Real Expert or Just Self Proclaimed

[MJK]

Originally Posted by MJL21193

Some "basic text on acoustics" wasn't going to give me the explanation I wanted and wanted to discuss; after all, Martin King himself couldn't adequately explain it.

I am not sure that is true. I just decided not to continue in the discussion. I did not believe you were making an effort to understand, you were just playing word games for your own amusement.

[MJK]

Originally Posted by godfrey

No, I think lumped parameter modeling is great for sealed boxes, reflex systems and the like.

It's just with transmission line enclosures (and horns, for that matter), where the internal path length is long, that transmission line modeling is more appropriate (using T/S parameters for the driver, of course).

Sometimes the boundary can be a bit blurred. For example a very tall reflex box would be better modeled as a transmission line, but I gather those sorts of enclosures are referred to nowadays as "mass loaded transmission lines" anyway.

While tapered transmission lines and horns look quite different and have separate groups of enthusiasts, the same software modeling techniques ought to be applicable to both.

It would be interesting to compare results between e.g. Hornresp and Martin's TL software when faced with the same speaker design.

The blurred boundary comment is an interesting observation. One could consider the lumped parameter sealed box and ported box enclosures as just simplifications of TL modeling. A routine that can predict behavior of a TL enclosure can also model a closed or ported box, but not the other way around. A routine that can model a TL can also model a horn, or you could say that a horn calculator can also model a TL if stuffing is ignored.

All of these enclosures can be modeled by the same math and therefore you don't need to be concerned about boundaries of applicability between models. If you are using one of the simple T/S based models, then you really need to be aware of the limits of applicability with respect to geometry and frequency.

Once you step up to a 1D wave equation TL style of model, then you need to be concerned about the other two dimensions and potential standing waves not accounted for across the TL (side to side or front to back) at higher frequencies. The mitigating factor is the stuffing, it will help suppress the modes you are not accounting for in your model so your results will probably still be accurate.

With respect to Hornresp, Augspurger's TL program, AJHorn, and my MathCad worksheets; a few years ago I ran tests of my MathCad models against these three other programs. If the same problem was formulated and run the results matched very closely, physics is physics. Each program has its own set of strengths and weaknesses so the user needs to determine what they want to accomplish and then pick the right tool. The accuracy of model results is driven by the knowledge and skill of the user.

[scott wurcer]

Has anyone tried this? COMSOL : Acoustics Module

I have access to a copy (it's very expensive). This does full 3D FEM on arbitrary geometries and it does multi-physics so the thermal effects at LF can be included.

[pkitt]

Not intending to minimize or disagree with the point you're trying to make, Ed, but your description of the second graph is not quite correct IMO. The graph shows the system response with the red line, the outputs of the driver and terminus/port combined in both magnitude and phase, whereas the dashed blue line is how the system response would behave in an infinite baffle arrangement. You need to show the 8th graph in the series to accurately show how the driver and terminus/port individually respond where the red line is the driver and the dashed blue line is the terminus/port. At least that's my thoughts, and if I'm all wet, I apologize.
Paul

Originally Posted by Ed LaFontaine

Just to move this along:

Vifa P17 Mass Loaded Quarter Wave designs








The above simulations use the same parameters for a subject driver. The top sim is intended to make use of room gain to arrive at a "flat" response in room. In my experience this response is less subject to "boom" due to room modes. YMMV

The bottom sim is an exercise in attaining a "flat" response from the baffle. The area under the red curve, above the blue dash line, is the gain of the system from the terminus.

[andy19191]

Originally Posted by scott wurcer

Has anyone tried this? COMSOL : Acoustics Module

I have access to a copy (it's very expensive). This does full 3D FEM on arbitrary geometries and it does multi-physics so the thermal effects at LF can be included.

It does linear acoustics, can couple to structures to an extent and can include some thermal effects. It does not simulate the full physics in particular the nonlinear fluid motion is not handled which limits the coupling to structures, acoustic dissipation and nonlinear effects, etc...

[PB2]

Originally Posted by scott wurcer

Has anyone tried this? COMSOL : Acoustics Module

I have access to a copy (it's very expensive). This does full 3D FEM on arbitrary geometries and it does multi-physics so the thermal effects at LF can be included.

Yes of course FEM is the modern compute, and purchase cost intensive solution. However, one should study the classics for insight into the nature of different well known alignments. I would welcome having an FEM program available but would probably still mostly use T&S type simulators where they apply. Non-linear modeling is another story.

[PB2]

Originally Posted by MJK

The blurred boundary comment is an interesting observation. One could consider the lumped parameter sealed box and ported box enclosures as just simplifications of TL modeling. A routine that can predict behavior of a TL enclosure can also model a closed or ported box, but not the other way around. A routine that can model a TL can also model a horn, or you could say that a horn calculator can also model a TL if stuffing is ignored.

All of these enclosures can be modeled by the same math and therefore you don't need to be concerned about boundaries of applicability between models. If you are using one of the simple T/S based models, then you really need to be aware of the limits of applicability with respect to geometry and frequency.

Once you step up to a 1D wave equation TL style of model, then you need to be concerned about the other two dimensions and potential standing waves not accounted for across the TL (side to side or front to back) at higher frequencies. The mitigating factor is the stuffing, it will help suppress the modes you are not accounting for in your model so your results will probably still be accurate.

With respect to Hornresp, Augspurger's TL program, AJHorn, and my MathCad worksheets; a few years ago I ran tests of my MathCad models against these three other programs. If the same problem was formulated and run the results matched very closely, physics is physics. Each program has its own set of strengths and weaknesses so the user needs to determine what they want to accomplish and then pick the right tool. The accuracy of model results is driven by the knowledge and skill of the user.

I agree and as I mentioned used hornresp to model a vented system with a long (30") port here. And of course a horn with no flare is a TL. It did not seem to be completely accurate but I believe that no losses are included:
AudioKarma.org Home Audio Stereo Discussion Forums - View Single Post - AK Design Collaborative - Insubnia

[Ed LaFontaine]

pkitt,

I appreciate your comments. Together, we'll get it right. I've added them to the things I'll review this weekend.

I'm looking forward to diyIndiana. Hope to see you there.

[soongsc]

Originally Posted by scott wurcer

Has anyone tried this? COMSOL : Acoustics Module

I have access to a copy (it's very expensive). This does full 3D FEM on arbitrary geometries and it does multi-physics so the thermal effects at LF can be included.

It seems there is a lot of work involved with the setup. Please let us know what you think.

[gedlee]

Originally Posted by scott wurcer

Has anyone tried this? COMSOL : Acoustics Module

I have access to a copy (it's very expensive). This does full 3D FEM on arbitrary geometries and it does multi-physics so the thermal effects at LF can be included.

I think that most of the FEA programs will handle acoustics. The problem is the cost and the learning curve. They use ANSYS here in China where I am at the moment and it does a good job, but of course they ripped off the software and they spent endless hours learning to use it. Then the problem is that it's not a flexible model when its done. It's A model, of A single design. Very detailed for that specific design, but not very extensible in terms of changing the model substantially. FEA has its good and bad points, but I still think the same way about it today that I did 40 years ago when I used it in my thesis: its too detailed and cumbersome for rapid development. Its fine if you are trying to squeeze out the last few pennies from a design, as would be done in automotive or the like, where millions of the same product are made. But I have hardly used it since my school days.

I replaced the ANSYS models here with a T-matix model which is far less detailed, but detailed enough, and runs instantly. It will even search hundreds of designs for the best one - something that FEA would have real trouble with.