PIEZO NXT type panel

[el`Ol]

Not so interesting. The weight / surface ratio even of the lightest version is factor three to high anyway.
This looks MUCH more interesting.
Metalleido Components - Monocore, main page
Wonder what the price is.

[LineArray]

Hi,

independently from the interesting materials
suggested in this thread i am a little suspicious
regarding materials of the

"ultra stiff / ultra low mass/ low damping"

style, when trying to build a high quality
panel type speaker.

Such a material has fast propagation of bending
waves, resulting in a low coincidence frequency
and also a low modal density in the range of
coincidence.

IMO modal density should be as high as possible
when reaching the coincidence frequency.

Otherwise the discontinuities in the preferred
direction(s) of radiation may be audible as
well as the bumpy frequency response.

Especially in combination with a low damping,
such a material simply sounds "nasty".
One has to take countermeasures to tame it
(damping, larger sizing of the panel).

Regarding efficiency those materials may be
very well suited for the application as a panel
speaker. Regarding quality i doubt it.

For moderately sized panels i would prefer materials
of higher modal density (slower wave propagation) and
appropriate damping.

However, by choosing the right thickness of the panel,
the material properties can be compensated to some point.

But there is always a compromise to be found, which
balances efficiency and quality.

[pol_bct]

Hi all,
can you explain this a little more
or give a direction where I can het more infos on this aspect of DML

Originally Posted by LineArray

Such a material has fast propagation of bending
waves, resulting in a low coincidence frequency
and also a low modal density in the range of
coincidence.

IMO modal density should be as high as possible
when reaching the coincidence frequency.

can we overcome the bumping response by multiples points of excitation?

POL

[mkstat]

Originally Posted by LineArray

Hi,

independently from the interesting materials
suggested in this thread i am a little suspicious
regarding materials of the

"ultra stiff / ultra low mass/ low damping"

style, when trying to build a high quality
panel type speaker.

Such a material has fast propagation of bending
waves, resulting in a low coincidence frequency
and also a low modal density in the range of
coincidence.

IMO modal density should be as high as possible
when reaching the coincidence frequency.

Otherwise the discontinuities in the preferred
direction(s) of radiation may be audible as
well as the bumpy frequency response.

Especially in combination with a low damping,
such a material simply sounds "nasty".
One has to take countermeasures to tame it
(damping, larger sizing of the panel).

Regarding efficiency those materials may be
very well suited for the application as a panel
speaker. Regarding quality i doubt it.

For moderately sized panels i would prefer materials
of higher modal density (slower wave propagation) and
appropriate damping.

However, by choosing the right thickness of the panel,
the material properties can be compensated to some point.

But there is always a compromise to be found, which
balances efficiency and quality.

Hi Oliver!

I'm very interested in the physics behind dml operation. I puzzled around a little with numbers (calculating the coincidence frequency/speed of bending wave propagation for different solid materials of different e-modul/mass/thikness) and I agree about the effects you described (observation of a panel with constant size). But the fact that the speed of bending wave propagation is increasing with frequency boggles my mind and I have dificulties in estimating the mode density.

Increases mode density in the panel faster than it would with constant c? I have some problems to get a picture of this in my mind.

I read somewere that above coincidence frequency a rise in efficency takes place (at least for a free radiating panel). It seems that for higher frequencies the panel progessively acts more like a bipole (sound energy from the back is not strongly phase-related to that from the front). I suspect the point where this effect starts is somewhere below or around the coincidence frequency so basically a low coincidence frequency would be maybe desireable. As you noted this makes only sense when mode density and (even more important for the lower frequency range) the mode distribution (mixture of axial and tangential modes) is as even as possible. I think there is some comparability to acoustics in small rooms where the mode density is also very low in the lawer frequency range.

This whole principle leaves a lot of questions open for me - very interesting.

regards
Markus

[LineArray]

Hi Markus,

yes modal density increases faster with frequency,
due to c being frequency dependent.

This effect allows sufficient reproduction quality in
the range > 10 modes/ocatve and above as a rule of thumb.

Modal density is one important quality measure
for a frequency range reproduced by a DML.

As you point out, there is an analogy to
room acoustics. Modal density is one important
quality measure there too.

Concerning the "chaotic bipole" notion, favoured by
some researchers for a free radiating DML panel:

IMO if phase correlation decreases above coincidence
frequency, one could with same justification call
the free radiating DML panel a "chaotic dipole" ...

For practical purposes it is important to note, that
for high frequencies the directivity of a free radiating
DML panel is lower than the directivity of an equally sized
panel which (ideally) radiates in phase over its whole area.

IMO for "wavelength << panel dimensions" this holds independently,
whether you compare our free radiating DML to a single
phase coherent panel of same size radiating as a dipole
or you compare it to two coherent panels, each of them
built face to face into a flat enclosure and acting in
bipole manner.

[mkstat]

Originally Posted by LineArray

yes modal density increases faster with frequency,
due to c being frequency dependent.

This effect allows sufficient reproduction quality in
the range > 10 modes/ocatve and above as a rule of thumb.

Modal density is one important quality measure
for a frequency range reproduced by a DML.

This is what I suspected.
So for long and narrow panels there will be a limitation in modal density, because the width comes into play at a quite high frequency and thus a relatively high usable frequency limit will be the result.
Do you think that controling the wave propagation velocity for width/hight independently by the use of an anisotropic material makes sense? It would be possible to equalize the mode distribution while keeping the physical w/h ratio in a more practical/compact range.

Originally Posted by LineArray

Concerning the "chaotic bipole" notion, favoured by
some researchers for a free radiating DML panel:

IMO if phase correlation decreases above coincidence
frequency, one could with same justification call
the free radiating DML panel a "chaotic dipole" ...

Okay - this helps for my understanding.

Originally Posted by LineArray

For practical purposes it is important to note, that
for high frequencies the directivity of a free radiating
DML panel is lower than the directivity of an equally sized
panel which (ideally) radiates in phase over its whole area.

So a closed back DML has a narrower dispersion?
What happens with the quite strong hf side lobes (between about 65°-80° off axis)? Is this due to "chaotic dipole" behaviour?

-Markus

[LineArray]

Originally Posted by mkstat

...
Do you think that controling the wave propagation velocity for width/hight independently by the use of an anisotropic material makes sense? It would be possible to equalize the mode distribution while keeping the physical w/h ratio in a more practical/compact range.
...

Yes.

Originally Posted by mkstat

...
So a closed back DML has a narrower dispersion?
What happens with the quite strong hf side lobes (between about 65°-80° off axis)? Is this due to "chaotic dipole" behaviour?

-Markus

Below coincidence the open baffle dipole version has narrower
dispersion. Behaviour is similar to a conventional open baffled
speaker.

At coincidence and above i have not investigated this systematically.

But since at coincidence the main portion of radiation is towards the
sides (coincidence lobe), the side (and top, bottom) lobes should be
stronger when using a closed back panel.

Since in a free radiating panel the signs of velocity and pressure
should be inverted between front (we should have "non chaotic"
modes of lower order in this frequency range) and rear, i would
expect the side lobes to be less pronounced or partially cancelled
out in the free radiating version.

(Thereby possibly causing a notch in the overall power radiated at
coincidence ? I have observed that with some of my free radiating
prototypes.)

For frequencies >> coincidence the radiation for the closed back
panel should be narrower, since there is no radiation to the rear
halfroom.

At least i feel you are asking the right questions ... i have no
assured answers based on my own experience.

[LineArray]

Originally Posted by pol_bct

...
can we overcome the bumping response by multiples points of excitation?
...
POL

Hi POL,
if the bumpy frequency response comes from
modal density being simply too low in the
frequency range under question, it cannot be
healed by applying more exciters IMO.

If the bumpy frequency response comes from
a single excitation point having an unbalanced
mechanical impedance over that frequency range,
averaging the impedance using differing excitation
points surely can help.

On the other hand many state that a single
excitation point gives best results in the highs.

Low pass filtering some of the exciters can be
a compromise, but personally i'd like to to get along
without crossovers completely in a panel shaped
bending wave transducer.

[LineArray]

Originally Posted by LineArray

...
Since in a free radiating panel the signs of velocity and pressure
should be inverted between front ... and rear, i would
expect the side lobes to be less pronounced or partially cancelled
out in the free radiating version.
...

That was too sloppy maybe:

I meant velocity and pressure in the air surrounding the panel to
be inverted in direction/sign between the front and rear side of the
panel, when the panel is vibrating in the range of coincidence frequency.

I imagine the nearfield of front and rear side to be mirrored, but with
inverted direction of air motion and inverted sign of pressure.

This is why i think at least partially cancellation of the side lobes
can be expected, when comparing the free radiating panel to a
"back closed" one.

[mkstat]

Originally Posted by LineArray

Hi POL,
if the bumpy frequency response comes from
modal density being simply too low in the
frequency range under question, it cannot be
healed by applying more exciters IMO.

If the bumpy frequency response comes from
a single excitation point having an unbalanced
mechanical impedance over that frequency range,
averaging the impedance using differing excitation
points surely can help.

I absolutely agree to this. It's just possible to excite modes that are inherent to a particular panel (depending on size, w/h ratio, mass and other mechanical properties). There is again the analogy to room acoustics (for the lower modes up to about 5th order, above this mode density is sufficient enough that there will be no perception of discrete frequencies) - the position of the excitation point will determine which modes will be excited more and which one less. If the exciter is positioned in a node (don't mix the term "node" and "mode") position for a particular mode there won't be any exitation (at lest no mode excitation - maybe some pistonic action of the panel, depending on mass). You have to move the exciter position to excite this particular mode or you have to terminate a point on the edge (in the particular direction) of the panel (fixing it) to shift the node in it's position. The problem is that there are many, many modes (hopefully) and it is unlikely to find a position where it's possible to excite all of the lower modes in equal manner at just one point. It's only possible to find a compromise here.
Using more exciters will certainly help to improve the bumpy frequency response, but only if the panel has an optimum w/h ratio and if the exciter positions are optimum according to this particular panel - these positions/size ratios are highly deterministic and that's why NXT uses software to find these.

Originally Posted by LineArray

On the other hand many state that a single
excitation point gives best results in the highs.
Low pass filtering some of the exciters can be
a compromise, but personally i'd like to to get along
without crossovers completely in a panel shaped
bending wave transducer.

I also agree to this.
I've tested various configurations - regarding sound quality it's best to have just one exciter running for the high frequencies. If you're using more than one exciter the practical upper frequency limit is where the distance between the exciters is in the range of lambda (actually lambda/2). Above this frequency serious discontinuities in directivity start which are clearly audible to me. So if you take for example the Dayton exciters (45mm diameter) and space them as close as you can (45mm between the acoustical centers of the exciters) the limit would be about 7650Hz (for lambda/2 3822Hz). I used a single series capacitor on my small test panels (and an active HP filter to prevent over-excursion of the exciters) - this improved sound quality much.

Originally Posted by LineArray

...i am a little suspicious regarding materials of the
"ultra stiff / ultra low mass/ low damping"
style, when trying to build a high quality
panel type speaker.
...Especially in combination with a low damping,
such a material simply sounds "nasty".
One has to take countermeasures to tame it
(damping, larger sizing of the panel).

Regarding efficiency those materials may be
very well suited for the application as a panel
speaker. Regarding quality i doubt it.

I've tested a small (50x30cm) glassfibre - nomex honeycomb material (4mm thickness) which should have near to perfect properties for the use in a DML.
Regarding efficency and hf output this may be a good choice - regarding sound quality I'm not so shure. It has a foggy, blured sound character. I think this is due to massive ringing of the material - to my perception it adds a sort of "short reverb" to the sound. I compared it with a custom made laminate which has lower stiffness, higher weight and much higher internal damping which sounded more open altough it has less hf output. I doubt the ringing of the glasfibre honeycomb can be sufficiently controlled just by damping on the edges of the panel - even for larger panels of the same thikness where ringing should shift to a lower frequency range.

regards
Markus