It is not actually free at all. If it is used in a profit generating project, the author expects the user to purchase a license. The author has graciously and generously made the software available to the DIY community for use at no charge for non-commercial activity. So it is very far from an open source software program (as in GNU GPL)...
Since I am not attempting to generate any money from my design activities, I am "free" to use it at no charge.
But I understand your point. As a user, I must view VituixCad as a "black box". I can test the inputs/outputs against known science/math, and I can compare actual measured results against the software predictions. I have done this, and so far the software works as expected. But the actual secret sausage making that happens inside the box is not visible to me.
Right, so what did your findings say about the contribution off the quality or issues of those cabinet resonances and radiation relative to the total sound quality of the loudspeaker design as a whole?
Also, a lot of this information can also be found in plain old books and papers about acoustics?
..and with VituixCad - it's Arta for measurements (for easier off-axis IR (impulse response) transfer to VituixCad).
It is a bit steeper learning curve though.
For something a bit easier to start with look to the Basta (and REW or even Sirp). ..and you could even use a UMIK-1 with REW for ease of use, but remember that it is a bit more limited (UMIK-1) though still useful for a starting point or for quick easy IR measurements.
Here are my thoughts on mic.s and sound cards though note that Clio Pocket does apparently have a software timing reference for proper off-axis phase result (..that I was unaware of when writing what I have in the link below), fundamentally it's still limited with respect to near-field distortion testing though:
https://www.diyaudio.com/community/threads/arta-and-usb-interface.367198/page-4#post-6537908
Last edited:
You can obviously measure a driver in a standard test box for acoustic comparisons between drivers and determine the T/S of the driver alone. When designing a speaker system you can use the T/S to design the volume and tuning of a reflex cabinet. As far as the drivers actual response in the designed box the system as a whole needs individual driver measurements done in the designed box you want to use. Reasons are baffle width/ driver placement on the baffle, physical spacing between drivers and differences in the acoustic centers as examples. These all are important.
So yes use the basic driver T/S parameters to design and build a prototype and take measurements. Use these in box measurements to engineer the crossover in simulation software. As far as need advanced math and physics can't hurt but good software does the heavy lifting. For example with LEAP there are extensive tutorials that get you up to speed. There is no substitute for time using the software.
Of course you could find cabinet design issues such as resonances where you could make mods as required. Also with cabinet are you going to use round overs to reduce diffraction are the drivers flush mounted are you going to use a stepped baffle to help align acoustic centers and so on.
Rob
Last edited:
There was a time when it was the only way, and necessity would drive us forward. Is that the difference..
https://www.diyaudio.com/community/threads/geddes-on-waveguides.103872/post-2630197
However even now, there is no tool for certain parts of a design. In certain areas and in certain cases it is necessary, and a person can't limit themselves to what is available.
FEM isn't necessarily the ultimate tool, not in it's present form. It often gives a difficult to discern result since it doesn't differentiate the wanted from the unwanted. How can it, since the tool is being used for lack of that very information?
More often it is to give a yay or nay, a litmus test. Some blindly use it to empirically teach themselves how to manage diffraction, but this puts the horse after the cart.
More often it is to give a yay or nay, a litmus test. Some blindly use it to empirically teach themselves how to manage diffraction, but this puts the horse after the cart.
I suspect you're understating the importance of diffraction, since after the initial response shaping is done, the question turns to 'how should this case differ from the obvious smooth curve?'
Not sure I understand your point. Are you saying that because FEM generates values everywhere throughout the solution region and requires the user (or software) to perform an extra processing step to get a subset of that information into a form that is useful to evaluate what is currently of interest this is a poor feature?
For one thing, I'm not typically seeing it used to display only diffraction rather than the primary wavefront, not that it necessarily could. I'd question it if that was claimed. The same could be said about differentiating between wanted and unwanted diffraction.
Also we often see it used at only a single frequency, which limits the usefulness.
Also we often see it used at only a single frequency, which limits the usefulness.
If you can define it then it can be calculated from the FEM solution field and displayed (subject to assumptions involved in the simulation). Assuming your "primary wavefront" (vague term) is the fundamental mode (something precise) then the energy in that and the other modes could be evaluated and displayed. This brings in physics and maths that are part of the way engineers view things but not necessarily hobbyists. Perhaps calling it the fundamental mode rather than primary wavefront might help?
For a linear system calculating a set of frequencies can be substantially faster than calculating over a long enough period of time to get sufficiently accurate low frequency information and with a sufficiently small time step to resolve high frequency motion. An even more efficient approach for a linear system if applicable is to calculate a set of modes and work with those. Even if the modes vary with frequency as they do with viscoelastic materials an approximated modal approach can still be the most efficient (hence my comments about free FEM software above). What makes computational efficiency important for those of us simulating complete loudspeaker systems at home on our desktops is keeping the memory size to a few gigabytes and run times to less than a few hours and preferably only a few minutes. There is no difficulty performing time varying simulations but large models with small numerical errors will consume a lot of computer time if reasonably accurate low frequency information is required.
There is no distinct fundamental mode in a waveguide.
Instead what I often see is cabinets with not so distinct blobs in front of them.
I can answer them but no conversation was taking place so I moved on.
The objective is to bring the radiated sound down to inaudible levels. This is doable without much of a stretch for most designs if one goes about it in a methodical evidence based manner which detailed numerical simulation facilitates. I don't know how to evaluate total sound quality and presume it to be some audiophile notion? Sound radiation from cabinet designs that are not too silly is usually at a low level and not particularly intrusive as people have mentioned earlier. Pushing it down to inaudible levels is a detail that helps distinguish high quality designs from average ones. This matters to some hobbyists but not to others. It is a hobby.
The engineering knowledge to go about things easily would be around that of a graduate that had taken a course or two in acoustics and vibration. Not aware of much literature on loudspeaker cabinets specifically but there are many text books on sound and vibration in solids and sound radiation from solids. It is basic stuff for an engineer in a relevant field but less so for non-engineers although scientists, mathematicians,... that think in a similar rational scientific way could swiftly pick up the parts they are unfamiliar with. The challenge is getting it across to those that dropped the science stuff at school as soon as they could and don't view things in a "scientific manner".
Fluid flow is nonlinear but if you introduce the notion of sound then some form of linearisation is taking place which leads onto various ways to decompose into modes. One needs to be careful and precise about what one is referring to when it comes to modes but they do exist at least to the extent sound exists outside mathematics.
Blobs of what? Most numerical schemes will have numerical errors in the form of diffusion and dispersion which will smear a sharp wavefront. Increased resolution will reduce these errors to whatever is considered acceptably small. Fancy schemes can track the front and eliminate any numerical error associated with it but they are rare.
You are talking about a structural FEM to assess cabinet radiation. @AllenB is, I believe, talking about boundary element models which assess the radiation pattern of a driver diaphragm and a waveguide/baffle shape.
Thanks Jim, I might have missed that.
Andy, when the primary wavefront of a waveguide can be described by a single mode, the device is called '1P'. Few specific examples are that ideal.
This post may help to describe 1P from the beginning of the waveguide - https://www.diyaudio.com/community/threads/geddes-on-waveguides.103872/post-1877500
..however the termination adds more issues.
Andy, when the primary wavefront of a waveguide can be described by a single mode, the device is called '1P'. Few specific examples are that ideal.
This post may help to describe 1P from the beginning of the waveguide - https://www.diyaudio.com/community/threads/geddes-on-waveguides.103872/post-1877500
..however the termination adds more issues.
I am talking about all numerical methods. Allen is referring to simulating sound in waveguides which I understood hence the comments about modes. Compressible flow can be decomposed into modes like in structures but there are several ways to go about it because fluid flow is nonlinear and acoustic waves are only one of several in a fluid. Don't know how much of this is involved in whatever Allen is referring to but simulating sound in something like a waveguides is fairly common.
This appears to be something like geometrical acoustics with jargon. Hard to say without the details. What generates the wavefronts with blobs?
Modal analysis can't be used with a typical acoustic BEM code where a surface is discretised but it can with typical CFD, CAA or acoustic FEM/FVM/FDM/... codes where a volume is discretized. The latter can be preferable when things get complicated or the BEM computational size gets too large. The optimum is often a coupled FEM/BEM if both the near and far acoustic fields are of interest.
@andy19191
I find the conversation more and more confusing by the minute.
It started from your point that BEM/FEM was essential in (any) professional design.
Now all of a sudden you're saying it goes beyond DIY and it's just an hobby?
How do you know these levels are inaudible and where is the boundary?
The basic relationships can be easily understood by anyone who is able to follow just a logic way of thinking.
You most definitely don't have to be a graduate for that.
However, when designing professional speakers I might hope that the engineer in question has at least a basic understanding of those things.
In fact, everyone who spends his/her time into BEM/FEM simulations needs this basic knowledge to understand what is BS within the simulations or not.
Without this knowledge, how is that any different from someone who just goes to town with a certain approach, by just making a cabinet extra heavy and thick?
Is it overkill? Yes
Is it effective? Probably
Is it easy to implement and build? Yes
Result is the same?
I find the conversation more and more confusing by the minute.
It started from your point that BEM/FEM was essential in (any) professional design.
Now all of a sudden you're saying it goes beyond DIY and it's just an hobby?
Can you provide what those levels are?
How do you know these levels are inaudible and where is the boundary?
I don't know that all depends on the skill and knowledge of the reader.
The basic relationships can be easily understood by anyone who is able to follow just a logic way of thinking.
You most definitely don't have to be a graduate for that.
However, when designing professional speakers I might hope that the engineer in question has at least a basic understanding of those things.
In fact, everyone who spends his/her time into BEM/FEM simulations needs this basic knowledge to understand what is BS within the simulations or not.
Without this knowledge, how is that any different from someone who just goes to town with a certain approach, by just making a cabinet extra heavy and thick?
Is it overkill? Yes
Is it effective? Probably
Is it easy to implement and build? Yes
Result is the same?
Perhaps I failed to make my previous point about waveguides not having a single discrete mode of interest.. a room, perhaps. (Not to mention the modal complexity increasing continually as the wavefront evolves.) If I understand you, and I'm not sure I do, we're going in circles.
No I don't think we have understood each other which is perhaps not surprising given our different backgrounds. I am interested though in what you consider a mode to be because I suspect it might be different to what I would consider a mode to be (i.e. the modes in the compressible flow in aeroengine compressors and turbines). Is it some form of geometrical construct coming from Earl as suggested in your link? Is it defined somewhere in a mathematical way that you can reference?