DIY: Omni Directional Speakers

Surely this is further mimicking what happens with a live instrument, and therefore beneficial to the impression of realism...
Sure? Speaker are about reproduction... I'm not convincing at alla that a speaker should "work" like a music instrument. So many different instruments with their individual way of emitting sound in different directions. A speaker model for solo violin, an other for solo congas? Hm...

//
 
  • Like
Reactions: jaapVL and MrKlinky
if its so difficult to combine good time response and omnipolar response still the classic configuration can be a problem solver:

like this one but with a dsp'ed fullrange like Fane Sovereign 12-250tc.

but a concentric design would do, too.

th-522410462.jpg




th-379794764.jpg



th-489391314.jpg




last one from canon s70 has the advantage of not collecting dust over time
 
Last edited:
I'm not convinced at all that a speaker should "work" like a music instrument. So many different instruments with their individual way of emitting sound in different directions.
Of course, however a conventional forward-firing loudspeaker will have a polar distribution quite unlike most acoustic instruments. I'm quite open to the possibility that the ideas cited above have real merit, but of course any omnidirectional speaker is only omnidirectional to a single listener if it has something to reflect off, and that's where its effect - for good or bad - gets difficult!
 
A speakers radiation must have something to do with how microphones "listen"... I suppose there must be some symmetry in the rec/repl chain in order to make a faithful reproduction. Multi mic/ studio/electrical instruments can only hope to be "nicely" played.

//
 
In post 52 I argued that adding four tweeters to the omni's could solve some problems.
Although I still think this is possible I am not going to continue down this path for the following reasons:
1) the approach requires a lot of (measurement) work.
2) the end result is still quite uncertain.
3) during the project I cannot listen to music in my living room.

I am now thinking of building new omni's with a sphere shape controlled by active electronics. I expect less risk with this approach and better results.
 
in this construction the tweeters do the omni effect. Close to back wall the 20cm bass gets some compensation of baffle step for free.

And it sounds good. I understood well why these sonab boxes were so popular in northern europe in the 70ies. They nearly had a monopole standing in sweden.

https://www.diyaudio.com/community/...oupled-up-firing-speakers.121385/post-6707993

It sounded much better with the felt below the tweeters.

IMG_20201027_204305.jpg
 
  • Like
Reactions: Juhazi
@jaapVL

if you take one of your small drivers and put an acoustic lense on it like 3-4cm diameter opening it would beam less and not worse than some tweeters.

With nearly 90 to 120 degree dispersion you would only need two of them one on front and one in the back keeping the box as little as possible like a small ball.

With a subwoofer it would certainly give a nice response if DSPed linear.
 
if your driver is 2.5 inch just put a round opening before it with only one or 1.5 inch. You get dispersion like with smaller driver.


its like the opening of a carlson coupler but only round and smaller than the driver.

if EQed it should give good results aiming at enhanced dispersion.

Just keep an eye on distortion. If the opening is too small there is too much compression.

BBC did that with an old model
 
As Visaton boxsim software can model a cube with a driver on each face I did that
Thanks for the hint. Although a cube is quite different from a truncated octahedron, I made an attempt to simulate the behavior of the omni with Boxsim.
For the simulation I placed the four drivers on a cube (riblength 20 cm) as below, resulting in the frequency response as shown.

1724158289760.png

The agreement between the simulation result and the REW measurement is clear (see post 33 of this tread).
1724158390978.png


Also the directivity plot of the simulation is not in disagreement with my REW measurements.

1724158488521.png


Next I gave the direct driver the amount of delay necessary to virtually place it in the plane of the indirect drivers. This plane is perpendicular on the listeners axis (see post 19 of this thread), The applied delay was 350 usec.

1724158592990.png


1724158635180.png


It is clear that the nasty dip at 1300 Hz on the listening axis has disappeared, as expected.

The above exercises have given me confidence that Boxim can be a useful tool in the further development of the omni.
As stated before, I am already happy with clear and open sound of the omni. The aspect I want to improve is quality of the stereo image.
To achieve this, I want to reach the following:
  • a reasonable flat SPL function on the listen axis
  • better reproduction of high frequencies for better directivity
  • minimize edge diffraction

What I have in mind is a spherical enclosure with 4 woofers and 4 tweeters, delayed in such a way that they all drivers virtually lie in the same plane. It then becomes, in effect, an omni directional version of a small 2 way speaker.
 
  • Like
Reactions: olsond3