Since you were able to manipulate the wavefront shape by varying the lengths of the passages and demonstrate it with simulations, wouldn't it be possible to model a phase plug and check out the performance in simulator to see the wavefront, or include as part of the waveguide model? Can we assume the inputs of such phase plug are positioned optimally and the sound enters the slits at same time everywhere (assuming the diaphragm was pistonic)?
I think at least my hf10ak had plastic phase plug that pops out and into pieces without destroying anything. A bit tedious to measure and model it though but might be easier for one off than a planewave tube stuff?
Tried to search images of phase plugs but not much available. Hasty observations of the images I found include most (annular) phase plugs seem to have equal width slot at the input as well as output, but the radiuses at output the output are much smaller of course. Here is about the only exploded image of (perhaps) modern phase plug I found Compression Molding of Multi-Part Phasing Plugs - Mexico - Globe Plastics, Inc.
Last edited:
David, Dmitrij_s, it is easy to play with the ATH script and pretty soon one makes observation that a mouth rollback, a smooth termination, benefits any horn/waveguide profile.
Dmitrij_s I think this is the basis how to do it
JMLC Inspired Horn Calculator | Sphericalhorns
And the problem is it beams pretty bad so I think there is no reason hold on to it since the profile can be better
Dmitrij_s I think this is the basis how to do it
JMLC Inspired Horn Calculator | Sphericalhorns
And the problem is it beams pretty bad so I think there is no reason hold on to it since the profile can be better
Last edited:
Sorry, I'm not interested in a conversation that won't teach anyone a thing.
Sorry, but I'm too lazy to learn mabat's ATH, so I wrote my own tool in Excel/VBA
. I have long wanted to read about the JMLC algorithm, but I did not come across its description.
The mouth rollback is a matter of a smooth curvature change, there's nothing really profound about that. It's obvious that you don't want an abrupt change in the wall curvature anywhere along the profile. I do it by a simple algorithm that first came to my mind, and that works for me. There will be many ways how to achieve basically the same.
Power of the ATH script seen here ATH - Advanced Transition Horns - Gallery , an explosion of profiles. Anything seems possible including JMLC I guess, but we don't need other than the best one for given application do we? There is the real power, ability to make and optimize a profile just for the need and not care about what the name is.
Last edited:
What could be quite useful is a hypothetical component that would somehow average the input wavefront, whatever that would be, and transferred this average in a form of a flat wavefront on its output. Any ideas how to do that?
Perhaps some rigid piston in the way? (that's probably plain stupid)
Or maybe mash the wavefront and sum it more chaotically / run it through a duct with open cell foam...
Last edited:
Or be able to measure the waveshape coming out of existing compression drivers at multiple frequencies and designing the phase plug accordingly, so it converts the wavefront with least damage.
I'll hopefully have some measurements this weekend.
My prints are consistantly failing on the .2mm nozzle, so I'm switching to a .4 on a prusa in my workshop which seems to do the job. About 20 hours remaining
I'll hopefully have some measurements this weekend.
My prints are consistantly failing on the .2mm nozzle, so I'm switching to a .4 on a prusa in my workshop which seems to do the job. About 20 hours remaining
Or maybe to run it through a duct with a set of concentric perforated tubes - only the zeroth mode (i.e. the flat wave) would pass without attenuation. Perhaps filled with foam as well. The tubes could even seamlessly continue as the vanes of the "spherical wavefront plug".
A wavefront purifier
Last edited:
A theoretical question: Would it be possible to get the exit wavefront modal decomposition, as we talked about in the past (the first few axial modes), but from steady state measurements in a closed chamber placed over the driver? We could of course still simulate the chamber when driven with the individual axial modes alone. I'm just not sure about the summing. We concluded it's possible in a free field situation with a waveguide but honestly that's still too much of a hassle... With a small chamber with several microphones fixed in the walls it would be much more practical.
Yes, that's what I meant, that the "waveguide" starts at the diaphragm and that the phase plug should be part of that, but if it is short enough then it is less important.
As to:
Of course its possible, but making the right choices is design would be key.
I see no reason why you could not simulate any situation mode by mode, with mode as defined at the throat intersection. The question is: is this surface solid or not, as seen by the model. That would have to be looked at. But then you take a field measurement as close as possible to the models assumptions, like free-field etc. This field measurement can then be decomposed into the constituent parts, i.e the modes at the interface. - math not too complicated.
Another complicated assumption is that the modes are orthogonal and not coupled. This may not be true, but it's certainly the first assumption to make. If the first model turns out to not work so well then the modes can be tested for independence.
What inner shape of the chamber would you suggest to begin with? A duct section of the same diameter as the throat, or something different?
Unfortunately I have no idea how to look at that. But I think it sees it as solid. That would make it unusable, right, because that's not what's happening.
Last edited:
I don't follow your top question.
I thought about it and it would look like a tube coupled to a volume creating a resonance. If the tube is short enough then its a simple mass - until it's self resonance. Modeling this would be fairly important since it represents the delay of the reflection from the diaphragm. Without it and you might get erroneous results making the analysis less stable.
I thought about it and it would look like a tube coupled to a volume creating a resonance. If the tube is short enough then its a simple mass - until it's self resonance. Modeling this would be fairly important since it represents the delay of the reflection from the diaphragm. Without it and you might get erroneous results making the analysis less stable.
I'm glad this topic has come around again and mabat is handling this very well. There is something I wonder.. where the exponential part meets the conical section, will there be a reflection due to the change in expansion, or no reflection according to the proper wavefront.. or option 3, it depends on what the driver is able to feel?
My initial thought was that it may be possible to model it without the driver internals (diaphragm, phase plug), which is not true as I see, so that somewhat loses its attraction.