Except that is not what happens. I'm not just making it up I have run the simulations. Once you go finite excess flat baffle area just adds more mess.
Nice mabat! Thanks posting these flat disk sims
With quick look it is quite obvious that flat disk would benefit from a waveguide. Also interesting to see how much difference from the depth and back corners make, surprisingly much effect on frequency response graph.
What to do with back of the box diffraction? Make a box meter deep to get good 5ms delay for the back diffraction?
make triangle box with pointy back tip to have both side sound cancel the other side by diffraction? Thinking is that as positive amplitude goes around the tip to otherside, negative amplitude from diffraction "reflection" goes back toward front, and this happens both sides simultaneously making positive and negative amplitude go to same direction (towards listener) at same time on both sides. Would this get rid of the cabinet backside diffraction?
Perhaps such is not needed, because the response is fine <1kHz for box of that size no matter what the roundovers are, just use freestanding waveguide above that
With quick look it is quite obvious that flat disk would benefit from a waveguide. Also interesting to see how much difference from the depth and back corners make, surprisingly much effect on frequency response graph.
What to do with back of the box diffraction? Make a box meter deep to get good 5ms delay for the back diffraction?
make triangle box with pointy back tip to have both side sound cancel the other side by diffraction? Thinking is that as positive amplitude goes around the tip to otherside, negative amplitude from diffraction "reflection" goes back toward front, and this happens both sides simultaneously making positive and negative amplitude go to same direction (towards listener) at same time on both sides. Would this get rid of the cabinet backside diffraction?Perhaps such is not needed, because the response is fine <1kHz for box of that size no matter what the roundovers are, just use freestanding waveguide above that
Would be great to see 14,16,18,20,22,24in baffle widths with 0,1,2in rads. Sometimes the data informs gut feeling.
I don't think you understand quite what you are asking for. These models have high resolution and that is needed for waveguides. When you extend that size out to 32" even with a progressive reduction in element size, the number of elements and consequent solving time goes up quite considerably, on an ordinary machine to 2 or 3 hours per sim even managing the elements to keep them down.
mabat maybe willing to make the models if limacon or someone else wants to crunch them on their computer.
Ok I see now, I'd crunch it, if possible, I have akabak and ath but am a noob at both.
I dont know if this is fast enough
I dont know if this is fast enough
I'd want to go wider than that but yeah, the data that is being shown is pretty interesting, One day I'll be able to mock up the simulations on my own...that day isnt today though lol
Last edited:
In a case of a 3D model, ATH generates a GEO file which is read by Gmsh to generate a MSH file which is used in ABEC as the BEM mesh.
That was only extended to 20 kHz.
I tend to agree[...] Perhaps such is not needed, because the response is fine <1kHz for box of that size no matter what the roundovers are, just use freestanding waveguide above that
Perhaps the next logical step would be to implement a simulation of a passive cardioid woofer structure. The only trouble is that it may depend on damping, which can't be dealt with in BEM. What can be independently controlled in a simple BEM sim however, is the radiation of the rear side of the diaphragm. Maybe that would suffice to produce something usable.
Any ideas how should such general and parameterizable structure look like are welcome. I'm sure there's already a plenty of experience out there.
Any ideas how should such general and parameterizable structure look like are welcome. I'm sure there's already a plenty of experience out there.
Last edited:
This is interesting. Will it be possible to try out a double chamfer and set individual angles? Clearly, the slanted edge produces a massive localization of diffraction effects higher up in the passband. Maybe something for a three way, but a two way would suffer dearly.
Or add helper drivers on either side or just the back but that means DSP...
However probably easier to simulate?
An easier way to model cardioid is to model the side vents and then incorporate that model into a Vituix sim that also includes the front facing drivers. Such a model would have radiating surface in place of the side or rear vents. In Vituix, you add filtering, delay, attenuation, and 180 degree phase shift for the side vents to optimize the response. You still have to figure out how to translate that into absorbers behind the vents or you can take the easy way out and add active side drivers and DSP instead of vents.
Having said that, since I know you can include crossovers in an ABEC sim even if I don't know how to do it myself, you could use that capability to model the absorber structure of a passive cardioid. Even so, I would first want to optimize that filter structure in Vituix before incorporating it in ABEC. An iteration in Vitux takes perhaps a second, in ABEC it would take minutes to hours.
In ABEC, once is the BEM part solved (which is the time consuming part), you can change source parameters (levels, delays, additional signal filtering, etc.) without the need to re-solve the model, so this is not an issue - you need only to refresh the observation stage which is very fast.
To me, adding more sources and DSP doesn't have the appeal.
To me, adding more sources and DSP doesn't have the appeal.