Acoustic Horn Design – The Easy Way (Ath4)

limacon

Member
2019-10-19 10:09 am
What, ripple tank in 3D?:D Nice!
Yeah. There are many ways to look at this in BabylonJS. I can fill the waveguide space with 100 million GPU particles and then apply a pink noise vibration to the mouth. Quick example of a block moving through through a point cloud. Gives you an idea of where I'm heading.

https://playground.babylonjs.com/#UI95UC#21
I'll start a new thread I think. The app is quite mature now.
 
Fascinating. If one now could see in real time the effect of the sliders in your standard report (e.g. 39-48.png above) it would be a fun excersie :)

//
While the sims probably take some time not to have it exactly real time this kind of thing should be possible on browser. Both desmos and matlab/octave have API. I have no idea how much work it would be :D probably at least few evenings worth + expensive server to get performance up, or use localhost.
 
Fascinating. If one now could see in real time the effect of the sliders in your standard report (e.g. 39-48.png above) it would be a fun excersie :)

//
I've envisioned this too and am trying (struggling) with MMM in python/cupy to run stuff on the gpu. My CPU is not that powerful and parallel (optimization) does not work (at least out of the box) in Octave.
 

limacon

Member
2019-10-19 10:09 am
Still some bugs to resolve with the waveguide curves / rendering.
 

Attachments

  • capture.gif
    capture.gif
    8.2 MB · Views: 51
Remember that in your case it's an infinite baffle simulation and that the DI will be higher towards low frequencies when placed in an enclosure. If there's a midrange narrowing in an infinite baffle, it probably will get worse in a finite baffle. So I wouldn't try for the flatter DI.

That's one of the reasons I like free-standing horns - one can easily simulate and optimize the real free-space performance without the need to model an enclosure. Besides that they work like a charm.
 
Last edited:
Tweaking GCurve.Dist = 0.39 / 0.41 / 0.43 (animated gif)
Is it possible to have this sort of optimization happen with a baffled device, maybe even a box with roundovers?

I still have the problem with a waveguide I had been simulating for many hours. Recently, fluid was so kind and made the effort to do a limited simulation in a baffle with ABEC. I needed to see how the performance changes on on a baffle. Drastically so. This showed to me that it is rather useless to optimize a baffled waveguide without regard to the diffraction effects generated by the box itself. Its impacts are too severe. Is it possible to do it with your new "API"?

edit: mabats cross post underscores the demand for such a solution