Acoustic Horn Design – The Easy Way (Ath4)

[fluid]

Look at styles within VACS as that seems to offer a way to easily apply range and layout options.

 

[limacon]

Working on baffle with HF10AK + 8PR200 . Still tweaking...

 

[limacon]

Seemingly resolved with a minor change to Morph.CornerRadius.

 

[limacon]

Updated charts.

Mesh.AngularSegments = 120
Mesh.LengthSegments = 64
Mesh.CornerSegments = 6
Mesh.ThroatSegments = 6
Mesh.ThroatResolution = 4 ; [mm]
Mesh.InterfaceResolution = 4 ; [mm]
ABEC.f1 = 800 ; [Hz]
ABEC.f2 = 16000 ; [Hz]
ABEC.NumFrequencies = 60
ABEC.MeshFrequency = 1000 ; [Hz]
ABEC.Polars.Dist = 2.5 ; [m]
ABEC.Polars.Step = 5 ; [deg]

 

[fluid]

Looks good, I imagine with 60 frequencies and 4mm resolution it took a while to solve? How many elements did you end up with?

Reducing the mesh length segments to 20 or so and adding this line as recommended by mabat helped me

Mesh.ZMapPoints = 0.5,0.1,0.5,0.98

 

[limacon]

Threadripper and walk away. Thanks for the tip, will give it a try.

 

[limacon]

Mesh.ZMapPoints = 0.5,0.1,0.5,0.98

Is this a gmesh function? What is it doing to the mesh?

 

[mabat]

Here is the explanation: #1119.

('Mesh.DepthSegments' renamed to 'Mesh.LengthSegments' in the meantime. Also, as fluid said, it is now recommended to decrease it for ABEC mesh, 20 would be a good start.)

 

[limacon]

Ah, thanks!

 

[limacon]

Tweaking this band. Slowly working towards a flatter response horizontally +/-45 deg.

 

[mabat]

Does it involve intentional diffraction? Sure I'm eager to see where you can get just by shaping the horn.

 

[limacon]

Pastel colours.

 

[limacon]

Does it involve intentional diffraction? Sure I'm eager to see where you can get just by shaping the horn.

I was going to have a look to see if the mesh could modified with meshmixer is sclupt mode, pull and tug in a few places to just see what happens on the charts.

 

[mabat]

Hmm, that's probably not so easy to keep the mesh file usable as prepared by Ath - you would have to modify the ABEC scripts for your modified mesh files(s). For Meshmixer you need STL and by converting from MSH you lose the tagging necessary for ABEC scripts to work. Maybe I could split the one MSH file into several seperate files, MSH or STL (waveguide, interface, driving, etc.) but you would need to make sure it all fits together.

- The waveguide looks good, BTW. Maybe too deep for my taste but I'm definitely looking forward to see the measured data. Good job.

 

[limacon]

I ended up at 232Wx218Hx98D

 

[mabat]

It looked bigger on the previous picture. This is still fairly small then.

 

[mabat]

Finally I eliminated the constraint on quarter-symmetry.

 

[Stéphane B]

WOW ! I this model asymmetric ? It looks like so but my eyes and the 3D model might be deceiving.

EDIT : maybe you can post a front view ? Thanks

 

[mabat]

It has seven vertices which is pretty asymmeric in my book
(It's just a random shape put together by some superformula numbers for m=7.)

It was necessary to remove this limitation to be able to incorporate the enclosure, which usually isn't quarter-symmetric. So I eliminated the constraint altogether. Now I can put the waveguide into a (rounded) box, basically anywhere on the front side (symmetry still utilized where applicable).

 

[mabat]

Anyone?