Acoustic Horn Design – The Easy Way (Ath4)

[mabat]

No, there's almost never any additional smoothing in the polars I show.
Attached is the ABEC project and my results obtained using ABEC 3.6.0b7 (64bit demo) as calculated a while ago using the script from #10,653 -

 

[mabat]

I saw that the setting for ABEC mesh resolution is only at 1000Hz in the script. Is that correct? I always assumed it must be at least equal to the highest frequency of interest?

That's of course true, in general. The aim here is to setup Ath to generate the mesh with a high-enough resolution already, so the nominal value in ABEC can be kept this low (i.e. it won't do any additional mesh subdivisioning, which is all it could do).

 

[Olombo]

Just running your project to see if there are differences. In setting up ATH I assume not many mistakes can be done as long as it proceeds the input script...!?

That mesh issue is interesting: I was not aware that Abec/Akabak will only sub-mesh if necessary. This explains horrible computing time of some other sims I did.....

 

[mabat]

In setting up ATH I assume not many mistakes can be done as long as it proceeds the input script...!?

Yeah, it's all in the script (the Mesh.* items).
Anyone using the script as it is should get the same results.

- I have the "Preferences - Processing - NUC" set to "None" in ABEC. That's the only possible difference I can think of.

 

[fluid]

I get very scattered results. I tried to normalize it; however the scattering behaviour remains. What am I doing differently maybe?

Something is wrong in the script to cause the response to look like that. It is possible sometimes for the mesh to have a hole in it, or just a few triangles with the normals reversed, and produce garbage results.

 

[mabat]

That could be a Gmsh bug - a different version perhaps (I used 4.6.0 for the project posted above). That's something I can't ensure.

 

[Olombo]

Hi

different...... if I run your project first I wonder that axis for observation seem a bit curious as out of axis:

Is that intended?

Gmesh I have 4.10.5. I will try to find your version online. However the scattered results remain. See e.g. radiation impedance:

Will try the "NUC" setting next....thanks for inspirations!

 

[Olombo]

Forgot to add: same ABEC Version like yours and lates ATH beta for 2-profile rectangular horns. Anyway must be within ABEC if your script works on your machine....

 

[lbstyling]

BTW, the bended board construction also allows an easy making of a big mold - I want to try concrete casting. All in one piece, including the adapter (3D printed part of the mold).

I build some floorstanders from concrete about 10 years ago.
I recommend making super lightweight concrete if you can, as stopping any boards used for forming the concrete from bending under the weight is the problem.

 

[TNT]

How do one make lightweight concrete?

//

 

[Olombo]

no change with NUC setting. Any ideas?

 

[mabat]

I've tried it again on a different machine, maybe already with a slightly different settings. Anyway, it should be a lot smoother.
This is without any smoothing, 40 points from 200 to 5k -

 

[Olombo]

Uninstall.....re-install Abec (new download).....now waiting. Can try another machine tomorrow if this also did not help. Thanks for your effort helping me!

 

[mabat]

BTW, this was one of the iterations, don't know which one anymore..

 

[mabat]

Uninstall.....re-install Abec (new download).....now waiting. Can try another machine tomorrow if this also did not help. Thanks for your effort helping me!

It's strange and I have no clue what can be wrong. I doubt that will help though.

 

[mabat]

if I run your project first I wonder that axis for observation seem a bit curious as out of axis:

Is that intended?

I don't see anything suspicious, that seems correct.

 

[Olombo]

looks better:

Not sure it did accept the "NUC" setting only after re-start or if it was new installation. Anyway time to play! Thanks a lot!

 

[fluid]

Not sure it did accept the "NUC" setting only after re-start or if it was new installation. Anyway time to play! Thanks a lot!

What you had before was not a Non uniqueness problem so NUC would not help just take longer, it looked just like a mesh generation problem but whatever you did fixed it

 

[augerpro]

How do one make lightweight concrete?

//

For loudspeakers, I think Aircrete method might work well. Or using Styrofoam as aggregate.

 

[mabat]

As the shape is more or less conical, I don't expect it to be easy to deform but I'll see.