Someone in Canada had created one, but there is no further information available.
And, some additional information from another source.
And, some additional information from another source.
Last edited:
@EarthTonesElectronics
Hello Dave,
Thank you for the videos you posted on Youtube. I watched them; here is my feedback.
Interface and ergonomics : It is very clear, pragmatic and seems easy to use. My only suggestion would be (if not already done!) to have the materials and the exciters in some separate text files to be easily extended or customized. As a Linux guy, I would appreciate your simulator works on it (emulation with Wine is a possibility depending on what language is currently supporting it). If necessary, I could find some Windows no more in use on my computers.
Exciters/Materials : For them, I was wondering which degree of models you have.
For the material, I have no doubt that some viscous (linear?)damping is used, is there a visco-elastic damping in the model (influence on the HF)?
About the exciter, its inductance and the voice coil mass are supposed according to papers about DML to create some low pass filters, in combinaition with the electrical resistance and the mechanichal impedance. We have no direct evidence in this thread of that. Is it part of the exciter model (with the spider stiffness and losses)? Or maybe according to your experience neglectable.
Exciter position/velocity map : The velocity map is of first importance to understand what happens. The exciter position graph is more an information to see where there are. In my tentative of simulation (FDM with python), both exciter position and mode shapes, were on the same graph... in case you want to save space to add other information (I have one suggestion below). At the screen, we see those graph scales are up to 0.4m. Our panels can have one dimension up to 1m or even more. Is the scale automatically adapted?
SPL graph : Here, I am not sure of which SPL it is... The on axis response? Some energy one, which is more common speaking about DML? I will come back later on the post on that. The SPL scale is large (120dB). We are used to share the frequency response on a 50 to 60dB range with a 5dB (major)/1dB(minor) grid. You have introduced on this graph the coincidence frequency (I am happy to see what I introduced recently here about it and the directivity plots make sense) and something suspected here but not as clear as you explain it about the HF limit due to the voice coil diameter
Directivity plot : I discovered the power of this plot some weeks ago (an the possibility for a DIYer to get it). This is really a big plus of the tool. As I just read your thesis and some other papers about the way to get the pressure response (I have experimented the Rayleigh integral in my FDM tentative), seeing the polar plot in the video (no null at 90°), I understand it is the response in a half space, the DML being on an infinite baffle. Is my understanding correct? If yes, it is important to explain it is not the response that a DIYer will get with an open back DML (which is the majority of the designs here). In addition, I am not sure to understand if it is possible to get this response in practice. Maybe with a fully supported panel (CCCC or SSSS) for which some back load might be possible but for some variant around the FFFF, it might be far from what it is possible in practise. I am maybe wrong here, please, tell me.
Some suggestions of improvement:
Impedance curve : Eric already suggested it. To have the impedance curve would be a plus as it is very easy for a DIYer to measure it. It would be a good way to check how close the simulation is from the reality or to adjust the parameters for some new material or exciter.
Directivity index
There are strong (solid?) design rules for loudspeakers coming from the work of F Toole and others that lead to a straight on axis FR and a smooth directivity (to say it short). Of course DML were not part of the loudspeakers tested but I don't think they are different enough in their behavior to escape to that.
We have now tools based on DSP that work well enough to be include in the audio chain to customize the FR. The point is it modifies the FR in all the directions.
So currently I am in the idea that the most important target is the control of the directivity, not the FR. Of course, the FR can't be anything but we have a certain margin on it where there is no way (as far as I know) to modify afterwards the directivity.
I am even more in this idea after watching your video where free edge panels (with the limitation of a possible half space representation and no dipole effect) seem to show less lobes than supported edges panels.
My suggestion here is to add a directivity index (difference between the mean SPL in the listening window and the +/-90° window, see B Zenker's paper or humbly my paper here about the coincidence frequency and the directivity plots) to have the possibility to optimize in directivity... nothing says for now this optimisation is possible.
Christian
Hello Dave,
Thank you for the videos you posted on Youtube. I watched them; here is my feedback.
Interface and ergonomics : It is very clear, pragmatic and seems easy to use. My only suggestion would be (if not already done!) to have the materials and the exciters in some separate text files to be easily extended or customized. As a Linux guy, I would appreciate your simulator works on it (emulation with Wine is a possibility depending on what language is currently supporting it). If necessary, I could find some Windows no more in use on my computers.
Exciters/Materials : For them, I was wondering which degree of models you have.
For the material, I have no doubt that some viscous (linear?)damping is used, is there a visco-elastic damping in the model (influence on the HF)?
About the exciter, its inductance and the voice coil mass are supposed according to papers about DML to create some low pass filters, in combinaition with the electrical resistance and the mechanichal impedance. We have no direct evidence in this thread of that. Is it part of the exciter model (with the spider stiffness and losses)? Or maybe according to your experience neglectable.
Exciter position/velocity map : The velocity map is of first importance to understand what happens. The exciter position graph is more an information to see where there are. In my tentative of simulation (FDM with python), both exciter position and mode shapes, were on the same graph... in case you want to save space to add other information (I have one suggestion below). At the screen, we see those graph scales are up to 0.4m. Our panels can have one dimension up to 1m or even more. Is the scale automatically adapted?
SPL graph : Here, I am not sure of which SPL it is... The on axis response? Some energy one, which is more common speaking about DML? I will come back later on the post on that. The SPL scale is large (120dB). We are used to share the frequency response on a 50 to 60dB range with a 5dB (major)/1dB(minor) grid. You have introduced on this graph the coincidence frequency (I am happy to see what I introduced recently here about it and the directivity plots make sense) and something suspected here but not as clear as you explain it about the HF limit due to the voice coil diameter
Directivity plot : I discovered the power of this plot some weeks ago (an the possibility for a DIYer to get it). This is really a big plus of the tool. As I just read your thesis and some other papers about the way to get the pressure response (I have experimented the Rayleigh integral in my FDM tentative), seeing the polar plot in the video (no null at 90°), I understand it is the response in a half space, the DML being on an infinite baffle. Is my understanding correct? If yes, it is important to explain it is not the response that a DIYer will get with an open back DML (which is the majority of the designs here). In addition, I am not sure to understand if it is possible to get this response in practice. Maybe with a fully supported panel (CCCC or SSSS) for which some back load might be possible but for some variant around the FFFF, it might be far from what it is possible in practise. I am maybe wrong here, please, tell me.
Some suggestions of improvement:
Impedance curve : Eric already suggested it. To have the impedance curve would be a plus as it is very easy for a DIYer to measure it. It would be a good way to check how close the simulation is from the reality or to adjust the parameters for some new material or exciter.
Directivity index
There are strong (solid?) design rules for loudspeakers coming from the work of F Toole and others that lead to a straight on axis FR and a smooth directivity (to say it short). Of course DML were not part of the loudspeakers tested but I don't think they are different enough in their behavior to escape to that.
We have now tools based on DSP that work well enough to be include in the audio chain to customize the FR. The point is it modifies the FR in all the directions.
So currently I am in the idea that the most important target is the control of the directivity, not the FR. Of course, the FR can't be anything but we have a certain margin on it where there is no way (as far as I know) to modify afterwards the directivity.
I am even more in this idea after watching your video where free edge panels (with the limitation of a possible half space representation and no dipole effect) seem to show less lobes than supported edges panels.
My suggestion here is to add a directivity index (difference between the mean SPL in the listening window and the +/-90° window, see B Zenker's paper or humbly my paper here about the coincidence frequency and the directivity plots) to have the possibility to optimize in directivity... nothing says for now this optimisation is possible.
Christian
Last edited:
@EarthTonesElectronics
Dave, as I was thinking more about the high frequency limit you show in the SPL plot (the blue line), I have had a deeper look to the math behind.
Below is an extract of your thesis showing the wave length in the panel (8.18 below) and the coincidence frequency (8.20). Combining both it comes if we search the frequency at which the exciter diameter is equal to one wave length in the panel :
d_exciter = C / sqr(fh.fc) with C the speed of the sound in the air, fh the frequency at which the wave length is the exciter diameter (HF limit), fc the coincidence frequency.
As you explain it in the videos, this seems to have an optimum when fh = fc = 20kHz (maybe a bit arbitrary at this step) so d_exciter = 17mm.
I won't conclude for now that a 19mm exciter is the best choice due to its low BL compare to a 25mm one and maybe other parameters but it could give a direction for a concentrator diameter... Having it a bit smaller would allow a higher fc.
HvdZ (@HvdZ ) concentrator is a 16mm diameter at the panel interface. Excellent!
Loud thinking!
Christian
Dave, as I was thinking more about the high frequency limit you show in the SPL plot (the blue line), I have had a deeper look to the math behind.
Below is an extract of your thesis showing the wave length in the panel (8.18 below) and the coincidence frequency (8.20). Combining both it comes if we search the frequency at which the exciter diameter is equal to one wave length in the panel :
d_exciter = C / sqr(fh.fc) with C the speed of the sound in the air, fh the frequency at which the wave length is the exciter diameter (HF limit), fc the coincidence frequency.
As you explain it in the videos, this seems to have an optimum when fh = fc = 20kHz (maybe a bit arbitrary at this step) so d_exciter = 17mm.
I won't conclude for now that a 19mm exciter is the best choice due to its low BL compare to a 25mm one and maybe other parameters but it could give a direction for a concentrator diameter... Having it a bit smaller would allow a higher fc.
HvdZ (@HvdZ ) concentrator is a 16mm diameter at the panel interface. Excellent!
Loud thinking!
Christian
@homeswinghome Just a quick note that these are great suggestions and ideas for discussion. I’ve already incorporated quite a few upgrades into the model. I’m on a trip this week and only have wifi on my phone, so I can’t type a response that would do your questions much justice. Before I left, my most recent task was incorporating a more complex model for sandwich panels as suggested by @Veleric. I promise I’m not ignoring these messages - I’ll have more updates when I’m back next week!
I was reflecting on the concept of resonating areas, a notion introduced by Stanley Rich in the early 1960s, which @spedge is quite familiar with. This principle is surely employed by both Sonance and Stealth Acoustics in their high-end in-wall speakers, and most probably by Amina Sound, even though they never mention that anywhere.
An intriguing aspect is that covering a "normal" speaker with a resonating panel (soundboard) effectively disperses the sound emitted from that speaker, creating an auditory experience that makes it difficult to pinpoint the source of the music. Since most people have "normal" speakers at home, this experiment can be conducted quite easily.
At the outset of the experiment, one may perceive that the sound from the covered "normal" speaker is somewhat subdued. However, this perception largely depends on the type of resonant board (soundboard) used in the setup.
Another intriguing aspect is that if one incorporates holes into the resonating board (soundboard), the resulting sound may exhibit a different, more pronounced bass quality. This effect likely depends on various factors, such as the size and placement of the holes, as suggested by the accompanying high-end speaker from Japan.
--------------------------------------------------------------------
I have been searching for Sonance and Stealth Acoustics literature that might reference their ideas on resonance boards for sometime. When one keeps on digging, one finds. Here are some of them:
https://patents.google.com/patent/US7292702B2
https://patents.google.com/patent/US8611575B1
https://patents.google.com/patent/US20130156243A1
https://patents.google.com/patent/US9008342B1
https://patents.google.com/patent/US9313570B1
https://patents.google.com/patent/US10587949B1
An intriguing aspect is that covering a "normal" speaker with a resonating panel (soundboard) effectively disperses the sound emitted from that speaker, creating an auditory experience that makes it difficult to pinpoint the source of the music. Since most people have "normal" speakers at home, this experiment can be conducted quite easily.
At the outset of the experiment, one may perceive that the sound from the covered "normal" speaker is somewhat subdued. However, this perception largely depends on the type of resonant board (soundboard) used in the setup.
Another intriguing aspect is that if one incorporates holes into the resonating board (soundboard), the resulting sound may exhibit a different, more pronounced bass quality. This effect likely depends on various factors, such as the size and placement of the holes, as suggested by the accompanying high-end speaker from Japan.
--------------------------------------------------------------------
I have been searching for Sonance and Stealth Acoustics literature that might reference their ideas on resonance boards for sometime. When one keeps on digging, one finds. Here are some of them:
https://patents.google.com/patent/US7292702B2
https://patents.google.com/patent/US8611575B1
https://patents.google.com/patent/US20130156243A1
https://patents.google.com/patent/US9008342B1
https://patents.google.com/patent/US9313570B1
https://patents.google.com/patent/US10587949B1
And, the most humorous video I have come across in quite some time. And it is certainly not off topic!
You can only hear what your ears are capable of perceiving—nothing more. You cannot hear what the microphone captures, nor can you access what the software calculates for you!
You can only hear what your ears are capable of perceiving—nothing more. You cannot hear what the microphone captures, nor can you access what the software calculates for you!
It took awhile but I've read through most of the messages in this thread. Based on what I've read decided to give DMLs a shot. So I got a pair of exciters, DAEX32EP-4's; and a couple of coin type ones if I decide I want to add additional high end response. I've settled on a design that is compliantly suspended on all 4 sides using butyl tape that is used for automotive purposes.
My (current) question is regarding the polarity of the drivers. The terminals on the ones I got are not marked. I haven't received the smaller ones yet but don't expect them to be marked, either. So: what IS the polarity of these things? Are they consistent between different styles of drivers?
My (current) question is regarding the polarity of the drivers. The terminals on the ones I got are not marked. I haven't received the smaller ones yet but don't expect them to be marked, either. So: what IS the polarity of these things? Are they consistent between different styles of drivers?
Hi Mark... Usually there'll be a red dot on the positive tab OR commonly, the positive tab will be larger than the neg...If unsure, do a DC test with a battery...+ve outward movement, neg inwards
Cheers
Eucy
Cheers
Eucy
Hello!
Hmm... low exchange activity those days! Here is some reading.
Directivity measurements applied to DML Edition 3
This is an update of the paper showing directivity measurements and coincidence frequency research (see previously #12704, the method, #12705 (link to pdf), #12708 (Ed1), #12712 (Ed1 pdf), #12729 Ed2
Thanks to Eric (@Veleric ) and Thomas (@Sandasnickaren ) who provided measurements and to Dave (@EarthTonesElectronics ) for the inputs from the videos.
The pdf is to large to be attached. Find it at the following place : Github : Directivity measurement applied to DML
Extract :
The edition 3 addon :
Conclusion
The table from the conclusion
The figures
Hmm... low exchange activity those days! Here is some reading.
Directivity measurements applied to DML Edition 3
This is an update of the paper showing directivity measurements and coincidence frequency research (see previously #12704, the method, #12705 (link to pdf), #12708 (Ed1), #12712 (Ed1 pdf), #12729 Ed2
Thanks to Eric (@Veleric ) and Thomas (@Sandasnickaren ) who provided measurements and to Dave (@EarthTonesElectronics ) for the inputs from the videos.
The pdf is to large to be attached. Find it at the following place : Github : Directivity measurement applied to DML
Extract :
The edition 3 addon :
Conclusion
The table from the conclusion
The figures
That's all quite nice, but can we truly hear what the microphone captures or what the software calculates for us?
It’s quite difficult to comprehend how such two corrugated pieces of aluminium foil can produce such a beautiful and engaging sound. Although the song is in Russian and you may not understand the lyrics, the emphasis is on the sound itself rather than the words.
This video was recorded using an ordinary mobile phone available in that region of the world, yet the sound produced by these corrugated pieces on the flattened coil is truly remarkable. (It is best experienced through earphones.) By the way, the individual who uploaded the video possesses considerable knowledge of various types of excellent audio equipment.
-------------------------------------------------------------
Oh, by the way, the NXT panels were cut at an angle. They were crafted from corrugated double-sided plastic material, which is sometimes used by @spedge In other words, the corrugations were not parallel to the vertical sides.
Last edited:
Interesting, thanks lekha. Looks like polycarbonate sheet to me.
https://anandplastics.com/multiwall-polycarbonate-sheets/
https://anandplastics.com/multiwall-polycarbonate-sheets/
lekha did indicate that the substrate is some kind of corrugated plastic (cut so the corrugations are not parallel to any edges). It's difficult to tell from the video but it's possible that the aluminum stuff is attached to the panel. Maybe acting as some sort of distributed whizzer?
I thought I was hearing some odd resonances in the vocals, almost a buzz. But that easily could be the crummy speakers in my laptop.
I thought I was hearing some odd resonances in the vocals, almost a buzz. But that easily could be the crummy speakers in my laptop.
lekha,
When you say "the" NXT panels, what particular panels are you referring to? Are you suggesting that corrugated plastic was a preferred material according to NXT? I never recall reading that. As far as I know, there was no one specific material used in speakers made under the NXT license. The "NXT" speakers I have used a nomex core with paper skins. I suspect there were "NXT" speakers made with lot's of different panel materials.
Eric
The answer to both questions is yes, we can hear much of what the microphone captures, and what the software calculates. But what is your point? Are you suggesting that trying to develop a science based understanding of speaker performance, as Christian is doing, is a waste of effort? And that you think it is akin to spending big bucks on cables that make no perceptible difference?
Eric
Christian,
The pages after 25 (Fig 31) won't load for me. Any idea why that might be? Anyone else have this issue?
Eric
Certain companies produced panels according to NXT specifications for which they paid a significant amount. However, they never disclosed what those specifications entailed. They also did not reveal the type of corrugated material used to manufacture those speakers, nor the angles at which they were cut. The bottoms of the panels were glued to a soft material around the perimeter. As you can see from the image, the "flutes" were left open. Why? I have no idea. @spedge is more knowledgeable about this.
This is a different matter; it pertains to the flattened coil that you might notice beneath the transparent material. There’s a link to another video on that page, also featuring flat panels. It’s fascinating to hear how such unrefined panels can produce sound.
That's good to know. However, we have not yet become robots! 😉
Stanley Rich was a U.S. naval engineer who held around 80 patents. His patents can be somewhat challenging to understand at times. Here’s one of them. The most intriguing aspect is Fig. 8, which he doesn’t elaborate on much, but it certainly provides food for thought.
Hello Eric
There are 2 ways to access to this document.
- thanks to the github pdf viewer which propose the pages by a set of 5 with a button at the bottom of the 5th page to load the next pages. What I see is the "more pages" button change at page 25. On my laptop it becomes a blue empty icon that I can click on to access to the next pages (it is maybe then OS or browser dependent)
- the other (probably better) possibility is to download the doc thanks to the small download (down load raw file) icon on the right just below "History". It should open the full doc in a new browser tab or maybe propose you to download it in some directory.