Over on my 'Square Pegs' thread, the subject of BEM came up*, and it seemed like an interesting subject that may be worth a fork.
I don't know much about the subject, but if anyone has anything to contribute (software, models, methods, results) please do!
To get started, here's some results created by Richard Taylor with Acousto (AcouSTO - Acoustics Simulation TOol - Homepage)
Data taken from Taylors site here: LX521 Simulation: First Steps | Richard's Stuff
* http://www.diyaudio.com/forums/multi-way/217298-square-pegs-61.html
I don't know much about the subject, but if anyone has anything to contribute (software, models, methods, results) please do!
To get started, here's some results created by Richard Taylor with Acousto (AcouSTO - Acoustics Simulation TOol - Homepage)
Data taken from Taylors site here: LX521 Simulation: First Steps | Richard's Stuff
* http://www.diyaudio.com/forums/multi-way/217298-square-pegs-61.html
Great! I have not tried that software, but have also been intrigued by Richard's work on the Lx521 baffle. I had hopes that he would take it to the next step and make a baffle with higher WAF with similar polar response 🙂
I have played around with Abec3, which combines lumped element and BEM. Its a bit rough on the UI side, but has great potential. Where I got stuck there was when trying to use imported mesh files (from gmsh). I couldnt get the logic on how to define the boundries in a "solid" mesh (i.e not zero thickness), which would be the output from my CAD software.
A full Synergy horn model in Abec3 would be great 🙂
I have played around with Abec3, which combines lumped element and BEM. Its a bit rough on the UI side, but has great potential. Where I got stuck there was when trying to use imported mesh files (from gmsh). I couldnt get the logic on how to define the boundries in a "solid" mesh (i.e not zero thickness), which would be the output from my CAD software.
A full Synergy horn model in Abec3 would be great 🙂
Hi Guys,
I have been using ABEC/2/3 for about 5 years now, partly professionally (I am a professional acoustician ) and partly for my personal interest in horns.
The learning curve is indeed very steep (took me about a year to really get going nicely), but the end results can be very rewarding.
My experience is that the models can be very accurate IF you input a good model, for this you need to be very aware of what's important and what is not, if you want to model a complete synergy horn you can do this fantastically with LEM (Akabak scripts) combined with the BEM engine.
I found that importing meshesh isn't (currently) the best way to work. I create XL sheets wich generate horn profile data and model parameters.
When I started learning ABEC the examples were sparse, however since version 3 there is a good collection of nice examples to play with, I really encourage you to download them and play around a bit (disclaimer; I will not be held responsible for insomnia..😉 ).
A lot of the stuff I created is paid for by somenone else so I can't share this.
At the moment I am at a very busy period, but when life calms down I will try to make a cleaned version of my worksheets so you guys can learn a bit from them.
Kees
I have been using ABEC/2/3 for about 5 years now, partly professionally (I am a professional acoustician ) and partly for my personal interest in horns.
The learning curve is indeed very steep (took me about a year to really get going nicely), but the end results can be very rewarding.
My experience is that the models can be very accurate IF you input a good model, for this you need to be very aware of what's important and what is not, if you want to model a complete synergy horn you can do this fantastically with LEM (Akabak scripts) combined with the BEM engine.
I found that importing meshesh isn't (currently) the best way to work. I create XL sheets wich generate horn profile data and model parameters.
When I started learning ABEC the examples were sparse, however since version 3 there is a good collection of nice examples to play with, I really encourage you to download them and play around a bit (disclaimer; I will not be held responsible for insomnia..😉 ).
A lot of the stuff I created is paid for by somenone else so I can't share this.
At the moment I am at a very busy period, but when life calms down I will try to make a cleaned version of my worksheets so you guys can learn a bit from them.
Kees
Kessito, great tip about Excel. I dissected the H400 example, used all the elements from there, but replaced them with my spreadsheet outputs. That exercise made me understand a bit more about how Abec3 works. Note: A bit, not a lot. 🙂
I next step would add the midrange entries to the Node structure. It should be doable, as they are on a flat surface. Just more time consuming than drawing it with a mouse.
Question: The example uses a lumped element model of a compression driver. The output of this is transferred to the BEM model via the Driving section in the solver. I understand that the output from the compression driver itself through the horn takes into account the small waveguide buiilt into the driver. I´m not sure, however, how that would work when I add midrange drivers to the horn sides. In other words, will the reflections from the midrange drivers bounce off the element "one can see" in the model (BEM only), or will they go deeper, into the compression driver (in the LEM model) before they bounce back? I suppose what I am asking is if I should remove the "compression driver waveguide" from the LEM-system and rather include it in the BEM system to ensure that the midrange radiation gets reflected from the right point.
Another question: It seems to solve impossibly slow (as in I couldn´t see ANY progress) when raising the MeshFrequency from 2khz to 3khz. Any ideas on why? Does Abec simply give up if there is not enough memory available or something? I am running the software on a win7 through parallels and have set aside 4gb mem.
Looking forward to seeing your worksheet examples! Perhaps you even have a template for a smooth round-to-square transition? 🙂
I next step would add the midrange entries to the Node structure. It should be doable, as they are on a flat surface. Just more time consuming than drawing it with a mouse.
Question: The example uses a lumped element model of a compression driver. The output of this is transferred to the BEM model via the Driving section in the solver. I understand that the output from the compression driver itself through the horn takes into account the small waveguide buiilt into the driver. I´m not sure, however, how that would work when I add midrange drivers to the horn sides. In other words, will the reflections from the midrange drivers bounce off the element "one can see" in the model (BEM only), or will they go deeper, into the compression driver (in the LEM model) before they bounce back? I suppose what I am asking is if I should remove the "compression driver waveguide" from the LEM-system and rather include it in the BEM system to ensure that the midrange radiation gets reflected from the right point.
Another question: It seems to solve impossibly slow (as in I couldn´t see ANY progress) when raising the MeshFrequency from 2khz to 3khz. Any ideas on why? Does Abec simply give up if there is not enough memory available or something? I am running the software on a win7 through parallels and have set aside 4gb mem.
Looking forward to seeing your worksheet examples! Perhaps you even have a template for a smooth round-to-square transition? 🙂
Attachments
do you have a absorbing boundary for free space coupling? - BEM really solves for the "modes" in the volume, subject to the boundary conditions so the absorbing boundary is critical to radiation problems
then you have to figure out how to calc transient from a series of frequency solutions if you think such matters...
then you have to figure out how to calc transient from a series of frequency solutions if you think such matters...
Last edited:
In the H400 example, the following are used:
1. An "edge" around the horn mouth, with normals pointing away form the horn throat)
2. An interface baffle, extending out from the edge in 1), and "closing the horn" (i.e. normals pointing into the horn space).
3. The driving element seals off the throat of the horn.
And there is also
4) A baffle rear, with the size of the horn mouth (but shifted 1mm towards the throat), with normals pointing towards the throat.
Aparently 4) is a trick to avoid creating the outside of the horn.
So it seems the horn boundaries are well sealed off through 1, 2 and 3. If you could explain how 4) works, I would be grateful. I also don´t quite understand why the five-sided bounding box ("Interface baffle") needs to have volume and not just be a flat element separating the sub-domains.
EDIT: Didn´t quite understand your second sentence - could you please elaborate?
EDIT2: Any explanations about how BEM works are very welcome! 🙂
1. An "edge" around the horn mouth, with normals pointing away form the horn throat)
2. An interface baffle, extending out from the edge in 1), and "closing the horn" (i.e. normals pointing into the horn space).
3. The driving element seals off the throat of the horn.
And there is also
4) A baffle rear, with the size of the horn mouth (but shifted 1mm towards the throat), with normals pointing towards the throat.
Aparently 4) is a trick to avoid creating the outside of the horn.
So it seems the horn boundaries are well sealed off through 1, 2 and 3. If you could explain how 4) works, I would be grateful. I also don´t quite understand why the five-sided bounding box ("Interface baffle") needs to have volume and not just be a flat element separating the sub-domains.
EDIT: Didn´t quite understand your second sentence - could you please elaborate?
EDIT2: Any explanations about how BEM works are very welcome! 🙂
Attachments
Last edited:
Ok, it runs at 3khz now. I changed f2 to 4000. Seems that if mesh frequency and top frequency of the simulation are equal it doesn´t work.
This shows the directivity of just a driver.
Compare with
Electro-acoustic models
for the case b = a.
Compare with
Electro-acoustic models
for the case b = a.
Attachments
ABEC 3 Help
Hello everyone. I am fairly new to both ABEC3 and AKABAK. Would anyone be able to give me some guidance in modelling a horn?. I have a compression driver with me. I need to model a Horn/Waveguide for it. However, I do not have some essential driver parameters. My plan is to measure them.
*Should I model the driver in Akabak first?
*Is there a way to import the frequency response of my driver to ABEC?
*Is there a better way to model the horn for my driver?
Please excuse me if my questions are not clear. I can explain in detail. Any help would be greatly appreciated.
Hello everyone. I am fairly new to both ABEC3 and AKABAK. Would anyone be able to give me some guidance in modelling a horn?. I have a compression driver with me. I need to model a Horn/Waveguide for it. However, I do not have some essential driver parameters. My plan is to measure them.
*Should I model the driver in Akabak first?
*Is there a way to import the frequency response of my driver to ABEC?
*Is there a better way to model the horn for my driver?
Please excuse me if my questions are not clear. I can explain in detail. Any help would be greatly appreciated.
* BUMP * Me too. I know HORNRESP, recently got interested in Unity | Synergy horns, and now have my mitts on AKABAK and have ABEC3 and have zero use on either. Which to try first? ABEC3 said to be nearly 100% compatible with AKABAK; however I want to sim with the great minds here the easiest ways possible 🙂
Sent from my NV570P using Tapatalk
Sent from my NV570P using Tapatalk
If you look in various synergy threads their will be Akabak models posted already. It seems difficult at first but is very logical and useful once you get past the basic interface and script. I suggest trying to learn Akabak with provided examples in manual and package for basic speakers like a sealed, a reflex, and a horn first. Then dive into a synergy. Abec3 requires a solid model of your system - not easy if you are not adept at CAD.
- Status
- Not open for further replies.