Yes pistonic in acoutics book refer to a whole body moving at the same speed (or with the same displacement). It is a model (a simplification of the reality). The panel is for sure bending in its central area.
Here the word "pistonic" says more that all the points of the central area are moving in same direction, not with the same displacement. Probably the important fact is all the points move in the same direction in that area so there is no sound cancellation due to points going in the opposite direction so a full production of acoustic waves from there.
My tentative to put figures on DML points in the direction of your experience Steve: light for the efficiency (at least), flexible to get a high coincidence frequency (my previous words were light but not too stiff). I try currently to understand the effect of damping. I would divide it in 2 or even 3 aspects : the linear damping (independant of the frequency) of the plate, its viscous damping (depending of the frequency), the edge damping (linked to the suspension, the edge conditions).
The directivity measurements show clearly that materials have more off axis SPL than others when the frequency increases which I understand currently as a lack of viscous damping.
When the linear damping increase, the overlap of the modes increases (low Q material). We can suspect there is a kind of optimum (loud thinking). After a certain value, why to increase the mode overlaping? I even wonder if we can speak of modes then. You have explained Steve that you built a panel with 2 glued sheets of aluminum that was very smooth in FR (low Q, high over lapping) but which doesn't sound good.
Christian
Hello Christian.
Damping usually improves measurements by producing a smoothed frequency response and fast short impulse response.
These measurements do not take into account the destructive influence of the damping on the panel sound.
This varies depending on panel materials.
Heavy damping on a heavy panel such as thick ply ,may not have as much effect as light damping on a light panel such as eps.
So generalising about how much damping is very difficult.
Each panel material should be treated differently.
Heavy panels are already heavily damped by their own weight.
The two sheets of aluminium and the glue used to glue the aluminium together produced an over damped panel with super smooth frequency response.
But it sounded boring and uninteresting, similar to a heavily damped cone driver using a mastic coating.
I am also not a fan of heavily damped cone drivers, they also sound boring 😴
I am afraid I also have to include BMRs in this category.
Steve.
Damping usually improves measurements by producing a smoothed frequency response and fast short impulse response.
These measurements do not take into account the destructive influence of the damping on the panel sound.
This varies depending on panel materials.
Heavy damping on a heavy panel such as thick ply ,may not have as much effect as light damping on a light panel such as eps.
So generalising about how much damping is very difficult.
Each panel material should be treated differently.
Heavy panels are already heavily damped by their own weight.
The two sheets of aluminium and the glue used to glue the aluminium together produced an over damped panel with super smooth frequency response.
But it sounded boring and uninteresting, similar to a heavily damped cone driver using a mastic coating.
I am also not a fan of heavily damped cone drivers, they also sound boring 😴
I am afraid I also have to include BMRs in this category.
Steve.
Hi Steve,
Looking at the clip it is quite obvious that the bouncing exciter bounces on the beat/peak of the kickdrum, and the other exciter is not working, or has the capacitor hipass filter (inadvertently?) in place.
If the panel would accelerate/decelerate the exciters and cause this big difference, it would have to have(show) a lot more movement.
Hans
I have in mind the smooth FR you shown. For the IR, I think too but I don't remember IR of over damped panel. Maybe I should make one to measure it? 2 sheets of 2mm plywood PVA glued together may do the job? What do you think?
Most of the measurements show the power response (they include the reflections sent by the room). The problem is to understand the mechanism behind your experience and find the right measurement... If we can!
Yes. I am very surprised by the measurements from the Depron (without any coating) that seems to be a good candidate when the same Depron PVA coated gave bad panels.
This is the challenge: try to find the right figure to describe such behavior.
From physics vocabulary, the weight is not in the damper familly. There are 3 basic famillies : the spring for which the force is proportional to the displacement, the damper for which the force is proportional to the speed, the mass (inertia, weight) for which the force is proportional to the acceleration. Of course, all those forces play "against" the force from the exciter.
I have also this comparison with the damped cone in mind. I wonder if it is the same mechanism? A cone is a mass loaded device for which any additional weight on the cone reduces the bandwidth.
Anyway, it is in the next step to go deeper in the role of the damping.
Christian
Hello Christian.
All those forces play "against" the force of the exciter 😃
I like it. 👍
Steve.
All those forces play "against" the force of the exciter 😃
I like it. 👍
Steve.
So , both you and Eric think I am either a lier or incompetent.
I can not do this test again at the moment as my room is in disarray.
I will come back to this subject when I have Done these tests again.
I think it was tectonic (I'm not sure) that showed how two exciters closely mounted could be affected differently by the panel vibrations.
It is a pity they have removed all their information of dml panels from their site.
But I will come back to this point at a later date.
Steve.
Haha, yes. My mistake. The correct time is shown in the figures. In the example I modeled, the PIP effect was clear at 0.14 ms (milli!), and the modal behavior established clearly at 0.26 ms. Of course, this was with a pretty high driving frequency (11.55 kHz) and at lower frequencies it would take longer.
Eric
I don't think that. We all make mistakes. And it is especially easy when what we observe appears consistent with our internal model.
I would be very interested to see it if you can reproduce it.
Eric
I'm really sorry that I have offended you, I don't think you're a liar or incompetent, but the clip seemed to show so obviously one exciter jumping because of low peak overload, and the other not, that I thought you must have made a mistake with the wiring. That's all. Again my apologies. Hans
Q is sqrt(k*m)/b... where k is the stiffness of any particular mode - I think k was D*Psi^2 in my papers, basically a "modal stiffness" metric. So b is mode-dependent.
I'm not familiar with the nu terminology, but if I had to guess, it's a percentage of the Young's modulus. Damping is often modeled using a complex E term.
Hello Dave,
Thank you for the access to your tool!
Here is a quick feedback of my installation of PETTaLS.
Our laptops run usually under Linux so I did a quick test with Wine but something was missing. I didn't push further.
So I have restarted a previous laptop under Windows (a Lenovo Yoga 2 I3, windows 8...).
No real issue in the installation. Just the need to download some C++ library (VCRUNTIME140.dll). That done, PETTaLS was installed with an icon on the desktop as I requested.
At 1st launch, I was wondering if it was ok due to the long time to launch the Matlab runtime... but yes, it was ok.
On this laptop (I don't remember the screen resolution), I had to move the taskbar usually at the bottom to the left of the display to get an almost full screen. Almost because the top of the titles of the upper plots are under the matlab window top bar. Not a big issue.
A first simulation up to 2k was done in about 3s (it is not a powerful laptop!).
About the GUI, you could think maybe to add a mark in the plot to remind the simulation was limited in modes.
All the fields are in International Units except the Box Depth... maybe a change? Neither a big issue.
It is very easy to change any field in the GUI.
Probably due to my slow laptop and my first use, I didn't find easy to get information on a point of the FR or the IR. It is much more pleasant in the impedance plot. In fact, I won't be able to say how I got the "delete data tips" menu...
I also tested a simulation up to 20k... about 25mn... I understand the 2k by default. It was about 1s per frequency (mode?). Waiting for the plots update, I was wondering the reason... I ran on this laptop my first DML simulation based on the Finite Difference Method with some large arrays and linear system solving, it was slow also but here I imagine it as to loop on the modes, on the frequency range for each (probably on a log scale?) and on the 2D of the plate to get the velocity. All being based on formulas, under Python it is the job of Numpy to work on arrays (matrix)... sorry this is my developer side who think loud!
For the SPL 80dB scale for the SPL is too large. 50dB is more usual.
In the IR plot, I have a change at about 1500Hz like a scale effect (suddenly, the yellow/green level expend on time).
You will have maybe to think to several time scale for the IR. A 200ms is maybe relevant for a 20Hz frequency but not from a auditory point of view. 50ms is already a long time. When we will look at time EQ for DML, the horizon will be in my opinion much below 10ms.
For further test, I will probably wait for a Linux version...
Of course I point here issues or possible improvements. For the rest : thank you again to share all of that with us.
Christian
I added a Linux installer to the webpage. I compiled the app on Ubuntu - hopefully it works on your Linux distribution.
There is apparently a bug where dropdown menus don't work unless you add "export ENABLE_QWEBWINDOW=true" before calling the app. This fix worked for me. The app actually seems to run more quickly in Linux!
Believe it or not, there is only one loop in the code - over frequency. Everything else is 4-dimensional matrix operations and such. I'm surprised that you were able to get it up to 25 mins - must've been a very large and very thin panel, or perhaps just the older computer wasn't happy.
Hi Shido
You'll have to be able to flex your panel to release it from the rigid glass... Is that going to be possible??
Hi Dave:
Thanks for the release.
I messed around with it and have a few questions :
1/ Materials - I figure the pre-sets just prefill the properties boxes and we can adjust them to suit other materials ? ie - start with anything and change to your desired material properties via the boxes. In that case, to make it clear, why not add a custom material which leaves the boxes blank?.
2/ I ran a few compare cycles and after a couple the comps slowed to a crawl (my desktop is pretty quick) - then I found there was no way to abort a run so I had to ctrl-alt-del - stop matlab process? - Is that correct or did I miss something?
3/ I hope the next release will allow selective mixed support cases.
4/ I really hope the next release will allow anisotropy 🙂
5/ It would be good if comparison graphs changed colours.
No doubt more to come both ways
Thanks again
Eucy
Hello Christian
About the perforated materials, I use the Libra:
It is intended to be used in front of sound absortion sheets (you can try whatever material specs you want and, also, vary the distance from the perforated material to the sheets) for best results, although it can be used solely (with about half the absorption rating) as I use in home.
As for appearance it looks like a regular window light blackout (I use it also for that purpose).
@Christian:”The third |document| one enters quite easily in my knowledge field. I recommend to read it. It is really interesting the idea the brain can build a representation of the sound of a room. (…)The third document opens to an other hypothesis which is by its wide radiation, the reverberation field of a DML playing in a room is close enough of the representation build by the brain with other sounds that the DML disapears.”
Precisely, and that`s why I prefer the dml close to the walls (with the toe in they are almost on the corners) just with the Libra taming the back wave and the glass wall doing is job.
@Christian:”For the reverberation as a tool, I have no experience...”
Not with the tool (yet I will dare to say), but tons of listen experience.
it`s never to much to praise the effort of members like you, Valeric or Leob, and a special welcome to EarthTonesElectronics, to find the DML physics objectiveness. With those capable hands in the task, I prefer to put my attention in the psychoacoustic area.
So let´s start form the sound reverberation experience and do a kind of reverse engineering, which lead to:
@Christian: “Have you made tests in this idea |mimic the dml sound in a conventional speaker|? Which kind of sound treatment?”
Since the early days the sound engineers try to capture and/or mimic the natural environments (i,e, mostly reverberation) in the recordings (the plethora of sound capitation – mic positioning - techniques or even room constructions in the studios with “wrong” reverberations far from the scientifical acoustical design studio paradigm is an interesting reading), the most appraise effort was (is for many) and, to my knowledge, the biggest DML speaker ever build, the 1957 EMT 140:
https://www.vintagedigital.com.au/emt-140-plate-reverb/
When I did the first DML setup, the joke turned serious because of the sound propagation characteristics correlated to a reverberation coherence that have the ability to enhance the recorded sound to a listening pleasure.
That reminded me the EMT sound and the quests of sound engineering design (I`m from the time of big multitrack studio tape recorders and exceptional Harrison analog mixing consoles).
For instance, listen from 2.30” to 5.17”:
Don`t we have that flavor in our DML reproduction?
The time maths:
Subtle after the maths (i,.e, the importance of the non important)?
(Just for the joke of it, here is the raw XPS/EPS sound:
)
Now let`s move to the serious listening:
Good, isn`t it? Namely at 15.33” with the two mono channels were panned towards the center (again that flavor in our DML reproduction).
So the importance of the non important is, in Warren Hurt perspective, and for the matter in almost all good sound engineers, the quest for randomness, which is a very hard work based on knowledge. Dml give us that at taste with little to no effort or knowledge, and that was what strikes me.
Now that we have the context of studio techniques to create sound reproduction illusion (so not only objectiveness but psychoacoustic coherence), I can answer to your: @Christian: “Have you made tests in this idea |mimic the dml sound in a conventional speaker|? Which kind of sound treatment?”
I did made tests in house (not treated) and in the studio (very well treated), the next step will be to record one recorded track played by DML in studio and then analyze, by reverse engineering, the differences between them to capture what`s going on.
Good auditions
Rui
About the perforated materials, I use the Libra:
It is intended to be used in front of sound absortion sheets (you can try whatever material specs you want and, also, vary the distance from the perforated material to the sheets) for best results, although it can be used solely (with about half the absorption rating) as I use in home.
As for appearance it looks like a regular window light blackout (I use it also for that purpose).
@Christian:”The third |document| one enters quite easily in my knowledge field. I recommend to read it. It is really interesting the idea the brain can build a representation of the sound of a room. (…)The third document opens to an other hypothesis which is by its wide radiation, the reverberation field of a DML playing in a room is close enough of the representation build by the brain with other sounds that the DML disapears.”
Precisely, and that`s why I prefer the dml close to the walls (with the toe in they are almost on the corners) just with the Libra taming the back wave and the glass wall doing is job.
@Christian:”For the reverberation as a tool, I have no experience...”
Not with the tool (yet I will dare to say), but tons of listen experience.
it`s never to much to praise the effort of members like you, Valeric or Leob, and a special welcome to EarthTonesElectronics, to find the DML physics objectiveness. With those capable hands in the task, I prefer to put my attention in the psychoacoustic area.
So let´s start form the sound reverberation experience and do a kind of reverse engineering, which lead to:
@Christian: “Have you made tests in this idea |mimic the dml sound in a conventional speaker|? Which kind of sound treatment?”
Since the early days the sound engineers try to capture and/or mimic the natural environments (i,e, mostly reverberation) in the recordings (the plethora of sound capitation – mic positioning - techniques or even room constructions in the studios with “wrong” reverberations far from the scientifical acoustical design studio paradigm is an interesting reading), the most appraise effort was (is for many) and, to my knowledge, the biggest DML speaker ever build, the 1957 EMT 140:
https://www.vintagedigital.com.au/emt-140-plate-reverb/
When I did the first DML setup, the joke turned serious because of the sound propagation characteristics correlated to a reverberation coherence that have the ability to enhance the recorded sound to a listening pleasure.
That reminded me the EMT sound and the quests of sound engineering design (I`m from the time of big multitrack studio tape recorders and exceptional Harrison analog mixing consoles).
For instance, listen from 2.30” to 5.17”:
Don`t we have that flavor in our DML reproduction?
The time maths:
Subtle after the maths (i,.e, the importance of the non important)?
(Just for the joke of it, here is the raw XPS/EPS sound:
Now let`s move to the serious listening:
Good, isn`t it? Namely at 15.33” with the two mono channels were panned towards the center (again that flavor in our DML reproduction).
So the importance of the non important is, in Warren Hurt perspective, and for the matter in almost all good sound engineers, the quest for randomness, which is a very hard work based on knowledge. Dml give us that at taste with little to no effort or knowledge, and that was what strikes me.
Now that we have the context of studio techniques to create sound reproduction illusion (so not only objectiveness but psychoacoustic coherence), I can answer to your: @Christian: “Have you made tests in this idea |mimic the dml sound in a conventional speaker|? Which kind of sound treatment?”
I did made tests in house (not treated) and in the studio (very well treated), the next step will be to record one recorded track played by DML in studio and then analyze, by reverse engineering, the differences between them to capture what`s going on.
Good auditions
Rui