Acoustic Horn Design – The Easy Way (Ath4)

mabat, in the instructions there is mention of simulating a freestanding guide as an exterior subdomain only without an interface. Have you been able to do this successfully? I have tried with Ath and my own meshes and just generated garbage.

To get good results I needed to have the interface be on the last slice and offset enough to not interfere.

I have been trying to get the report function to work on a meshed xy freestanding horn but I get these errors. The vacs output looks OK, any idea what is going wrong?

attachment.php
 

Attachments

  • Report Error.jpg
    Report Error.jpg
    105.6 KB · Views: 898
This is the config

Code:
; Ath version 4.7.0
ABEC.MeshFrequency = 1000
ABEC.NumFrequencies = 48
ABEC.Polars:SPL = {
  MapAngleRange = 0,180,37
  NormAngle = 10    ; [deg]
  Distance = 3      ; [m]
}
ABEC.SimType = 2
ABEC.f1 = 600
ABEC.f2 = 10000
Coverage.Angle = 40
Length = 160
Mesh.AngularSegments = 64
Mesh.InterfaceDraw = 10
Mesh.InterfaceOffset = 30
Mesh.InterfaceResolution = 12.0
Mesh.LengthSegments = 20
Mesh.RearResolution = 10.0
Mesh.RearShape = 1
Mesh.SubdomainSlices = 19
Mesh.ThroatResolution = 5.0
Mesh.WallThickness = 5.0
Output.ABECProject = 1
Output.STL = 1
Output.SubDir = 4.7
Report = {
 PolarData = "SPL"
 NormAngle = 10
 MaxAngle = 90
 Width = 1024
 Height = 768
}

This is the observation script

Code:
Driving_Values
  DrvType=Acceleration; Value=1.0
  401  DrvGroup=1001  Weight=1.00 Delay=0.0

Radiation_Impedance
  BodeType=Complex; Range_min=0; Range_max=2; RadImpType=Normalized
  402   1001 1001   ID=8001

BE_Spectrum
  PlotType=Polar; GraphHeader="PolarMap_SPL"
  BodeType=LeveldB; Range_max=5; Range_min=-45
  PolarRange=0,180,37
  Distance=3m; BasePlane=zx
  NormalizingAngle=10
  501  Inclination=0  ID=5001
 
Yes the graph above is the same guide, (a different set of frequencies 600 to 10K to reduce time when fully meshed) but otherwise no difference.

I had to reduce the mesh density more than usual to get the elements reasonable, and the interface made more of a difference this time. Where did you place the interface when the sims were similar?

The spiky ripple is definitely the sim rather than the guide.

Circysm test for comparison

attachment.php
 

Attachments

  • Test4C.png
    Test4C.png
    129.3 KB · Views: 514
Then I suspect something must be different between the two models. This doesn't look like a typical numeric issue (only).

IIRC, I placed the interface at the apex of the mouth (in the full-3D case) but don't remeber the details. But I agree that these details have weird effects sometimes. With the CircSym, I settled on using just one (exterior) subdomain.
 
Last edited:
hi


here is couple of my tries, one OS-SE for 1" and two CircArc, 1" and 1.4". All for asymmetric free standing WG's. They are looking promising, I hope so.
There is one thing what is not clear for me. OSSE show significant rise in DI when wavelength become smaller than throat diameter. On other side CircArc is almost flat. Is that some simulation error or inherent feature of CircArc design. If is it later is there any explanation for it
 

Attachments

  • osse.png
    osse.png
    110.9 KB · Views: 109
  • circ.png
    circ.png
    103.5 KB · Views: 119
  • circ14.png
    circ14.png
    104.8 KB · Views: 122
You are definitely on the right track :)

The CircArc examples would be exceptionally good, IF you used a practicable source at the thoat - I suspect that you left the default, matching spherical source there (i.e. a pulsating one). Then it surely would be this good but such source doesn't really exist - no available compression driver will do that at the moment, I'm afraid (and maybe we still could try to make a one).

- Try to connect a common throat extension of a compression driver to the above models (Throat.Ext), driven by a flat wavefront (Source.Shape) and see what you get. Or just a flat wavefront directly, without any extension.
 
Last edited:
I would use either flat or a spherical wave but with much larger radius (and in that case it's maybe best to try to simulate an exit section of a specific driver).

What is your opinion about lack of DI rising in CircArc?
As I said, I find it realistic with that source. That's the reason I tried to develop the spherical wavefront phase plug - it could be this good, no matter the throat size, at least in principle. (In the end it probably woudn't because of other reasons.)
 
Last edited:
I'm just guessing here because I don't know the wavefront shapes at the exits of compression drivers, but if there's a conical extension (as almost always is, although there are drivers without it), I assume a spherical wave corresponding to that exit angle could be used as the first approximation. Or just a flat wavefront..
 
Last edited: