PIEZO NXT type panel

[mikadosan]

Elmers Black is nothing but just a normal polystryrene board with black paper skins. Of course, it would be interesting to test it if it is more stiff than other polystyrene boards available.

[jzagaja]

I use hot glue when I want remove exciter without stress. Can you confirm that on glass windows sound is somehow artificial like with an echo?

[LineArray]

Hi jzagaja,

i guess there is the same effect as with all underdamped
structures ...

Regards

[mkstat]

Just some info if anybody is interested in DML simulation:

http://www.randteam.de/_Docs/Aes104%...on%20model.pdf

The contained skript runs on Akabak.

Download - Akabak

Manual: http://home.comcast.net/~rhcamp0217/...21_Main_US.pdf

There is an syntax error in the 3rd line - "**" (Bending stiffnessBs) instead of "B"

Wow! - the censoring algorithm works great here!

regards
Markus

[sled108]

Well I finally got these to Sound Good...First, Round the corners and cover the edges with Black Tape. That takes care of any Buzzing. Then, remove the drivers from the Spider Cases...With the Daytons this is easier than Shucking a Clam....Next, attach them to the Panels with double sided tape. I used some that I use for covering windows with plastic for the winter. This Stuff will stick them on for Keeps. Finally, you need to Crank the Bass up and the Treble down a Click...They will play Bass if it's Cranked. These sound quite nice now from a Sure TA2024 Amp...And Remarkably Loud to Boot...All for a Total of $21....

[LineArray]

Originally Posted by mkstat

Just some info if anybody is interested in DML simulation ...

Hello Markus,

i did run the script, but there was something which makes me wonder:

The element types "DMPanel" , "Exciter" , "DMCoupler"

are neither part of the documentation nor are they integrated into
the forms of the "Def" menu, are they ?

So those DML Elements in Akabak 2.1 seem to be undocumented
builtins ?

Maybe Jörg Panzer had already integrated some DML modeling into
that early Akabak 2.1 version, since he already worked with NXT.

It is a very nice tool to check out some basics concerning panel and
exciter paramers and the effect of both on the behaviour of a DML.


But since i feel, the description of those elements mentioned above
is hidden, it is not a good base for further
understanding and discussion and maybe future refinement and
customization of the model ...

e.g. simulation of panels which have different stiffness dependent on
direction cannot be simulated with that "DEF_DMPanel" function.

I am sure that current software from Panzer/NXT will have evolved.

For those who are interested in DML modeling, this early AkAbak version
may serve as a starting point for pre selecting materials, panel sizes
and so on.

Kind regards

[mkstat]

Hello Oliver!

You are right - there is obviously no documentation of the DML components included in the standard AkAbak Manual - I have not found any further information.
I just posted the link of that specific Manual version as it gives the following information on its' first page:

"NXT has selected AkAbak to be the base for its software front end 'NXT Designer'. The latter is part of the NXT licensee package and incorporates the DML components and tools for panel design. AkAbak is developed in parallel with the NXT Designer, i.e. both packages are compatible. The only differences are the DML components."


The manual gives information about the structure of the script language so that anybody who is interested can at least get an idea how to modify this DML script to his needs. I have no idea if there are more DML functions included in Akabak or not.
There is no doubt that the capabilities of this example script are very limited and some things are not taken into account. It is certainly not the last step in Joerg Panzers modelling approach - actually it is one of his first steps.

It's definitely not my intention to start a discussion on this topic - I just wanted to share what I have found out - that there is a tool you can play with - for free. Maybe it can help to get more insight in this technology as the interactions of the design parameters lead to effects that can not be easely captured in an intuitive way, as one might think. For me it definitely cleared up some things.

with kind regards
Markus


PS. I'd like to share some more information i found - for those who are interested:

http://www.randteam.de/_Docs/Aes105%...Simulation.pdf
http://www.randteam.de/_Docs/Aes106%...Enclosures.pdf
http://www.randteam.de/_Docs/Aes107%...k%20Solver.pdf

Some infos about Joerg Panzers projects in the past (from the R&D Team Website):

"AkAbak for DMLs This special implementation of AkAbak is able to simulate the response of multiple exciters on a plate. The plate was assumed to be infinite and some assumption about the radiation were made. This project was done for Mission UK (1996).

PanSys This software simulates the fully coupled system: Lumped elements for the exciter and electronics, modal analysis for the plate, the enclosure and the radiation. PanSys is very powerful in prediction and a particularly useful tool for the design of rectangular plate speakers. The code-base of PanSys is huge and it took us several years to develop (1997).

AFR-Designer simulates the radiation and motor parameter of a fully coupled motor-plate-acoustic loudspeaker driver. The model is based on modal-analysis and axis-symmetry. AFR-Designer makes use of Vacs as output device. This approach dramatically reduced the cost of development because the design concentrates only on the implementation of the mathematics and on the input part. AFR-Designer was designed after specification for Nxt (2003)."


Some info out of the PanSys Manual (this was posted some years ago in the Visaton Forum by somebody who has the NXT System Designer Software):

"The material properties of a panel control the modal distribution of the plate. They are directly linked to the coincidence effects of the acoustic radiation. They describe the material properties of a ready-made panel, which is cut-out in a certain direction (see Panel materials - Width/Height stiffer chapter for details).

The input parameter for the panel material, which PanSys needs are not generic, such as Young's modulus etc. Rather effective parameter of ready made and cut-out panels are used, such as area density and bending rigidity. It has been found that this approach yields more realistic and realisable results than generic parameter. This is especially true for layered panels.

A simulation is only as good as the quality of its input parameters. This is, of course, especially true for the panel parameters. Therefore PanSys comes with an extensive panel database. These parameters are determined and confirmed using a scanning laser interferometer together with a modal identification software tool. As the identification tool uses the same model as implemented in PanSys, this method yields the best fitting parameters.

At very low frequencies the total mass of the plate together with the compliance of the exciter-suspension and supports determine the rigid body resonance frequencies. A panel’s first rigid body mode can be a piston motion as with the conventional loudspeaker. The next possible rigid body mode allows the plate to rotate. If all sides of the plate have Free boundary conditions (FFFF) the first bending mode is close to f0, which is the fundamental engineering parameter of a DML. However, the real first bending mode is shifted because of the coupling to components and to the acoustics. In FFFF mode the on-axis response of the DML is controlled by the first rigid body mode only. The bending modes support the off-axis response. However, in practise the coupling to components masks this ideal effect.

At low and medium frequencies the plate is in pure bending and the vibration pattern is determined also by the plate stiffness in x- and y-direction and some other material constants. The finite size and mechanical conditions at the edges of the plate enforce a field of standing waves, or, so-called bending modes. At higher frequencies shear wave fields add to the bending wave field. The bending wave field is dispersive, whereas the shear wave-field is not. In this frequency range the thick plate theory is applied and gives reasonable good results.

At very high frequencies the plate and its components are vibrating as an elastic volume. At very high frequencies the implemented plate theory stops to be valid.

The simulation model is based on the Kirchhoff-Mindlin plate theory. In principle this theory is valid for flat, thin and homogeneous plates. Any curvature increases dramatically the resonance frequencies of some of the modes due to coupling between bending and membrane effects.

The plate must be thin enough to ensure that the front and rear surfaces of the panel move in parallel. As a rule of thumb, the thickness should be smaller then a sixth of the bending wave-length.

Homogeneous, means that section parameters are constant along any direction. In the transverse (z) direction this constraint is not satisfied by usual sandwich constructions. However, since there is assumed to be no wave field in the z-direction, the different material parameters of each layer are lumped into so called section parameters, such as the effective bending rigidity of the panel.

The panel can be stiffer in one direction than the other, which is taken into account by the two components of the bending rigidity, D1 and D2. If D1 = D2 then the plate is said to be isotropic, otherwise an-isotropic. In the context of the program, D1 and D2 are assumed to be aligned with the edges of a rectangular cut-out. Panel parameters are therefore always annotated by the degree under which the panel was cut-out (for example: 0/90 or +/-45 degree).

In a plate, the bending wave field cannot be completely separated into independent fields in x and y direction. This is because there is coupling resulting from shear-mechanics and lateral/direct strain. Coupling is controlled by the in-plane-shear (Shr) and the Poisson ratio parameters, which are also dependent on the way the panel is cut-out.

Although the panel is assumed to be infinitely thin with respect to transverse waves and with respect to diffraction, there are some important effects related to the thickness, which modify the bending wave field. The thicker the plate, and the higher the frequency, the more influential the effects of finite transverse shear rigidity and the inertia of rotation of a section. The controlling parameters are called “thick plate parameter” (see thickness, t, and shear modulus, Gz).

In addition to the bending and shear wave-fields, there is a local elasticity at any driving point. The effect of elasticity can be observed at high frequencies where the local stiffness goes in resonance with the mass of the voice-coil and the mass-type reactance of the aperture effect.

At very high frequencies (usually around 10kHz) the driving point mechanics become very complex. With the help of the in-build plate model extensions it is possible to go beyond the limits implied by the usual plate theory and to display all effects involved. However, in the upper frequency band the uncertainty of the result-curves increases with frequency."

[mkstat]

Btw - there is the parameter AbsorbCoeff missing for the Radiating component in the example script.

It's possible to simulate a closed back DML with it (I guess) when it is set to 1. Default setting is 0 for radiation from both sides.

Syntax:

|Radiating component

DMPanel 'DM1' Def='DM1' AbsorbCoeff=0
x=0 y=0 z=0 HAngle=0 VAngle=0


regards
Markus

[LineArray]

Originally Posted by mkstat

Btw - there is the parameter AbsorbCoeff missing for the Radiating component in the example script.

It's possible to simulate a closed back DML with it (I guess) when it is set to 1. Default setting is 0 for radiation from both sides.

Syntax:

|Radiating component

DMPanel 'DM1' Def='DM1' AbsorbCoeff=0
x=0 y=0 z=0 HAngle=0 VAngle=0


regards
Markus

Is it possible that 'AbsorbCoeff' models the
inherent mechanical loss of the panel ?

I do not know, maybe your interpretation is right.


Kind regards

[mkstat]

Originally Posted by LineArray

Is it possible that 'AbsorbCoeff' models the
inherent mechanical loss of the panel ?

I do not know, maybe your interpretation is right.


Kind regards

I don't know.

When i inspect the directivity plot in the 360° mode i suppose it's the absorbtion of the radiation. It's possible to use values between 0 and 1 for "AbsorbCoeff".

regards
Markus