The lower part of the throat did not scan well. If there is more interest I can redo this scan. Got some white chalk markers meanwhile. This should improve scanning on black anodized section. But on holidays...so will take ~2 weeks.
I can also scan the upper part that contains the HF diaphragm and is attached to the thorn. Then we can re-design right from the start of the HF diaphragm...
Right, there was no data close the throat in the scan. However, we know that the throat of dcx has a radius of 17.78mm and that it has a zero degree exit angle. I used that information to extrapolate the contour. The last/closest points of the scan to the throat have a distance of 18.5mm to the axis. The spline smoothly connects the this last points to the throat and its derivative is zero at the throat.
[EDIT:] Thanks for the offer. A new/cleaner scan of the dcx would improve the accuracy of the contours for sure.
[EDIT:] Thanks for the offer. A new/cleaner scan of the dcx would improve the accuracy of the contours for sure.
Last edited:
Good job. The whole adapter is different for the 4552ND, I've uploaded a new one (T520-4552-EXT). The starting diameter is 16.3 mm, that's even a tiny bit smaller than for the 4554.
Last edited:
Nice writeup in #16,546 on how to do this 🙂
It's a dang shame that straight impulse inversion, which is so easy, is also the quickest way to overcorrect (or mess up worse.)
A really low tap count can save the day ..Lol
But seriously, if using straight impulse inversion, I'd strongly suggest checking the electrical impulse response of the FIR filter. They can look quite nasty sometimes, despite relatively decent looking acoustic response....makes for a form of noise ime.
And of course look to see if corrections hold up off-axis / off-measurement.....(but that's true for any corrections, even the mildest IIR.)
Just as I was starting to print some parts for a BMS 4552 on a A460G2 my printer decided to poop the bed. Take a dump. Go on the fritz. Gone tango uniform. It broke...
So while I work on that, what are the current preferred slicer settings, infill, perimeters etc for these waveguide parts? There has been mention of using sparse infill and injecting stuff inside the parts for damping purposes but I don't recall seeing anyone actually doing that. ?
So while I work on that, what are the current preferred slicer settings, infill, perimeters etc for these waveguide parts? There has been mention of using sparse infill and injecting stuff inside the parts for damping purposes but I don't recall seeing anyone actually doing that. ?
I've been using a .4mm nozzle, petg filament, 90c bed temp, 245c hotend temp, and 15 percent infill. This means long print times for everything. 21 hours per petal. I think Mabat is using a .6mm nozzle and has print times for a petal of about 7 hours.
@mabat How do you calculate the real and imaginary Impedance at the throat of the Horn? I have no knowledge of AKABAK, so I am not aware if there is any inbuild functions.
This Article (posted by @TNT) from audio express, gives a very interesting inside in how Jörg Panzer does it and he is the author of AKABAK.
I am also curious because your impedance plots look really really great! Usually I don't expect them to be that smooth.
This Article (posted by @TNT) from audio express, gives a very interesting inside in how Jörg Panzer does it and he is the author of AKABAK.
I am also curious because your impedance plots look really really great! Usually I don't expect them to be that smooth.
Can anyone guide me why I get this kind of beaming in the highest frequencies? Is there any fatal error in this configuration?
OSSE = {
r0 = 21
a0 = 60
a = 64.8
L = 17
k = 0.4
s = 1.9
q = 0.9983
n = 2
}
Source.Contours = {
dome WG0 35 11.5 3.5 1 5 2
}
Mesh.ZMapPoints = 0.3,0.2,0.7,0.9
Mesh.ZMapElementSize = 0.1,0.3,0.25,0.85
Mesh.Enclosure = {
Spacing = 15,30,15,100
Depth = 40
EdgeRadius = 10 ; 10
EdgeType = 2
FrontResolution = 8,8,12,12
BackResolution = 16,16,16,16
}
Mesh.Quadrants = 14 ; =14 for 1/2 symmetry
Mesh.AngularSegments = 64
Mesh.LengthSegments = 16
Mesh.SubdomainSlices =
Mesh.ThroatResolution = 5
Mesh.MouthResolution = 8
Mesh.InterfaceResolution = 2
Mesh.RearResolution = 16
Mesh.FrontResolution = 4
ABEC.SimType = 2
ABEC.f1 = 1000 ; [Hz]
ABEC.f2 = 20000 ; [Hz]
ABEC.NumFrequencies = 40
ABEC.MeshFrequency = 1000 ; [Hz]
; --------------------------------------------------------------
; waveguide polars
ABEC.Polars:H_WG = {
SID = 0 ; waveguide -> always SID = 0
Distance = 2
MapAngleRange = -180,180,17
NormAngle = 0
FRDExport = {
NamePrefix = wg_hor
}
}
ABEC.Polars:V_WG = {
SID = 0 ; waveguide -> always SID = 0
Distance = 2
PhaseComp = 2
MapAngleRange = -180,180,17
NormAngle = 0
Inclination = 90
FRDExport = {
NamePrefix = wg_ver
}
}
; --------------------------------------------------------------
Output.ABECProject = 1
Output.STL = 0
OSSE = {
r0 = 21
a0 = 60
a = 64.8
L = 17
k = 0.4
s = 1.9
q = 0.9983
n = 2
}
Source.Contours = {
dome WG0 35 11.5 3.5 1 5 2
}
Mesh.ZMapPoints = 0.3,0.2,0.7,0.9
Mesh.ZMapElementSize = 0.1,0.3,0.25,0.85
Mesh.Enclosure = {
Spacing = 15,30,15,100
Depth = 40
EdgeRadius = 10 ; 10
EdgeType = 2
FrontResolution = 8,8,12,12
BackResolution = 16,16,16,16
}
Mesh.Quadrants = 14 ; =14 for 1/2 symmetry
Mesh.AngularSegments = 64
Mesh.LengthSegments = 16
Mesh.SubdomainSlices =
Mesh.ThroatResolution = 5
Mesh.MouthResolution = 8
Mesh.InterfaceResolution = 2
Mesh.RearResolution = 16
Mesh.FrontResolution = 4
ABEC.SimType = 2
ABEC.f1 = 1000 ; [Hz]
ABEC.f2 = 20000 ; [Hz]
ABEC.NumFrequencies = 40
ABEC.MeshFrequency = 1000 ; [Hz]
; --------------------------------------------------------------
; waveguide polars
ABEC.Polars:H_WG = {
SID = 0 ; waveguide -> always SID = 0
Distance = 2
MapAngleRange = -180,180,17
NormAngle = 0
FRDExport = {
NamePrefix = wg_hor
}
}
ABEC.Polars:V_WG = {
SID = 0 ; waveguide -> always SID = 0
Distance = 2
PhaseComp = 2
MapAngleRange = -180,180,17
NormAngle = 0
Inclination = 90
FRDExport = {
NamePrefix = wg_ver
}
}
; --------------------------------------------------------------
Output.ABECProject = 1
Output.STL = 0
Yes, AKABAK does it. It's the most basic function of a simulator, I would think.
It's because the horns are really good. I guarantee that you would see the same curves in the article if they used the same horns. I understand the article mainly as a comparison of two different numerical solvers/approaches, not so much focused on the DUT itself.
Probably the mesh is already too coarse for that high frequencies.
BTW, I've already measured my TW034 waveguide, and it doesn't behave as predicted, but even at much lower frequencies than I would anticipate. I need to analyze it in more depth because at the moment I don't understand at all what's going on (once again)...
I've finally assembled my A460G2. It went absolutely smoothly with a slow foaming PU glue.
The completion of the full petal circle was literally a matter of a couple of minutes -
It's still without any surface finishing, but it seems I'm still not able to repeat the ultimate midrange smoothness of the A520G2.
This is the BMS 4554 adapter:
A520G2:
A460G2:
The A520G2 is bigger and the mouth termination is quite a bit more gradual in absolute terms. But it's strange that there's virtually no difference in the simulation. Maybe the model has its limitations as well, but I'm more and more suspitious that the warped surface of my A520G2 has someting to do with this. Maybe it's just the missing finish, there are still some tiny holes in the petal joints.
The completion of the full petal circle was literally a matter of a couple of minutes -
It's still without any surface finishing, but it seems I'm still not able to repeat the ultimate midrange smoothness of the A520G2.
This is the BMS 4554 adapter:
A520G2:
A460G2:
The A520G2 is bigger and the mouth termination is quite a bit more gradual in absolute terms. But it's strange that there's virtually no difference in the simulation. Maybe the model has its limitations as well, but I'm more and more suspitious that the warped surface of my A520G2 has someting to do with this. Maybe it's just the missing finish, there are still some tiny holes in the petal joints.
Last edited:
No but the T34A being a hard dome poses no problem for BEM simulation being accurate at least up to 10K.
A460G2 proportions:
A460/520G2 prototypes
A460/520G2 prototypes
Seems unattractive, but probably you would need to print another 520 to verify. If it is actually the sunburn, this would require verification, sunburn should be implemented as a surface pre-print treatment in fusion. I like the looks of the current 460 assembly.
Last edited:
I see, I was wondering the in detail explanation of how they derived the acoustic impedance. I am unfortunately not able to follow how they combine equation 1 and 2 though
What do you mean by midrange smoothness? In my interpretation FR smootheness is similiar. The difference I can discern from these measurements is that the A460G2 seems to have a higher SPDI than the A520G2.
With the velocity fixed the program solves for pressure. What else would you need to know?
To me, smooth means simple. The less is going on regarding the radiation, the better. We're talking differences on the order of 1dB here (or even less), but it's obvious that there's something more going on in the case of 460 at least between 1 - 4 kHz; it's more complex, less smooth.
The polars shown were not taken at the exact same places, I just don't bother at this point. Actually, the radiation patterns of both should be virtually identical (i.e. including SPDI) down to 600 Hz or so.
To me, smooth means simple. The less is going on regarding the radiation, the better. We're talking differences on the order of 1dB here (or even less), but it's obvious that there's something more going on in the case of 460 at least between 1 - 4 kHz; it's more complex, less smooth.
The polars shown were not taken at the exact same places, I just don't bother at this point. Actually, the radiation patterns of both should be virtually identical (i.e. including SPDI) down to 600 Hz or so.
I already thought about taking a hairdryer treatment of the 460 🙂
If it's to make a difference, it should show up.
This is the A460G2 with the EQ kept from the previous A520G2 session (i.e. without any PEQ changes):
(on-axis and some off-axis angle)
A520G2 in comparison (the same driver + adapter):
I suppose it should be fun to figure this out 🙂
(on-axis and some off-axis angle)
A520G2 in comparison (the same driver + adapter):
I suppose it should be fun to figure this out 🙂
Last edited: