Geddes on Waveguides

Regarding the debate on "CD or not to CD", I've been using this approach for over three decades now. I stumbled across it somewhat by accident, wanting to build a speaker like the Klipsch cornerhorn but incorporating constant directivity. My first cornerhorns made this way were built in 1980, and everyone that listened to them found them to be very pleasant. I still use this approach today.

This is of course, purely subjective and its discovery was empirical, at least on my part. I wasn't expecting to find improved imaging, I was hoping only for improved coverage. My idea was that the room walls would form a giant CD horn that would work at the lower end where smaller horns lost directivity control. Of course, room modes take over below a certain point but that's a different subject. The subjective impression was so good, I soon started making freestanding speakers with similar characteristics, woofers and tweeters with directivity matched at the crossover point, just like what many describe here in this thread. It's been my favorite approach for several decades.

The point is, I think that whether or not you prefer a horizontal beamwidth of 60°, 90° or 120°, the idea of constant directivity and toe-in making the axes cross just in front of the listener is becoming popular with many manufacturers and owners alike. Whatever mechanisms make this work is probably something we could all debate, but I think the fact that so many have begun to embrace it says volumes about how good it sounds. Most of the participants on this thread embrace constant directivity in some form, but we often disagree on the methods used to achieve it. One thing we all agree on though, is the importance of CD for both coverage and for imaging.

To me, the biggest issue of debate isn't even related to the horizontals, it's the verticals. We all agree about CD in the horizontal, and like catapult said, the differences we're talking about there are minutia. But verticals are a different matter entirely. Some place a good deal of emphasis on it, others none at all. That's my biggest point of contention with Geddes, to disregard the verticals is to disregard the whole point of CD, in my opinion. I should not be able to move up and down a few inches and hear a difference. When making polar charts, one should include vertical off-axis measurements, not only horizontals. Failure to do so is needlessly incomplete and is incorrect, showing a very misleading picture of polar response.

I'm not sure I even want to qualify that statement with "in my opinion". To not show the verticals is to not show the whole picture, and that's all there is to it. I suppose you can argue they aren't important, but I can't see that as anything but standing on thin ice. When I can sit down and hear a different sound than when I stand up, that's a problem. When the nulls formed are in the frequency range easiest to detect, right in the peak of the Fletcher-Munson curve, that's something to avoid. I think it's important to place those nulls outside the listening area as far as possible. The floor and ceiling are the closest room boundaries, so reflections from them are going to influence the reverberent field. I just think the verticals are important to consider when designing a loudspeaker, especially one promoted as having superior directional characteristics.
 
In Toole's book there are references to experiments conducted in anechoic environments with adjustable and moveable secondary sources standing in for wall reflections. An experimental setup like that should be able to remove catapult's concerns over directivity differences being swamped by other differences between two speakers since you're only testing one speaker, varying the (simulated) reflections. As Earl has mentioned the test focused on listeners preferences for "spaciousness" not "imaging".

I'm unable to provide bentoronto with the links to those studies that he so strongly believes don't exist because I lent the book to a friend months ago but the references are in there. The book is a pretty wide summary of research done on the subject of correlating listener preferences to hard measurements, not an ad for Toole's work after "he became a figment of Cosmic Audio Corp." as bentoronto put it. He had retired from Harman before writing the book.

Bentoronto, you came here asking for evidence, we told you where to look. How that is offensive is beyond me. Look at what work has been done or don't, it's doesn't matter to me.
 
Last edited:
Regarding the debate on "CD or not to CD", I've been using this approach for over three decades now. I stumbled across it somewhat by accident, wanting to build a speaker like the Klipsch cornerhorn but incorporating constant directivity. My first cornerhorns made this way were built in 1980, and everyone that listened to them found them to be very pleasant. I still use this approach today.

This is of course, purely subjective and its discovery was empirical, at least on my part. I wasn't expecting to find improved imaging, I was hoping only for improved coverage. My idea was that the room walls would form a giant CD horn that would work at the lower end where smaller horns lost directivity control. Of course, room modes take over below a certain point but that's a different subject. The subjective impression was so good, I soon started making freestanding speakers with similar characteristics, woofers and tweeters with directivity matched at the crossover point, just like what many describe here in this thread. It's been my favorite approach for several decades.

The point is, I think that whether or not you prefer a horizontal beamwidth of 60°, 90° or 120°, the idea of constant directivity and toe-in making the axes cross just in front of the listener is becoming popular with many manufacturers and owners alike. Whatever mechanisms make this work is probably something we could all debate, but I think the fact that so many have begun to embrace it says volumes about how good it sounds. Most of the participants on this thread embrace constant directivity in some form, but we often disagree on the methods used to achieve it. One thing we all agree on though, is the importance of CD for both coverage and for imaging.

To me, the biggest issue of debate isn't even related to the horizontals, it's the verticals. We all agree about CD in the horizontal, and like catapult said, the differences we're talking about there are minutia. But verticals are a different matter entirely. Some place a good deal of emphasis on it, others none at all. That's my biggest point of contention with Geddes, to disregard the verticals is to disregard the whole point of CD, in my opinion. I should not be able to move up and down a few inches and hear a difference. When making polar charts, one should include vertical off-axis measurements, not only horizontals. Failure to do so is needlessly incomplete and is incorrect, showing a very misleading picture of polar response.

I'm not sure I even want to qualify that statement with "in my opinion". To not show the verticals is to not show the whole picture, and that's all there is to it. I suppose you can argue they aren't important, but I can't see that as anything but standing on thin ice. When I can sit down and hear a different sound than when I stand up, that's a problem. When the nulls formed are in the frequency range easiest to detect, right in the peak of the Fletcher-Munson curve, that's something to avoid. I think it's important to place those nulls outside the listening area as far as possible. The floor and ceiling are the closest room boundaries, so reflections from them are going to influence the reverberent field. I just think the verticals are important to consider when designing a loudspeaker, especially one promoted as having superior directional characteristics.

Of course, the directivity cutoff of a horn is correlated to the pathlength, which is significantly shorter in the vertical dimension with horns like yours, so around the crossover frequency there's less vertical directivity applied by the horn than in the horizontal dimension.

Which is not to say that negates the approach, but it certainly adds a dimension to the analysis of round vs. asymmetric horns. Vertical polar aberrations are largely a function of cutoff slope, since it's the 2 source interaction that's the major issue. Might be a reason to continue pushing for higher slopes in horn system filters.
 
Of course, the directivity cutoff of a horn is correlated to the pathlength, which is significantly shorter in the vertical dimension with horns like yours, so around the crossover frequency there's less vertical directivity applied by the horn than in the horizontal dimension.

Which is not to say that negates the approach, but it certainly adds a dimension to the analysis of round vs. asymmetric horns. Vertical polar aberrations are largely a function of cutoff slope, since it's the 2 source interaction that's the major issue. Might be a reason to continue pushing for higher slopes in horn system filters.

I see it as a balance of competing priorities. Horns that are larger in the vertical dimension have the potential for greater vertical directivity control, but at the expense of a narrower forward lobe. Horns that are not as tall can't control directivity down low, but they allow closer spacing which, in turn, increases the distance between vertical nulls. And as you said, crossover frequency and slope comes into play as well, with its own sets of tradeoffs to balance.

I tend to favor widely spaced vertical nulls because nulls in the listening area are show stoppers to me. The response anomalies are enormous, quite audible, and they trump any other benefits. To keeping nulls out of the listening are is a first priority. Spacing them outside the pattern is a must. But pattern control must be established as soon as possible, and a horn with too small dimensions is not much use. Ideally, the speaker is designed so the nulls are outside the HF pattern, perhaps cutting into a widening upper MF pattern.
 
In Toole's book there are references to experiments conducted in anechoic environments with adjustable and moveable secondary sources standing in for wall reflections.

Are you talking about the paper by Flindell et al? That would be
AES E-Library: Subjective Evaluations of Preferred Loudspeaker Directivity

More interesting are the papers by Naqvi et al:
AES E-Library: Active Listening room simulator: Part 1
AES E-Library: The Active Listening Room Simulator: Part 2
AES E-Library: The Active Listening Room: Part 3, A Subjective Analysis
 
In Toole's book there are references to experiments conducted in anechoic environments with adjustable and moveable secondary sources standing in for wall reflections. An experimental setup like that should be able to remove catapult's concerns over directivity differences being swamped by other differences between two speakers since you're only testing one speaker, varying the (simulated) reflections. As Earl has mentioned the test focused on listeners preferences for "spaciousness" not "imaging".

I'm unable to provide bentoronto with the links to those studies that he so strongly believes don't exist because I lent the book to a friend months ago but the references are in there. The book is a pretty wide summary of research done on the subject of correlating listener preferences to hard measurements, not an ad for Toole's work after "he became a figment of Cosmic Audio Corp." as bentoronto put it. He had retired from Harman before writing the book.

Bentoronto, you came here asking for evidence, we told you where to look. How that is offensive is beyond me. Look at what work has been done or don't, it's doesn't matter to me.

That's pretty harsh, eh.

But thanks much for providing that background information. I dropped into Toole's test space more than once - I am truly shamed to think that the Nat. Res. Council provided such a minuscule anechoic-kind-of workspace. Maybe I wasn't on the Board of Directors of the division at the time! If I recall, his appointment was in Physics, but his lab was in our building, the Division of Building Research, if I recall correctly. And he was always gracious with me in providing reprints and advice, albeit long ago.

I know this will seem odd, but I urge people to do simple "experiments" but these can't be interpreted beyond their foundations. Such is the case with the somewhat casual tests that Toole did at the NRC. At one point, he tried to simulate living room listening spaces, if I recall correctly. Wonderful to have this kind of research, groundbreaking and important, but still pretty "exploratory" (which is a common euphemism). Please note that many of his projects were "published" in magazines - a good thing for dissemination to our crowd but not the same thing as the Acoustical Society journal.

I've read some of Earl's stuff and am in his debt for posting his interesting and compelling arguments for all to download.

I have Toole's book on order and look forward to becoming far better informed. I think it is exactly precise to say that for the few days I've been asking for references here and excepting Toole's book, the total number of references I have gotten from this thread is: zero.

For sure I am not the only newbie that would like a tiny bit of orientation and background. Always a choice at a forum of being open or being.....
 
Last edited:
Thanks markus.

bentoronto, the tone of your previous posts gave me the impression you were more here for a hit and run argument than truthfully interested in what had been discussed so far. If you are reading the thread then I apologize. Hopefully you find what you're looking for.
 
I have Toole's book on order and look forward to becoming far better informed. I think it is exactly precise to say that for the few days I've been asking for references here and excepting Toole's book, the total number of references I have gotten from this thread is: zero.
If we aren't satisfying your demands, you could always try doing your own search for articles. Going to the AES page and typing in the keyword Harman would be a good place to start. I found 167 papers, some of which might interest you.

Floyd Toole is a legend but a lot has happened at Harman since he retired. Sean Olive is the head of R&D now. Not that Harman is the final word on acoustics but they have done more work than most on correlating measurements with listener preferences and they have a better research facility than any university I know of.
 
The level of the far side reflection will be similar to what you might expect from nearside in a non-toed in CD setup, however, there will be a greater delay..


buzz.. 😀

The far side reflection is NOT similar to a near-side reflection. Near side reflections (..assuming they are uncorrelated), tend to provide the subjective sense of an additional "source" at higher freq.s that CAN expand the apparent sound-stage or "width" of the venue. (..correlation refers to boundary's within the wavelength of the particular bandwidth, most notably 2-8 kHz - which is rare beyond the baffle itself. You do still have to be concerned with power-response and channel balance however.)

By contrast far side reflections tend to marginally augment "depth" at the EXPENSE of "width", by increasing monophonic emphasis (..which is better served with other reflections).


Also,

While the delay is somewhat greater, so is the delay from direct sound.
 
Last edited:
buzz.. 😀

The far side reflection is NOT similar to a near-side reflection. Near side reflections (..assuming they are uncorrelated), tend to provide the subjective sense of an additional "source" at higher freq.s that CAN expand the apparent sound-stage or "width" of the venue. (..correlation refers to boundary's within the wavelength of the particular bandwidth, most notably 2-8 kHz - which is rare beyond the baffle itself. You do still have to be concerned with power-response and channel balance however.)

By contrast far side reflections tend to marginally augment "depth" at the EXPENSE of "width", by increasing monophonic emphasis (..which is better served with other reflections).


Also,

While the delay is somewhat greater, so is the delay from direct sound.

"level of". If you want to allow early reflection for width that's fine but my understanding is that more effective decorrelation between the nonreflected wavefront and the reflection occurs when there's a greater delay. It's the preference issue again.
 
I think it is exactly precise to say that for the few days I've been asking for references here and excepting Toole's book, the total number of references I have gotten from this thread is: zero.

Obviously you just read what you want to read. There're plenty of references in this thread, e.g. just one post before yours: http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-461.html#post2237580

While waiting for Toole's book you might want to read the (very) short version:
http://www.harman.com/EN-US/OurComp...p/Documents/Scientific Publications/13686.pdf
You'll find lots of references in there.
 
Last edited:
In continuation of the wave guide posted here,
http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-423.html#post2162280
http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-425.html#post2196251
http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-425.html#post2196636
I have finally slipped in some time to extend the baffle of the wave guide beyond the lip to a round baffle with a diameter of 40cm. I should be doing some measurement either Friday or the comming Monday. Are there any guesses what this will do to the response?😀 The curves up to the blend with the baffle remains the same. The diameter of the WG where is blends with the baffle is about 22.5cm.
After some quick runs of measurements, it seems Earl was correct. What looks like an on axis hole is due to the lip curvature, not the abrupt cutback after the lip is tangent to the baffle. Scott's observation seems also correct that it does not seem to be exactly a hole.

What it seems like is that the lip is causing diffraction such that the dip and peak shift towards a higher frequency as the mic is moved away from the WG.
 
Hello Soongsc:

What it seems like is that the lip is causing diffraction such that the dip and peak shift towards a higher frequency as the mic is moved away from the WG.

As you describe it, this behavior could be also interpreted as related to reflexions or HOMs appearing deeper on the wall of the waveguide when frequency rises.

MigeO's simulations have shown that the longer the flat part of the profile is, the larger is the "axial dip" (also if memory doesn't miss me the same variation with frequency was shown).

Best regards from Paris, France

Jean-Michel Le Cléac'h
 
As shown in the attached pic,
magenta - previously measured @33cm
red - 98cm
blue - 141cm
green - 46cm
The dip occurs only outside the WG as previously measured.
 

Attachments

  • outside WG.gif
    outside WG.gif
    14.4 KB · Views: 292
Hello Soongsc,

.The dip occurs only outside the WG as previously measured.

Probably because at the highest frequency showing dip the (oblical) reflexions ( considered as rays for an easier purpose) can only interfer with the direct wave from a certain distance from the throat .

(See the somewhat simplified graph showing the principle of the oblical rays that may interfer with the not shown direct rays...)

Best regards from Paris, France

Jean-Michel Le Cléac'h
 

Attachments

  • rays_OB.gif
    rays_OB.gif
    3.5 KB · Views: 285