Pseudo-coaxial with narrow directivity (and Horbach-Keele filters)

*UPDATE*: Here is now my final build. :)


I want to show you my prototype of a pseudo-coaxial design. It is the 10th of its kind. It is not easy to achieve a constant directivity that is narrow at the same time.

520389d1450687425-pseudo-coaxial-narrow-directivity-horbach-keele-filters-_mg_2018_klein-jpg


It took me months to get the directivity like it is. And it still could be better under high angles. But since the overall amplitude is so low under these angles I doubt that it counts much in a medium damped room.

The cabinet dimensions are 50 x 50 x 20 cm. The final version will be a bit more shallow. This is the reason why a synergy horn was not an option. Depth is what I don't have in my room.
The horn is my own design and was 3D printed. The final version of this speaker will have a CNC-milled baffle which is made of one piece. But for the prototype this was not important.

The crossover is linear phase and is based on a Linux PC with BruteFIR. It uses optimized Horbach-Keele filters to achieve the directivity. First I tried minimum phase filters, but without phase correction the result was never as good as with linear phase filters.

Horbach-Keele filters:
520393d1450687717-pseudo-coaxial-narrow-directivity-horbach-keele-filters-filter-png


The whole speaker was designed to be used with a subwoofer system and should be crossed at about 100 Hz. I also evaluated several possibilities to integrate a 4th way (the woofers). But the cabinet would be much larger and I don't see any benefit for a narrow dispersion in the low frequencies inside a room. So I decided to install a subwoofer system later.

If more maximum SPL is desired the number of low-mid-drivers can be doubled without influence on directivity.


Measurements

Directivity normalized to 0°:
520390d1450687425-pseudo-coaxial-narrow-directivity-horbach-keele-filters-abstrahlung-png


Directivity normalized amplitude (10° steps, 0 - 90°):
520391d1450687425-pseudo-coaxial-narrow-directivity-horbach-keele-filters-abstrahlung-amplitude-png


Directivity index:
520392d1450687425-pseudo-coaxial-narrow-directivity-horbach-keele-filters-ndelungsma-png


The whole story of the prototypes can be read here (in German).

Sound
I can not tell too much about sound, since I have only one prototype and never listened to stereo with it. But the perception of distance is impressive. Even at 2.5 m in a small underdamped room it sounds like a nearfield monitor. And you can go very close to the baffle and the image never collapses to multiple point sources. This is truly an advantage of the pseudo-coaxial design.

I think this speaker works very well with larger distances and underdamped room. And it can play loud (low distortion), because there are many drivers and a high crossed horn in the middle.

When I build the final version I will post pictures. It will be looking a bit better. ;)
 

Attachments

  • _MG_2018_klein.jpg
    _MG_2018_klein.jpg
    210.1 KB · Views: 10,327
  • Abstrahlung.png
    Abstrahlung.png
    47.4 KB · Views: 3,801
  • Abstrahlung Amplitude.png
    Abstrahlung Amplitude.png
    16.6 KB · Views: 3,744
  • Bündelungsmaß.png
    Bündelungsmaß.png
    9.1 KB · Views: 3,655
  • Filter.png
    Filter.png
    9.8 KB · Views: 3,739
Last edited:
  • Like
Reactions: 1 users
Hi FollGott and hello neighbor ;)
I'm working in Hanover and went there studying some years ago.
You've created a very interesting design. Are the small mids Aurasounds NS-2?
From my point of view: well done!

On this forum the guys are more experimenting with the Synergy-Horn.
 
  • Like
Reactions: 1 user
Hi FollGott and hello neighbor ;)
I'm working in Hanover and went there studying some years ago.

Hi neighbour. I studied here in Hanover, too. Maybe we met some day and didn't know it. ;)

You've created a very interesting design. Are the small mids Aurasounds NS-2?

These are Aurasound NSW2-326-8A. The lower midranges are Tang Band W4-1320 and the horn is loaded by a Celestion CDX1-1425.

On this forum the guys are more experimenting with the Synergy-Horn.

Yes, I noticed that. But maybe an alternative with smaller depth could be interesting for some. :)
 
  • Like
Reactions: 1 user
Wow........that's really cool! I've thought it would be fun to try an array like that but I had no idea that the pattern could get that narrow. Very attractive if you don't have much depth for sure. You said the woofers are what limits the spl....what is the array of mids capable of? I'd imagine they do ok working over 1khz.
 
  • Like
Reactions: 1 user
You said the woofers are what limits the spl....what is the array of mids capable of? I'd imagine they do ok working over 1khz.

I don't know exactly. I have to do THD measurements. But I need to put the speaker in the garden for that. Maybe I'll can do it in spring.

I don't think the Aurasounds are limiting in its frequency range.

What are your intentions as far as toe-in, and covering multiple listeners?

The sweet spot is not large. Since directivity gets much wider under about 800 Hz the sound changes under angles. The lower frequencies stay nearly constant in the whole room (because of reflections), but the higher doesn't. So this speakers must point directly to the listener. And it is clearly a one-man-speaker. ;)
 
Does anyone know of any more examples of this type of design?

I don't know any.

It is a 2D version of Horbach's and Keele's design. But they never showed a version with a constant directivity in the highs. The tweeter is always the problem, because it is alone and has no partner. And its bandwidth is higher than the other ways.
When doing 1D with waveguide the distance between the mid-high-drivers is too large. It doesn't work. So I switched to 2D and increased the grid depth of the mid-high-drivers to shift side lobes to higher frequencies.

I failed to design a pseudo-coaxial speaker that has a wider directivity. And I failed to design one with different horizontal and vertical directivities.

But I tried to separate horizontal and vertical with a ribbon driver and a horizontal waveguide. It doesn't work that good, because ribbons can not be crossed deep enough. And plane wave formers for compression drivers are usually too large. But the concept could still be promising.
Read the documentation here (again in German).
 
Slide 21 in Horbach-Keele Presentation Part 2 V4.pdf:

Yes, but they never proved that it worked. ;)

A directional tweeter (waveguide or ribbon) always has larger dimensions than a dome. And this increases the distance of the surrounding drivers (and shifting lobes down in frequency etc.). The whole concept is based on a certain distance between the drivers. A directional tweeter is contrary to the concept.
 
Last edited:
Nils - does the pattern hold up at oblique angles?

Not completely. It gets a bit worse, but not much. I have only simulations yet and have to measure the diagonal direction on the prototype.

Superb project! How have you determined the exact placement of woofs and mids? Or should I ask: what software did you use to come the the final design?

Thanks! I used ABEC to determine the best positions, sizes and crossover points of the drivers. Usually the prototypes were build after a promising simulation that I designed after an idea. And it always fit good with that.

I'm a bit proud of the optimized alignment of prototype 7. It looks weird and chaotic, but there is a system behind it. I considered the horizontal and vertical directions indepenently.
To shift the side lobes to higher frequencies the drivers must be placed near and equally spaced. You can consider the grid as "weighted". If one driver is on one of the fixed positions the weight is 1. If there are two the weight is 2. All positions must be equal weighted to get the best result.
Prototype 7 is the result to get a high density grid of midrange drivers and to weight all drivers equally in both directions.

attachment.php


With this trick I was able to shift the side lobes to >5 kHz. This is important, because to shape directivity you need both ways to widely overlap in frequency.

Anyway, I skipped the 2-way design and switched to 3-ways. I found that the big horn didn't sound as good as the small one. Sibilants sound much better with the current prototype.
 

Attachments

  • gitter.jpg
    gitter.jpg
    22.5 KB · Views: 4,197
Last edited: