Acoustic Horn Design – The Easy Way (Ath4)

[pelanj]

bodo: That is actually not that bad!

Patrick Bateman: If would be a meeting in Prague, I would like to attend as well (unless my work sends me somewhere else in June).

 

[Patrick Bateman]

BTW, the tool already has the functionality of morphing an arbitrary shape into a rectangle. You can start with whatever you like and tell the software to e.g. "start at half the depth and gradually deform it so that it becomes rectangle at the mouth". You would get this by setting the parameters "Shape=raw2rect" and "Shape.FixedPart=0.5". Maybe some interesting creations could be made this way. You might also discover you would need some more parameters to be more flexible - just tell me.


...And after all of this is tried and tested, I think it is quite likely that we return back to a good old big round OS WG, heavily rounded (still possible, BTW). And I think the reason you don't see these big devices much around, is because they are BIG.

When you do this, you wind up with a waveguide that's so close to the M2 waveguide, it makes me wonder if this is what JBL was doing:

Here's a comparison between a waveguide I generated with Mabat's software, and the M2 waveguide.

Here's how this works:

In Mabat's original software, you can 'morph' from a circular throat to a square, circular, elliptical, or rectangular mouth. This led to waveguides that look very similar to oblate spheroidal.

In Mabat's new software, you can add a shape in between. For instance, you can go "circle" to "x-shaped" to "rectangular."

Here's the config file I used to generate this waveguide:


; -------------------------------------------------------
; Horn Geometry Definition
; -------------------------------------------------------

ThroatDiameter = 33 ; [mm]
ThroatAngle = 0 ; [deg]
Coverage_Horizontal = 69; [deg]
; Coverage_Vertical = 40 ; [deg]
Depth = 114.3 ; [mm]
SE_s = 0.7
SE_n = 4.0
SE_q = 0.995
Depth.ConicSectionPart = 0.6
Shape = raw2rect
Shape.FixedPart = 0.5
Shape.CornerRadius = 10.0 ; [mm]

ShapeCurve = SF
SF = 1,1,4,0.8,8,2
SF.Rot = 0

Termination = baffle ; baffle | free-air

; -------------------------------------------------------
; Mesh Setting
; -------------------------------------------------------

Mesh.AngularSegments = 120
Mesh.DepthSegments = 16
Mesh.LipSegments = 6
Mesh.CornerSegments = 12

Mesh.ThroatResolution = 5.0 ; [mm]
Mesh.WallResolution = 25.0 ; [mm]
Mesh.InterfaceResolution = 10.0 ; [mm]
Mesh.IBInterfaceRadiusRatio = 2.0
Mesh.InterfaceStrip = 10.0 ; [mm]

; -------------------------------------------------------
; ABEC Project Setting
; -------------------------------------------------------

ABEC.RadiationConditions = IB ; IB | box
ABEC.f1 = 500 ; [Hz]
ABEC.f2 = 12000 ; [Hz]
ABEC.NumFrequencies = 40
ABEC.MeshFrequency = 1000 ; [Hz]

ABEC.Polars.Distance = 2 ; [m]
ABEC.Polars.Step = 7.5 ; [deg]
ABEC.Polars.Points = 8
ABEC.Polars.Horizontal = yes
ABEC.Polars.Vertical = yes
ABEC.Polars.Diagonal = yes
ABEC.Polars.DiagonalInclination = 0.0 ; 0.0 -> automatic angle for max radius

; -------------------------------------------------------
; Output
; -------------------------------------------------------

;Out.DestDir = "D:\Horns" ; current directory by default

Out.Write_STL = yes
Out.Write_MSH = no
Out.Write_ABECProject = yes

 

[mabat]

Thanks, that's cool!

BTW, this is the original shape that was morphed (see below). It has the property of having a smoothly terminatd OS profile everywhere. It is not an OS WG because it is not circular but it has the same profile everywhere you would cut it through the axis. That's the whole idea behing this tool. By morphing to rectangular it looses this property, at least in the part that becomes morphed - here it is 50% of the depth from the throat that stays fixed (Shape.FixedPart=0.5) - the rest is gradually transformed.

Simply put, this tool connects circle at the throat to an abritrary mouth curve by (smoothly terminated) OS profiles.

 

[mabat]

STL file attached.

 

[mabat]

By making the parameters "SE_*" functions of an angle around axis (now they are set once for all directions), the flexibility would further increase considerably. That's what I'm working on now. That will unleash your creativity

 

[mabat]

This is the formula for the "middle shape" Patrick created: superformula

(Note that this curve intersects the horn surface at 60% of the depth as Depth.ConicSectionPart = 0.6, i.e. it is not directly the shape of the mouth edge. For that the Depth.ConicSectionPart would have to be 1.)

 

[mabat]

This is a nice contest
- Used nothing else than Ath4.

 

[mabat]

Circle, anyone?
Add a rollback and you have an unbaffled M2 (Now this is ugly.)

 

[tmuikku]

Hehe, yeah starting to look like a butt crack Would be ironic if that was the best horn profile there is.

 

[jzagaja]

How does it perform?

 

[mabat]

How does it perform?

I didn't try. Here's the ABEC project -

 

[mabat]

Release update: http://www.at-horns.eu/release/ath-4.4.2.zip

New Parameters added:

Shape.AddMouthWidth = 100.0 ; [mm] - mouth width addition for Shape=raw2rect, default=0
Shape.AddMouthHeight = 0.0 ; [mm] - mouth height addition for Shape=raw2rect, default=0

Shape=raw2circ ; morph mouth to circle

 

[mabat]

M2 contest

This should do the trick:

ThroatDiameter = 33 ; [mm]
ThroatAngle = 0 ; [deg]
Coverage_Horizontal = 40; [deg]
;Coverage_Vertical = 40 ; [deg]
Depth = 114.3 ; [mm]
SE_s = 0.7
SE_n = 4.0
SE_q = 0.995
Depth.ConicSectionPart = 0.6
Shape = raw2rect
Shape.FixedPart = 0.25
Shape.CornerRadius = 2.5 ; [mm]
Shape.StretchExp = 1.5

ShapeCurve = SF
SF = 0.88,0.65,4,0.66,9.2,1.73
SF.Rot = 0

; -------------------------------------------------------
; Mesh Setting
; -------------------------------------------------------

Mesh.AngularSegments = 120
Mesh.DepthSegments = 28
Mesh.LipSegments = 6
Mesh.CornerSegments = 4
...


'Shape.StretchExp' plays a crucial role here. The default is 3 which makes too gradual morphing in this case. It wants more abrupt deformation at the "knuckles". The depth of those is set by 'Shape.FixedPart'.

 

[mabat]

This is the 'raw' shape - this is what you get for 'Shape=raw', i.e. the shape governed by a superformula before the final 'raw2rect' transformation:

 

[turk 182]

so does the "superformula" guarantee good sounding waveguide/horns?

 

[mabat]

ABEC results -
(2 dB per color division; polars à 7.5 deg@2m)

Size of the horn is 340 x 302 mm, throat diameter 33 mm. Not bad at all but I'd say we have seen even better.

 

[Drofdissonance]

I'd say that it shows what the waveguide is meant to do. work really well for a 1.5 inch throat. If you ditch that, you dont need it

 

[mabat]

Anyone is invited to tweak this design for perfection - there are something like fifteen parameters to play with

 

[mabat]

To stay in context, this is an OSWG of the same throat, about the same size and coverage (380 mm / 105 deg), everything else equal -
It's actually wider above 10 kHz. There's more pronounced narrowing in the midrange in the case of circular WG, as we already saw several times -that's where the rectangular mouth seems to come handy.

ThroatDiameter = 33 ; [mm]
ThroatAngle = 0 ; [deg]
Coverage_Horizontal = 105; [deg]
Coverage_Vertical = 105 ; [deg]
Depth = 112 ; [mm]
Depth.ConicSectionPart = 0.5
SE_s = 0.7
SE_n = 4
SE_q = 0.995
Shape = raw
SEExp = 2.0

 

[mabat]

The same sized horn, 1" throat: