arrival time at your ears as it only exists in one physical location (a good thing), the impulse you measure is not dispersed /extended to multiple sources or an extended source, only has the shape the minimum phase response dictates.
A rule of thumb is that when the source is about a quarter wavelength or less in size, then it is radiating like this.
A small full range driver like some of the Fostex on a large flat baffle radiates a hemisphere pattern* with the baffle proving an acoustical mirror image. This can be thought of as a 180 degree horn aka half space down to some frequency governed by the baffle size.
*up to the frequency where it’s radiator size causes makes the radiation shift to a more complex field shape.
If you have never heard a set up like this, I highly urge you to make some sawdust and knock some together.
They have limitations BUT they really demonstrate how much difference “how it radiates” and “where it is in time” makes. As you might guess, I am a diy’r so I encourage people to make some speakers if you can.
With the wavelength being about 5/8 inch at 20KHz, making a point source in the acoustic sense becomes more problematic.
As a result even with the small full range drivers, the best demonstrations will be with voices etc that don’t have a lot of hf content and are what your hearing is most attuned to.
If you were trying to drive a conical horn like we use at work, then the maximum source size is set by the same thing that governs the optic limitation called the “airy disk” (the wall of science exists everywhere).
Thankfully some compression drivers have an internal structure which continues in a taper down to a fairly small dimension and so up high where the expansion starts (governing the highest frequencies) is small enough to radiate a segment of a sphere within the driver before reaching the horn.
Funny too if you examine where the famous impedance transformation takes place, one finds that “up high”, that happens way inside, it’s all over well before reaching even a one inch exit.
Anyway, the point is that with that acoustic relationship based on the wavelength at whatever frequency you’re concerned with, one finds what reminded me of a stack of Russian dolls. All identical looking except for the volume they occupy and so it is with a lot of things in acoustics.
For example to make a given “thing” like a woofer work an octave lower with the same efficiency requires the volume it occupies be cubed.
Anyway, a person speaking is not a point source, your mouth and head are large enough to have changing directivity, a guitar anywhere above the lowest notes is not a point source because it is large AND is vibrating in multiple modes in many locations. Most loudspeakers are point sources down low but stop when multiple driver combine outside a quarter wavelength spacing or when the source is large enough or radiating from more than one acoustic location (like a woofer used up high etc).
Sound travels at the speed of sound haha so one can also deal with it’s propagation based on that speed and awareness of how large the wavelength is, do things like are done with optics.
The acoustic lens is an example, take the old potato masher, where a spherical wave front with a large radius of curvature is progressively delayed as you move away from the center.
That small delay bends the wave front as it has to travel a larger distance at the edges than the center. Viola, a wide angle source out of a narrow angle source. You have adjusted the “focal length” with a lens.
I can’t explain how publicly yet but it is possible to approach the problem of adding high frequency sources this way (a lens system) as well.
That combiner to do this was I think the hardest thing I have had to design in my life so far.
Anyway, the key word might be acoustic size.
For a measurement microphone to not have appreciable directivity at 20KHz it has to be pretty small, like 4-5mm, on the other hand, at 20Hz, the mic could be 12-13 feet across.
Best,
Tom Danley
Danley Sound Labs
The acoustic use of the word is a bit more.
A freely radiating point source in the acoustical sense is one who’s size is too small to impart directivity, it radiates omni directionally.
If presented with an impulsive signal, it only produces one arrival time at your ears as it only exists in one physical location (a good thing), the impulse you measure is not dispersed /extended to multiple sources or an extended source, only has the shape the minimum phase response dictates.
A rule of thumb is that when the source is about a quarter wavelength or less in size, then it is radiating like this.
A small full range driver like some of the Fostex on a large flat baffle radiates a hemisphere pattern* with the baffle proving an acoustical mirror image. This can be thought of as a 180 degree horn aka half space down to some frequency governed by the baffle size.
*up to the frequency where it’s radiator size causes makes the radiation shift to a more complex field shape.
If you have never heard a set up like this, I highly urge you to make some sawdust and knock some together.
They have limitations BUT they really demonstrate how much difference “how it radiates” and “where it is in time” makes. As you might guess, I am a diy’r so I encourage people to make some speakers if you can.
With the wavelength being about 5/8 inch at 20KHz, making a point source in the acoustic sense becomes more problematic.
As a result even with the small full range drivers, the best demonstrations will be with voices etc that don’t have a lot of hf content and are what your hearing is most attuned to.
If you were trying to drive a conical horn like we use at work, then the maximum source size is set by the same thing that governs the optic limitation called the “airy disk” (the wall of science exists everywhere).
Thankfully some compression drivers have an internal structure which continues in a taper down to a fairly small dimension and so up high where the expansion starts (governing the highest frequencies) is small enough to radiate a segment of a sphere within the driver before reaching the horn.
Funny too if you examine where the famous impedance transformation takes place, one finds that “up high”, that happens way inside, it’s all over well before reaching even a one inch exit.
Anyway, the point is that with that acoustic relationship based on the wavelength at whatever frequency you’re concerned with, one finds what reminded me of a stack of Russian dolls. All identical looking except for the volume they occupy and so it is with a lot of things in acoustics.
For example to make a given “thing” like a woofer work an octave lower with the same efficiency requires the volume it occupies be cubed.
Anyway, a person speaking is not a point source, your mouth and head are large enough to have changing directivity, a guitar anywhere above the lowest notes is not a point source because it is large AND is vibrating in multiple modes in many locations. Most loudspeakers are point sources down low but stop when multiple driver combine outside a quarter wavelength spacing or when the source is large enough or radiating from more than one acoustic location (like a woofer used up high etc).
Sound travels at the speed of sound haha so one can also deal with it’s propagation based on that speed and awareness of how large the wavelength is, do things like are done with optics.
The acoustic lens is an example, take the old potato masher, where a spherical wave front with a large radius of curvature is progressively delayed as you move away from the center.
That small delay bends the wave front as it has to travel a larger distance at the edges than the center. Viola, a wide angle source out of a narrow angle source. You have adjusted the “focal length” with a lens.
I can’t explain how publicly yet but it is possible to approach the problem of adding high frequency sources this way (a lens system) as well.
That combiner to do this was I think the hardest thing I have had to design in my life so far.
Anyway, the key word might be acoustic size.
For a measurement microphone to not have appreciable directivity at 20KHz it has to be pretty small, like 4-5mm, on the other hand, at 20Hz, the mic could be 12-13 feet across.
Best,
Tom Danley
Danley Sound Labs
I got you on the first one and the second one...third one, dunno there...lol.
First time I have seen full scale presentation of Radiohead mentioned in print.
Though I don't care for them, I can see why you say that.
First time I have seen full scale presentation of Radiohead mentioned in print.
Though I don't care for them, I can see why you say that.
Sorry David Exactly NOT like the demos we did in Paris. Vinyl and CD, integrated EL34 P-P amp of good, but not stratospheric quality. No waiting in line, no lack of bass or highs, everything from Electronica to Jazz to Chamber, Pop, Rock, Symphonic and Choral music was played. Nailed them all. You can't just make this stuff up if you haven't been there, that's not fair and is wildly inaccurate.FWIW, I've been in audio show rooms with a huge buzz and big crowd and been sorely disappointed - that includes horns, open baffles, single drivers, etc. Some of us don't judge with our expectations, we judge with our ears.
I know where you can hear a stereo pair 😉 They actually throw a nice soundstage, and sound very sweet, even if they are a bit "AM Radio" in SQ.
Hello Joe
The JBL technote is from 1990. The way they are doing Be today is quite different. The only guys doing the VD is Tad. As far as I know the rest is Be foil which is much more robust. As far as secondary resonances the Aluminum diaphragms will always have them at a lower frequency and more of them compared to Be. There is no getting around it. All compression drivers are going to have issues with this and the 555 is no exception.
Why you would chose that as a issue to claim superiority over Be drivers I just don't get. Read the attached link
Rob🙂
Truextent White-paper for Large Format Diaphragms
Last edited:
So prejudiced point of view from one objectivist?
At least this year in Munich wasn't like that at all. There was no line in front of the room and there was no eloquent salesman. In sutarday morning I walked in half empty room and took center seat in the first row of chairs 🙂
If you've ever been to Munich High-end you would know that the WE system look couldn't be elegant in any way especially compared to at least 20 other expensive "larger than life" systems. There was all kind of music played 70's rock, jazz (trio, big band) and some opera and symphonic music. To my ears it sounded wonderful despite the fact that the day before I was explaining to the couple of my friends that such an old loudspeakers really didn't have any chance to sound at least any good - interesting perhaps, but not good in any way. 😀
You really should hear them before getting into discussion about how they sound (like I did), and I'm sure you'll be surprised.
Good link, Rob
The Berylium curves are very compelling. Top Octave response is clearly superior.
It has always bothered me to see three way horn designs with high crossover points as you won't have time allignment and the crossover at 5 or 7 kHz (as is typical) will always be messy. This is true of your avatar (the Stereophile measurements clearly showed it) and would be true of the WE systems using a 7k crossover.
The JBL 2441 and later compression drivers were pretty good in the top Octave and those Be curves are essentially perfect, giving no reason, with a good CD horn, to add a tweeter and an unnecessary crossover point.
Progress.
David
I'm not sure that this is the correct definition of a point source ? This implies that only an omni directional radiator can be a true point source.
Later in the post you give an example of a full range driver as a point source, but even a small full range driver is anything but omni directional at high frequencies, yet I would consider it to be a point source.
This I think is a key attribute of a point source, not only should the sound originate from one location, (a virtual point in the middle of the driver) but there should not be any dispersed time delayed radiation from physically separated locations. The number one culprit here of course is baffle edge diffraction.
I would consider a wide dispersion tweeter mounted on a flat baffle not to be a point source, despite the small size of the driver itself, because the entire baffle outline is a radiating source of complex time delayed out of phase radiation which will introduce anomalies in both the time domain on any measurement axis, as well as directivity anomalies due to the large "size" of the composite radiating source. (The driver plus the baffle outline) Most people only consider the effects on the frequency response, typically only on one or a few axes, but that does not paint a picture of how damaging the diffraction is, nor do they consider that they are not getting a point source.
On the other hand you could mount a waveguide driver on the exact same baffle and provided that mouth termination etc was good, you could come very close to a point source because there would be little or no baffle illumination, and therefore no signal to diffract from the baffle edge. In practice if you can cut the radiation at 90 degrees down by about 15-20dB it seems to enough to approximate this.
By the way although it doesn't seem intuitive, it is the case that baffle diffraction is still minimum phase. (not that that's any excuse to allow it)
There is something appealing about a good full range driver, especially in the vocal range and more specifically the imaging. There is something that gets lost in multiway systems unless you're really careful with crossover design, driver choice, diffraction etc. Hard to describe what it is, but it's a sort of coherency and rightness about the sound - within the bandwidth and dynamic limits of the driver of course.
My current system is based on a classic 8" full range driver, but I don't normally use it full range, preferring to use it with a ribbon tweeter, and in the past also with a woofer for the bottom end. (not possible in my current cramped abode unfortunately) I can easily switch it to full range though and being able to do a direct comparison between full range and 2/3 way with the same cabinet and the same driver producing the midrange in all cases is extremely informative when it comes to figuring out what can frequently go wrong in multiway systems.
As i found, it's very easy to destroy that imaging coherency that you get from a single driver through poor choice of crossover frequencies, crossover design, driver time alignment, or diffraction effects, all of which contribute to losing the imaging coherency of a single driver. Typical design techniques of multiway box speakers (non time aligned drivers, wide dispersion tweeters that introduce a lot of baffle diffraction, crossover points in the middle of the midrange, etc) can often kill imagining coherency, but if you aren't able to do a direct comparison to the driver operating full range (due to the driver not being a full range driver) then you might not realise what is missing.
After a lot of playing around with full range driver versus full range driver plus tweeter I found that it is possible to have nearly the same imaging coherency of a full range driver with the tweeter, but there are a number of prerequisites;
* One driver must cover the entire midrange plus a bit on either side. In my case I settled on 250Hz to 4Khz as a 3 way or bass up to 4Khz in a 2 way configuration. Typical 1.5 - 2.5Khz crossover points I don't feel image as coherently regardless of driver ability at those frequencies, as you're splitting the midrange into two very dissimilar physically displaced drivers with very different baffle diffraction signatures, and our ears are not fooled into thinking its one sound source no matter how good the amplitude and phase tracking between the drivers might be.
* Elimination of sudden shifts in baffle diffraction signature at the crossover point seems to be very important. If you go from a woofer that is starting to beam below the crossover point (producing little baffle diffraction) to a wide dispersion tweeter producing massive amounts of baffle edge diffraction you don't have a smooth handover between the two drivers, you're going from a point source just below the crossover frequency to decidedly non point source above the crossover frequency.
Many people notice that designs like 8" woofer to 1" dome tweeter don't sound very good unless the crossover frequency is very low (and I would argue the low crossover frequency is a bad choice as well for other reasons above) and the finger of blame is usually pointed at either directivity mismatch at the crossover frequency or power response holes due to beaming of the large driver.
Using a waveguide tweeter with a controlled directivity makes this sort of design work much better, but why ? Is it really the power response hole that's the problem ? Is moderate directivity mismatch really that important ? My hunch is its not the power response hole or directivity mismatch that's the problem, it's the huge and abrupt shift in baffle diffraction signature between the two drivers, one illuminating the baffle edge and one not that is the major problem.
In other words if you could wave a wand and eliminate the baffle edge diffraction from both drivers then close directivity matching at crossover becomes far less important.
How to achieve this in practice ? For the woofer/mid you want a driver size that will just start to beam about an octave below the crossover frequency, so that there is little if any baffle diffraction from that driver through the overlap region.
For the tweeter you want a waveguide with good directivity control including through the whole crossover region, so that there is little if any baffle diffraction produced by the tweeter at any frequency - eg it is acting as a point source.
Do that and you now have one point source transitioning smoothly to another vertically displaced point source, no shift in baffle diffraction signature, because there isn't any diffraction. Will there be some inevitable narrowing in the polar response and a hole in the power response ? Yes, some. Does it matter ? No it doesn't seem to as far as I can tell, especially with a high crossover point like 4Khz, which is what makes me think it's the effects on diffraction that directivity mismatch produce that is important and not the actual effects on power response etc in the room.
Remember also that a full range driver by itself is very directional at high frequencies and many also do not have monotonically narrowing directivity with frequency, eg those with whizzers and/or radiating dust caps will narrow then widen a bit then narrow again sometimes twice, and often have severe spurious lobes at the high end and yet they can still have brilliant imaging, so constant or monotonically narrowing directivity does not seem to be required for imaging or coherency.
* Time alignment of drivers. A full range driver is inherently time aligned, even a dual cone driver which is technically an acoustic/mechanical 2 way has close time alignment, by measuring excess group delay I've determined that it's typically on the order of a few millimetres between low and high frequencies, even on a large driver.
For example the acoustic centre of the high frequency whizzer output on my 8" coral drivers is about 7mm forward of the low frequency acoustic centre, I also measured some FE207E and found their high frequency acoustic centre is about 14mm behind the low frequency acoustic centre. On a small single cone FR driver it's probably much less. Either way these are all figures that are much less than driver misalignment of most flat baffle multiway systems.
Conventional wisdom is that the typically found large acoustic centre offsets in flat baffle systems don't matter provided that the phase curves are bent to track properly in phase through the overlap region, but then most conventionally engineered multiway speakers don't have the same coherency as a full range driver either...
Research has been done into the audibility of peaks in group delay, such as those occurring at a narrow frequency range around the crossover point, but has any research been done into the audibility of a multi-octave shelf shaped step in the group delay ? It seems to me that group delay occurring over only a narrow range of frequencies is more likely to be unnoticed that that occurring over a wide frequency range, so the thresholds could be different, just like the thresholds of audibility for peaks in the frequency response reduce as the bandwidth increases. It would be interesting to know.
In any case, if we are trying to capture the coherency of a full range driver in a multiway system, time alignment does seem important. I don't think it has to be perfect, but keeping it under about 20mm or 90 degrees phase shift at the crossover point, whichever is smaller seems to be a reasonable target. With a crossover frequency of 4Khz and an offset of less that 20mm I've been able to achieve very close to the coherency of the FR driver on its own. It's deliberately offset this much because I'm using an in phase 3rd order network, when I get time one day I plan to try a 4th order network with drivers completely aligned to see if there is any further improvement.
* Phase tracking through the crossover region. This is an area where conventional wisdom has got it right. Even if all of the above is right if the phase tracking is poor single driver like coherency is instantly destroyed. Not only does it need to track, it needs to track in phase. Quadrature summing just doesn't work and sounds weird. Summed response also needs to be flat with no weird peaking off axis.
Most fullrange plus "super tweeter" approaches like full range unfiltered and tweeter running through a small single cap fail partly because the phase tracking through the rather large overlap region is horrendous. It doesn't matter what you do there are usually points of phase cancellation and this sounds bad. FR fans then declare both tweeters and crossovers evil, without understanding why it didn't work... 😀
I've tried these approaches before and they just don't work. They add sparkle at the top but the fullrange coherence is spoiled. I can actually get much closer to the coherence of a single driver with the treble improvement of a tweeter by simply using a proper phase tracking crossover and following all the points I've mentioned above. I feel like I'm getting about 90% of the single driver coherence with none of the drawbacks at the top end, with an overall imaging result that is significantly better than the single driver.
I would advise anyone wanting to investigate optimising coherence and imaging of multiway systems to use a good full range driver as your midrange/midbass driver, it really is quite enlightening being able to do direct A/B comparisons between fullrange and multiway with the same driver on the same baffle, instead of comparing two completely dissimilar systems. The fullrange mode provides a good control reference for what coherency should sound like, allowing you to see how much proposed tweeters, baffles, driver alignment, crossovers etc degrade the coherency, and how close you are getting as you tackle each issue.
Last edited:
Some questions that come to mind:
Is time alignment important at high frequencies?
Have there been any studies done on audibility of phase issues with high frequency crossovers?
Do the pro's outweigh the cons?
Since there are many listeners that like the sound of this speaker, I think it would be more constructive to start asking WHY this speaker sounds as good as it does, rather than saying it can't possibly sound good. It might even rescue this thread, which seems to be oscillating between informative and train-wreck...
B&W Nautilus would qualify...
So would the Linkwitz Pluto albeit with low crossover. However diffraction would be minimal at the crossover point since there is no real "baffle"
LS3/5A perhaps? The felt block around the tweeter greatly reduces diffraction.
All of the above are regarded as imaging champs, so maybe you are on to something? 🙂
Last edited:
I am referring to the poor frequency response and would consider it a response error rather than necessarily a phase issue. You will always have a Octave or so of comb filtering and this is an audible error (subjectivism!) as I can tell you from experience. In any system where you can displace the tweeter you will hear the comb filter effect as you slide its position fore and aft of the in-phase position.
In any design using horns, horn length is an issue to contend with. Crossing a short tweeter to a long mid horn can't be easily done. Even if you tried for nominal time alignment (drive units at the same depth) then the vertical spread has to be considerable and any listener position shift will throw off the allignment.
Shearer and others heard time allignment problems with the WE horn used as a 2-way but that was at the lower crossover point where its 7 ft length made blending to a woofer difficult.
The point is that there is no need for a crossover in the top two Octaves with modern compression drivers and good CD horns.
David S
The Blauert and Laws test frequently quoted was for the threshold of detection for a 2nd order all-pass so it was for a peak in group delay. A crude test, but the Shearer remeberance referred to several time in this thread had him stating that time allignment of better than 1 ft. made the bluring he heard on impulses go away. That would be a shelf type delay error (likely there was a response error involved too).
As I was saying with crossovers between mid horn and tweeter.
David S
That's what I thought. So what they were studying would give a good indication of the audibility of crossover induced group delay, but can't necessarily be extrapolated to a multi-octave group delay shelf caused by driver time alignment errors. So there is more research to be done here.
The crucial missing piece of information here is what was the crossover frequency, (sorry if I missed it as I haven't followed the thread closely) without knowing where the spectrum was divided we can't know the significance of the 1 foot/millisecond figure.
Everything I said about driver alignment and group delay was assuming a midrange to treble crossover, (I should have stated that) it could well be that a shelf in group delay is a lot less noticeable at a lower frequency, for example 300Hz or lower, as we become progressively less sensitive to group delay below that...(just look how much can be tolerated in the bass compared to 2Khz)
I'd be very surprised if it was possible to get single driver like coherency with a tweeter that is 1 foot behind the midrange driver unless you used actual digital delay. I don't think normal asymmetric slope phase bending crossover would cut it, even if you could manage phase tracking.
The way I would approach trying to quantify the audibility of a group delay shelf is digitally, as follows:
Implement an idealised digital 2nd or 4th order L/R crossover with a selectable crossover frequency, then sum the two sections and listen to the result on high quality headphones. This would represent a time aligned version of a L/R design, and should give results consistent with Blauert and Laws.
Now add digital delay to one virtual driver, change the filter slopes to get phase tracking through the overlap region exactly as you would in a real speaker design, essentially you are duplicating the acoustic centre offset and crossover "tricks" that you would use on the real speaker. Then sum the two sections and listen on headphones.
Because there might be some response ripple in the asymmetric non-aligned system, it might also be necessary to apply some minimum phase convolution to the summed response to restore a completely flat amplitude response so that we're not hearing small differences in frequency response, only the group delay shelf.
If I had the software/programming knowledge to do the above I would try it. Of course this does not account for other differences caused by physical driver offset such as errors in inverse square law tracking with distance, and changes to the polar response and off axis summing.
Another approach instead of headphones would be to do the same with an actual DSP based active crossover design with time aligned drivers - deliberately misalign them with a digital delay, then fix the amplitude response and phase tracking with asymmetric slopes etc, and listen to the result. The drawback here is you're limited to testing only a small range of crossover frequencies that the drivers can handle, but at least you include some of the room and spacial effects of driver delay in the comparison.
Last edited:
Simon, did you see janneman's posts a few months ago, where he had six randomized files from Hawksford, 3 of which simulated the allpass delay of an LR filter? The idea was to do blind testing via sorting. Two of us who bothered to try this (several of the usual suspects refused to participate since they "know" that DBTs are invalid) were able to successfully sort the files.
This certainly sharpened MY awareness of what I thought were inaudible phase shifts!😀
This certainly sharpened MY awareness of what I thought were inaudible phase shifts!😀
Hi Simon,
I believe that the most important factor for good imaging is the similarity of both speakers (or multiple speakers in a multichannel system). It doesn't matter much if there are large magnitude or time errors as long as they are the same in both speakers.
Unfortunately we're only halfware there by making both speakers behave exactly the same. Room reflections can be perceived as part of the direct sound. Now if distortion introduced by the room isn't the same for both speakers, imaging will suffer again.
Tackle both issues and pinpoint imaging is your reward.