Acoustic Horn Design – The Easy Way (Ath4)

"The script Marcel made does it automatically and you end up with a body."
Are you talking about the output.stl?
I've imported that stl and converted to BREP, still could not thicken. see image
I must be missing a step.

Axissymetric horns are easy with Marcel's FusionRotaryProfile.
I'm just stuck on thickening the Tritonia.

If you have, could you share a screenshot please? thanks
 

Attachments

  • Tritonia.png
    Tritonia.png
    503.5 KB · Views: 474
It's actually extremely simple. I will add this to the User Guide.


1) Export the slices from Ath.
Try to use as low mesh resolution as possible to preserve the shape fidelity. Fusion will interpolate splines through the mesh points.
This is what I used here:

Code:
Mesh.AngularSegments = 60
Mesh.DepthSegments = 20
Mesh.CornerSegments = 5

GridExport:f360 = {
  ExportProfiles = 0
  ExportSlices = 1
  Scale = 0.1
}

In this case this will create a file Tritonia_slices_f360.csv in the project output directory.


2) In Fusion install the Ath scripts(s):

attachment.php



3) Run the "Ath4_SurfaceImport" script. In the "File Open Dialog" select the exported csv file.

attachment.php



4) Add thickness to the created body (Solid - Create - Thicken):

attachment.php
 

Attachments

  • add-ins.PNG
    add-ins.PNG
    29.6 KB · Views: 1,067
  • surface.png
    surface.png
    320.9 KB · Views: 712
  • thickened.png
    thickened.png
    163.1 KB · Views: 740
Last edited:
  • Like
Reactions: neutrino_th
... "to use as low mesh resolution" - I suppose you ment a "fine" mesh resolution to preserve....

I suppose we would look for local eruptions of GD, indicating stored and later released energy in certain frequency areas. I do understand that such aberrations might not be know to be audible but at least, we can identify areas where the WG/drivers are diverting form the ideal. If one would like, por any reason, avoid these we would have the opportunity.

//
 
This implies that large local group delays may be more audible with level than smoother ones. I believe that this is a large part of "horn honk" that I think that we have all heard - it's perception increases with level until it sounds like nonlinear distortion - which it is not.

Shouldn't it be possible to confirm or falsify this by using a FIR filter to flatten the group delay? Whether or not using FIR filters in general is desirable or not is another story, but one should be able to do a more or less ceteris paribus A-B comparison regarding this hypothesis, no?
 
Yesterday I fought a bit with adding the scripts to F360. First, they are scripts and not add-ins (first confusion) and second, the folder names must be exactly the same as the script name and manifest (second onfusion) to show up in Fusion when copied to the proper directory. I would humbly suggest to rename the folders in the ATH package and add a line on how to install them. (sorry if I missed it in the docs)
 
Dear Mr. Marcel,

I am getting the following message (image attached) when I try running the ATH software on my computer.

If this has anything to do with the word-length (32/64 bit) of the computer, kindly share a 32-bit version of the program (if available). My operating system is 32-bit Windows Vista Home Premium.

Thanks in advance.
 

Attachments

  • Untitled-1.jpg
    Untitled-1.jpg
    102.5 KB · Views: 103
I am trying to achive a wide horizontal coverage for an 33mm driver and just playing around with differnt settings in ath4. Independent from the settings which one of those two coverage version is to prefer?

The first one is constant between 2-10khz but but then falling rapidly after 10khz. The second one does not have constant coverage between 2-10khz but coverage is falling smoother over the complete range 2-20khz.
 

Attachments

  • wide_dispersion.jpg
    wide_dispersion.jpg
    67.1 KB · Views: 394
  • smooth_dispersion.jpg
    smooth_dispersion.jpg
    68.7 KB · Views: 382
Last edited:
Dear Mr. Marcel,

I am getting the following message (image attached) when I try running the ATH software on my computer.

If this has anything to do with the word-length (32/64 bit) of the computer, kindly share a 32-bit version of the program (if available). My operating system is 32-bit Windows Vista Home Premium.

Thanks in advance.
Try this, it should be a 32bit executable. I had some difficulties compiling for 32 bits in the past, so I hope it works: https://at-horns.eu/release/ath-4.7.1beta-32bit.zip
 
32-bit version

Try this, it should be a 32bit executable. I had some difficulties compiling for 32 bits in the past, so I hope it works: https://at-horns.eu/release/ath-4.7.1beta-32bit.zip


Yes, the one you gave does compile on a 32-bit computer.

However, when I tried enabling the circular symmetry function, reducing the number of slices and angular segments, sampling frequency etc., the same did not reflect in the output ABEC project file.

Incidentally, other ABEC examples seem to solve and show spectra on VACS.

Is the 32-bit executable designed to have only the set parameters in the user guide PDF file? For example, the length of the horn is always 94mm.
 
Last edited:
... However, when I tried enabling the circular symmetry function, reducing the number of slices and angular segments, sampling frequency etc., the same did not reflect in the output ABEC project file.
If you really enable the circular symmetry (i.e. set 'ABEC.SimProfile') you don't have to worry about the computation time, it will be always pretty short. So there's no reasons to reduce the number of slices. Also, keep the mesh (sampling) frequency high, certainly above 30 kHz. This way you will get stable and quite accurate results up to 20 kHz.

It is the general, non-circular shape for which has this optimization much more importance, otherwise you may wait for many hours (and then even not get the result at all).
 
Last edited: