Correct. The shape of the curved wavefront or "bubble" at the horn mouth is automatically calculated and taken into account as part of the isophase wavefront model. This means that the overall horn length used in the calculations (the 'virtual length') is longer than the actual physical axial length.
When you asked if it was possible to manually add the bubble, I thought that you were referring to actually specifying the bubble in AkAbak, which cannot be done.
The small increase in the length of each segment affects the "transformation" of the mouth impedance back down the horn to the throat, not the value of the mouth impedance itself. I guess it could be argued that the overall increase in length defines the bubble to the extent that it represents the height of the bubble, but that is all. To calculated the mouth acoustical impedance, the overall shape and surface area of the bubble are required.
That's because Hornresp simulates multiple-segment horns using the plane wavefront model.
Actually I was asking just that. When I asked what it would look like, I meant what would the manual correction look like (for example adding an extra flare at the end of the horn or adding a bit of length to each segment.) Just to be really clear here, the former would not work, the latter would work?
Interesting. So multi segment horns in Hornresp use the Webster model like Akabak? The vast majority of the horns we sim and see posted would be multi segment horns. All tapped horns and most flh sims too.
What about compound horns? If there is a single segment on one end and a multi segment horn on the other end, does it use different models (isophase and plane) on the different horns?
What about ports when Ap and Lpt are used? That's a single segment but not a horn.
Should I already know this stuff? I searched the Help file quickly and didn't see any of this documented.
Yes, the bubble is the wavefront shape, that's what I understood. My Q is, how does the bubble detach from the confines of the opening it is leaving? The edges hit an abrupt discontinuity, what then?
I do not know what anything below "moderately high power output" is..we're talking about bass, yes? The whole intent is to stun small animals and create visual blurriness..
No nausea, drinks cost too much..I make the assumption that there is a distance(perhaps the bubble definition here, where further distance will feel the effect of the edge discontinuity. I was wondering if the edge effects can be tailored to stall that onset.
Hey, if we can figure out how to modify that discontinuity to make a virtual flare, we can be rich I tell ya...
jn
To echo Guy, for most practical TH's then the same planar model is used in both HR and AkAbak? In AkAbak the last segment can be a Horn radiator element which allows for non planar directivity calcs based on curvature of horn mouth.
An externally hosted image should be here but it was not working when we last tested it.
"The edges hit an abrupt discontinuity, what then?" Then what is the question. Once the wave exits the mouth what happens?
I am saying once the wave leaves the mouth we don't have to worry about it (other than possibly correcting for the bubble length at the mouth), the diffraction is the only thing we need to examine.
Other people are saying that the 60 inch square boundary itself is enough of a waveguide to add length to the horn. I don't agree with this at all.
Yes, at the boundary edge the wave will feel the effect of diffraction. The picture above is a picture of diffraction. Diffraction will have an effect similar to adding horn length but that's not what's happening, it's just a solid boundary and should be treated as such.
As far as I'm concerned there is no such thing as a virtual flare.
Out to roughly mouth dimensions in the forward direction, the pattern is still affected by the edges.
For me, I've seen no convincing argument one way or the other.
The edge will impact the entire velocity pattern. The Q for me is, does that impact further alter expansion, or moreso, can the impact be used to tailor the expansion.
Your entitled to your opinion. I've not seen enough to form an opinion either way.. But enough to form questions..
edit: I believe pictures of diffraction effects with wavelengths a small fraction of the opening is not consistent with the discussion.
jn
Last edited:
Maybe a series of pics will help. You can let me know at which step you disagree.
A multi segment 2 pi sim in Hornresp or Akabak would reflect this physical situation. A horn mouth flush with the ground, which is considered to be a single infinite boundary. Note there is no bubble drawn in, apparently there is no bubble included in the sim.
An externally hosted image should be here but it was not working when we last tested it.
Now a 1 pi sim in Hornresp (Akabak can't do 1 pi). A horn mouth flush with the dual boundary junction (despite the fact that it's physically impossible to build this correctly). Both boundaries are infinite size and no bubble included in the sim.
An externally hosted image should be here but it was not working when we last tested it.
Hopefully the 1 pi horn looks larger, it should. I don't know what's going on mathematically but the 1 pi sim of a single cab looks exactly the same as a 2 pi sim of dual cabs. Shown is a random tapped horn sim, the black line is 1 pi, the light grey line is a 2 pi sim in which the multiple speakers tool was used to show 2 cabs in parallel (double the cabs and double the power). If you can't see the grey line it's because the two examples are a perfect match and overlay perfectly. David rejected the term "reflection" earlier, but it does look a lot like a reflection at least in terms of 1 cab in 1 pi = 2 cabs in 2 pi with twice the power.
An externally hosted image should be here but it was not working when we last tested it.
What needs to be done to get an accurate sim of a boundary extension is to simulate the response of a finite boundary to show the effect of the boundary extension. A diffraction model does this, it takes into consideration the size and shape of the boundary and also the location of the sound source in relation the boundary edges. The boundary provides a forward directivity and increased radiation resistance, same as the 1 pi sim above but only to the waves small enough (frequencies high enough) to be affected by the boundary size, and the sim will show rippled response due to reflections (red arrow) and interference (different path lengths shown with green arrows). Still no bubble, no bubble is considered by the sim.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
That's what Xrk did, a normal sim but with diffraction added to reflect the 60 x 60 inch front face. Note the remarkable similarity between his sim and the Danley measurement. I take this as proof that a diffraction sim is already pretty close.
The only thing left to consider is the bubble. Today we learned that none of our multi segment sims in Akabak or Hornresp consider the bubble anyway, so we could ignore it.
Or we could consider it. I'm not sure how big the mouth is exactly on the Danley horn, but based on a 20 inch diameter the bubble would be 6 inches forward of the mouth. Due to the increased radiation resistance provided by dual boundaries (the ground and the horn boundary) it might be considered in the sim to be a bit more than 6 inches.
What we would not do is consider the bubble to extend all the way to the boundary edges, the boundary (180 degree waveguide) has no ability to contain the wave against the expansion of being released into open space in a way that could be considered a horn flare.
I think this matches what's going on a lot better than vortexes and virtual flares made of air.
The plane waves going through the hole are clearly much larger than the hole, they narrow as they go through the hole and then expand in a curved wavefront. If the waves were much smaller than the hole they would pass right through without touching the edges and would continue unaffected by the hole.
Last edited:
I have just realised that when we have been talking about adding length to a segment, I didn't explain that this is done in the isophase model by dividing the segment up into a number of smaller elements, each of which has a curved throat surface and curved mouth surface. When all the element lengths are summed together, the resulting total axial length is longer that the actual physical horn length. The curved mouth surface of the final element becomes the infamous "bubble".
The overall geometry is similar to that shown in the attachment - assume the green and red lines are individual elements, and that the final green line at the mouth is the bubble.
Getting back to your statement above, and taking into account my further clarification, hopefully it now makes more sense why simply adding an extra flare at the end of the horn will not work.
Correct. That's why the power response, diaphragm displacement, electrical impedance, etc, results are the same. The pressure response results will normally be different because the directivity models used in the two programs are different.
Yes.
The port is assumed to be a cylindrical tube, so the plane wavefront model is used. The isophase model defaults to the plane model anyway, when the cross-sectional area remains constant.
No. In the very early versions of Hornresp, the user had the option of selecting either the plane or isophase wavefront model, but I removed this feature after receiving a number of queries, and instead let Hornresp automatically make the decision in the background - less confusing that way. What the user doesn't know, won't hurt them
.Attachments
Last edited:
Correct.
Correct. Hornresp uses a different method to calculate directivity, which does not require the user to manually specify the horn mouth wavefront curvature.
To illustrate the above, attachments 1 to 4 show the throat acoustical impedance for a 100Hz exponential horn, where:
Attachment 1 shows the throat acoustical impedance when Cir = 1, calculated using the plane wavefront model.
Attachment 2 shows the throat acoustical impedance when Cir = 1, calculated using the isophase wavefront model.
The different between the plane and isophase results is minimal.
Attachment 3 shows the throat acoustical impedance when Cir = 3, calculated using the plane wavefront model.
Attachment 4 shows the throat acoustical impedance when Cir = 3, calculated using the isophase wavefront model.
The different between the plane and isophase results is now significant. The plane wavefront model has broken down and the results have become inaccurate. The isophase results are tending asymptotically towards the infinite case (light red and grey traces) as Cir increases, which is the correct outcome.
Attachments
Last edited:
It should. The math is identical (or it should be). We do this with magnets as well. Also, electrostatics does the exact same thing as well.
Again, the diffraction models shown uses wavelength significantly smaller than the openings. It assumes planar waves approaching the opening as well. David's jpg, while also using smaller wavelengths does show some of what I'm thinking of..
I believe the drawing you provided will create reflection at the boundary everywhere except at the hole. There will be energy crowding at the orifice, due to the boundary reflection increasing the pressure variations.
Thanks David. I note also the wavelength is significantly smaller than the mouth, but see no reasonable way to have the point made with lower frequencies.
Of significance to me is the shape (curvature) of the fronts as they leave the confining walls in your model. Note that the sim has no curvature within the mouth, and the further it travels, the more non-planar the wave gets. I would suspect this volume of space, where edge effects have not yet been eliminated, may be what Danley is thinking of. edit: If you look carefully, the area where the boundary effect goes away is somewhat triangular in shape. If you were to put two horns side by side, the midplane conditions mirror, and the triangular boundary effect area will extend to twice the length. This aspect is entirely consistent with the general knowledge that adding more and more horns extends the low frequency response. I'm not saying it's cause and effect, but that it is consistent with current knowledge.
edit: I do feel however, that at lower frequencies, the model needs to go non linear. Pressure/rerafaction at the edges cannot be symmetrical when wavelength approaches mouth size.
edit 2. I note you also do not include boundary effects within the horn. I would have expected curvature to start as a consequence of the wave travelling against the expanding boundary.(I see now that you do indeed have that, your isophase mode)..
Again, thanks guys, a really nice discussion.
jn
Last edited:
Again, no it doesn't. I made the plane waves approaching the exit bright red so you can see how big they are compared to the opening. They narrow as they go through and then expand on the other side. If the waves were small they would go right through the hole and be completely unaffected. (Or hit the wall beside the hole and bounce back the other way.)
David showed the exact same thing, as the large wave passes the boundary it wraps back around. I put a couple of curved red lines on the picture on the right to show the wrap around at the edge. This is all simple diffraction.
Diffraction is what happens at the edge. The finite boundary itself provides extra forward directivity and radiation resistance. This is all captured well in a diffraction sim.
No it's not, the waves are larger than the mouth.
If you still don't agree put up a pic of what you think a large wave going through a hole looks like.
Last edited:
In your drawing, the energy which hits the wall will reflect back. At the hole, it passes through. At the hole edges, the pattern changes due to edge/reflection effects on the drive side of the boundary.
You are confusing width with wavelength. Wavelength is the distance from center of red to center of red in David's simulations.
Diffraction happens even when the wavelength is smaller than the hole. It may be argued that is where it matters more - hence desire to make features around a dome tweeter very smooth and flush and have rounded baffle edges not sharp. Diffraction of larger wavelengths with features of size of hole or baffle don't matter as much because hole or feature is smaller than wavelength. Not the other way.
Width is a necessary component of wavelength. If you take Hornresp's Wavefront Simulator (the pic he showed) and increase the frequency, the waves stop wrapping around the edges, they proceed with full forward directivity in the forward direction.
Width is a necessary part of a 2 dimension simulation. It has nothing to do with the definition of wavelength.
Decoupling of the boundary conditions is a part of similitude.
Again, your previous posts confuse width with wavelength.
jn
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.