Anyone ever tried to make a DML really big like just damping the complete back wall with some 3-5 cm thick foam and putting a wooden frame along all sides and fitting some heavier plastic foil covering the complete area with just two (for stereo) more powerful DMLs behind it or on it and EQing it linear?
Idea is to have an elegant fullrange which goes deep and omitting the need for a box.
Its a room installation which must not be costly and is easy to realize from a technical point of view.
I understand that a small DML crossed to a subwoofer is more flexible / mobile but room installations are rarely realized.
Idea is to have an elegant fullrange which goes deep and omitting the need for a box.
Its a room installation which must not be costly and is easy to realize from a technical point of view.
I understand that a small DML crossed to a subwoofer is more flexible / mobile but room installations are rarely realized.
Good remark. Like if the HF in one case was able to escape from the central area, out of the mask of the exciter and not in this case... Maybe a difference in the damping versus frequency between both?
Christian
Hi Leob,
I am in the idea of 2 architectures. A DML which is far enough from a wall (about 1m?) that could work in open back and a DML that can't be far enough of the wall with some absorption of the rear wave. As the open back is the most common configuration that has the favor of the DML builders here and it is a possible configuration for the room I would like to have new loudspeakers, I decided to investigate it more before going to solution with a semi closed back which seems to be a full new area.
Christian
I have in mind the realization from CLS in this way. The dimensions were probably a limitation for more applications?
Christian
Thanks a lot for your feedback Eucy! I am sure there is a reason why you suggest to keep the panel width at around 300mm. I was hoping to keep the proportions according to the golden ratio (1:1.618) . According to that ratio, a panel with a width of 300mm would be just 485.4mm tall. I could apply the same ratio also to 600 or 900mm long panels but they would be a bit wider (370.8mm or 556mm). Is there any proportion you can recommend or do you think I should drop the idea about the golden ratio and go for panel size you recommended? Regarding making my own composite panel, I am not too worried about shrinkage since it's ply. However bending could happen but I believe that a hard curing glue like araldite could prevent the panel from doing that so easily. What are you referring to regarding the sealer? The glue used for the compoite panel, the varnish top coat or both? Once I have decided what size to go for, I will do some experiments and share the results here. Thanks a gain for any feedback
cheers
shido
I think it will be worth looking at these two posts and Ben Zenker's video they refer to.
#12,967
@Leob: “Is it some preference that open baffle enthusiast have that having reflections from a real wall somehow improves the sound in a hifi setup, or are you actually using them in scenario where you make use of the omnidirectional sound?”
Regarding the first idea, it could be but with extensive effort in room treatment, otherwise is luck.
I`m using the panels omnidirectionaly in my school when the students present their works in very terrible acoustic environments (RT headhach), for the sake of speech intelligibility.
In home, I use them open (no spine) with the best results close to the corners (the walls are 90% glass) precisely because of what you said: @Leob “since diffusing the already diffuse signal doesn't do that much to change the overall impression”.
What seems to me is that contrary to what you said, although I agree: @Leob “It would be similar to when you are producing music and mix in a little bit of reverb and you have to make quite drastic filtering to hear the difference”, in the DML case the reverberation is so psychoacousticaly present that the “diffuse filtering” is not drastic and can be tamed to seem “hipernatural” or, at least, pleasant as a UAU! factor.
@homeswinghome: “What I am pretty sure is that some peaks in the front FR (a time gated one that excludes reflections) might have this rear emission as cause ("might" because I have no evidence of that)”.
I think so either, and that’s why I use PMMA 4mm over other tested materials.
Which lead us to the backwave…
@homeswinghome: “My current thought for using plates indoor is to cover the back of the frame with some material that dampens the emitted sound without actually closing the back. So something like acoustic foam or rockwool, which will not impede the movement of the plate since it doesn't completely block the waves, and that will not reflect a lot, but reduce the SPL to the rear. It will not be effective over the whole range of the speaker, but should make it better and can be complemented with some acoustic treatment”.
As you might remember I don´t use frames but I did that experiments and, until 86/90 dB, what I enjoy the most is micro perforated blackouts on the rear because it`s a compromise to retain that lovely diffuse sound (in terms of objectiveness all of us have speakers who measure better). Expecting your results.
Although, as I did post prior, i´m confident on the research evolution of metamaterials and that`s the approach were I put “my eggs on the basket” regarding backwaves:
https://www.sciencedaily.com/releases/2018/04/180410132851.htm
Because of mechanical alignment with the 15inch OB woofers I have the DMLs 75cm from the corners, but when they were just crossed to the sub the distance was 15cm to the blackouts and 6cm from those to the wall glass, which integrates better with the room in a invitation to listen consistently at lower output levels (as if the music emotional content conveys as a whisper, which is a very hard task for conventional speakers).
Regarding the first idea, it could be but with extensive effort in room treatment, otherwise is luck.
I`m using the panels omnidirectionaly in my school when the students present their works in very terrible acoustic environments (RT headhach), for the sake of speech intelligibility.
In home, I use them open (no spine) with the best results close to the corners (the walls are 90% glass) precisely because of what you said: @Leob “since diffusing the already diffuse signal doesn't do that much to change the overall impression”.
What seems to me is that contrary to what you said, although I agree: @Leob “It would be similar to when you are producing music and mix in a little bit of reverb and you have to make quite drastic filtering to hear the difference”, in the DML case the reverberation is so psychoacousticaly present that the “diffuse filtering” is not drastic and can be tamed to seem “hipernatural” or, at least, pleasant as a UAU! factor.
@homeswinghome: “What I am pretty sure is that some peaks in the front FR (a time gated one that excludes reflections) might have this rear emission as cause ("might" because I have no evidence of that)”.
I think so either, and that’s why I use PMMA 4mm over other tested materials.
Which lead us to the backwave…
@homeswinghome: “My current thought for using plates indoor is to cover the back of the frame with some material that dampens the emitted sound without actually closing the back. So something like acoustic foam or rockwool, which will not impede the movement of the plate since it doesn't completely block the waves, and that will not reflect a lot, but reduce the SPL to the rear. It will not be effective over the whole range of the speaker, but should make it better and can be complemented with some acoustic treatment”.
As you might remember I don´t use frames but I did that experiments and, until 86/90 dB, what I enjoy the most is micro perforated blackouts on the rear because it`s a compromise to retain that lovely diffuse sound (in terms of objectiveness all of us have speakers who measure better). Expecting your results.
Although, as I did post prior, i´m confident on the research evolution of metamaterials and that`s the approach were I put “my eggs on the basket” regarding backwaves:
https://www.sciencedaily.com/releases/2018/04/180410132851.htm
Because of mechanical alignment with the 15inch OB woofers I have the DMLs 75cm from the corners, but when they were just crossed to the sub the distance was 15cm to the blackouts and 6cm from those to the wall glass, which integrates better with the room in a invitation to listen consistently at lower output levels (as if the music emotional content conveys as a whisper, which is a very hard task for conventional speakers).
I've recently purchased an RV with a permanent bed in the rear left corner which has as the foot board a closet wall. I have a couple of my Polyurethane panels (8"x14") that will be installed flush in that wall under the TV. Planning on making a 4" wooden frame that protrudes into closet with some aluminum mesh attached to the rear of that which can be taken on and off easily to test different rear materials or just go without and see what the hanging clothes do. Curious to see what comes of this as I've been listening to some of these panels open rear for well over a year now so will be able to compare easily.
Last edited:
Many times, Steve has suggested the possibility of comb filtering when multiple exciters are used on the same panel. I have tried to look for evidence of that on a few occasions, but didn't find it. But I finally observed it in a recent test.
In this test I used a carbon fiber/balsa sandwich composite panel mounted to a frame using a high damping Poron foam around the perimeter. Two DAEX25VT-4 exciters were placed about 7 inches apart along the horizontal centerline of the panel.
Shown below are REW results for various angles of rotation of the speaker relative to the mic at 1 meter.
On each SPL curve, the minima marked with red crosses correspond nearly exactly with the frequencies at which comb filtering "should" occur. That is, where the path length from each exciter to the mic differs by an odd multiple of half wavelengths, and hence results in destructive interference. The "number" placed at each minima is the number of half wavelengths that corresponds with that particular minima. The nearly perfect correlation of these minima with their expected frequencies is very convincing evidence that this is indeed the result of comb filtering.
Interestingly, the lowest frequency at which I could clearly identify comb filtering was at about 7k. Some of the dips seen at lower frequencies could possibly also be the result of comb filtering, but I couldn't be confident enough to "assign" them to comb filtering based on the sheer number of dips in the FR at lower frequencies.
Note that this may be a bit of an extreme example. People typically place exciters closer than 7 inches (I think), which would result in fewer minima than with the exciters this far apart. Also, the very high perimeter damping may cause localization at a lower frequency, and hence may exacerbate comb filtering at lower frequencies
But in any event, I have no doubt that comb filtering can occur with multiple exciters on a single panel.
Eric
In this test I used a carbon fiber/balsa sandwich composite panel mounted to a frame using a high damping Poron foam around the perimeter. Two DAEX25VT-4 exciters were placed about 7 inches apart along the horizontal centerline of the panel.
Shown below are REW results for various angles of rotation of the speaker relative to the mic at 1 meter.
On each SPL curve, the minima marked with red crosses correspond nearly exactly with the frequencies at which comb filtering "should" occur. That is, where the path length from each exciter to the mic differs by an odd multiple of half wavelengths, and hence results in destructive interference. The "number" placed at each minima is the number of half wavelengths that corresponds with that particular minima. The nearly perfect correlation of these minima with their expected frequencies is very convincing evidence that this is indeed the result of comb filtering.
Interestingly, the lowest frequency at which I could clearly identify comb filtering was at about 7k. Some of the dips seen at lower frequencies could possibly also be the result of comb filtering, but I couldn't be confident enough to "assign" them to comb filtering based on the sheer number of dips in the FR at lower frequencies.
Note that this may be a bit of an extreme example. People typically place exciters closer than 7 inches (I think), which would result in fewer minima than with the exciters this far apart. Also, the very high perimeter damping may cause localization at a lower frequency, and hence may exacerbate comb filtering at lower frequencies
But in any event, I have no doubt that comb filtering can occur with multiple exciters on a single panel.
Eric
Thanks for the feedback. I saw the video and it's a bit complex for a beginner like me to understand, so I was ending up asking chatgpt to help me designing the best panel with the components, materials and tools i have access too. It recommended to build the honey comb composite panel and recommended to use thin fibre glass fabric insted of veneer. It also recommended the panel dimensions, honeycomb size, spacing, glue, mounting etc. I can share the results if someone is interested. cheers shidoI think it will be worth looking at these two posts and Ben Zenker's video they refer to.
#12,967
#12,968
Sorry, but I would not trust any advice given by him. Studying and following this thread is complex and confusing enough, but your best bet is to look at what some of the members here are building and sharing. And rather ask them directly on this thread for best first hand advice, knowledge, and experience.
Last edited:
Eric -
This was interesting, and I couldn't help but plug this into my PETTaLS software and see what happens. I simulated a very thin steel panel to make sure that the coincidence frequency was above 20 kHz, and I tried three different exciter sizes: point sources, small exciters (DAEX19CT-4), and large exciters (DAEX25VT). For each case, I ran a simulation with only one exciter, with two exciters nearby, and with two exciters farther apart. The interference pattern results are pretty clear, and are largely based on the distance between the two exciters, as you observed, but also appear to be somewhat affected by the shape of the exciter. (For example, the large exciter essentially has its own interference pattern at a high frequency!) The point sources don't appear to really produce the pattern to any noticeable degree, for whatever that's worth.
Also - just waiting on final approval before I can release the "lite version" of this model for download. Shouldn't be more than just a couple more days.
Excellent Eric,
This question have been pending for a long time with different opinions. It is good to have evidences to close it or at least to get a common understanding.
Christian
@EarthTonesElectronics
Hello Dave,
Thank you for the simulation of the multiple ring sources.
The dark blue pattern (see below) is visible in almost all the directivity plots I made with a 25mm exciter. It is from that and the concentrator patent I suggested to have an interface less than 20mm in order to "reject" this pattern after 10k. For now, no interface have really worked!
The pattern below might be similar to the one of the Tectonic 500 that didn't fit with the coincidence frequency pattern in what I posted.
I would be interesting to have an equation of it. It would make possible its identification in a directivity plot.
Hello Dave,
Thank you for the simulation of the multiple ring sources.
The dark blue pattern (see below) is visible in almost all the directivity plots I made with a 25mm exciter. It is from that and the concentrator patent I suggested to have an interface less than 20mm in order to "reject" this pattern after 10k. For now, no interface have really worked!
The pattern below might be similar to the one of the Tectonic 500 that didn't fit with the coincidence frequency pattern in what I posted.
I would be interesting to have an equation of it. It would make possible its identification in a directivity plot.
Hello RMAM,
What is "micro perforated blackouts" ? Sorry if you have already shared it!
Other topic... very often, the DML are referred as to be diffused. This is understandable in the idea that the sound sources changed over the panel area with the frequency. From measurement point of view, I am not able to linked it to something specific. I remember a paper trying to explain it by the correlation of the IR off axis versus on axis (maybe I should read it again...). What I see from the directivity measurement is more a source with a very wide radiation up to very high in frequency with no phase shift or important delays. The IRs remain similar over a large angle.
Maybe like a source that lights the room with the same color in almost any direction?
Christian
Hello Christian
About the Bkackouts:
https://vectadesign.com/wp-content/uploads/2024/12/Perforation-brochure-ENG.pdf
Indeed, although let`s think that colour in reveberation terms:
https://www.soundonsound.com/techniques/how-use-reverb-pro-1
The point is that one can mimic the dml sound in a conventional speaker using sound engineer techniques but…
Is astonishing how with little effort and knowledge at a residual cost one can make dml sound impressive, mainly in the vocal region. Why?
If one listen close to the panel the reverberation is easily identifiable, so as if each panel was conveying bundled sound techniques. In the pro mixing they will say that reverb/delay and all the other effects must dwelled with parsimony to “bring the mix together”. In our case is to “bring the reproduction in the room together”, so it`s all about coherence.
Christian, about “What I see from the directivity measurement is more a source with a very wide radiation up to very high in frequency with no phase shift or important delays”.
I think that the sum of what is of no importance is, precisely what is important to explain the nature of the dml diffuseness.
In the sound design/mixing the best results came from the judicious blend of “non important” teaks per se, although the goal is the final product, so the sum of them.
And we can`t really think of sound reproduction coherence without considering:
https://www.pnas.org/doi/10.1073/pnas.1612524113
So the “light colour” comes from a source that due to it`s nature seems to be the result of a very competent sound engineer which we strive to know the details of his knowledge, i,e, the judicious effects applied (blended as such which seems as no important).
Just for fun, try to insert a reverberation tool in your system and anulate the dml reproduction nature.
Good auditions
Rui
About the Bkackouts:
https://vectadesign.com/wp-content/uploads/2024/12/Perforation-brochure-ENG.pdf
Indeed, although let`s think that colour in reveberation terms:
https://www.soundonsound.com/techniques/how-use-reverb-pro-1
The point is that one can mimic the dml sound in a conventional speaker using sound engineer techniques but…
Is astonishing how with little effort and knowledge at a residual cost one can make dml sound impressive, mainly in the vocal region. Why?
If one listen close to the panel the reverberation is easily identifiable, so as if each panel was conveying bundled sound techniques. In the pro mixing they will say that reverb/delay and all the other effects must dwelled with parsimony to “bring the mix together”. In our case is to “bring the reproduction in the room together”, so it`s all about coherence.
Christian, about “What I see from the directivity measurement is more a source with a very wide radiation up to very high in frequency with no phase shift or important delays”.
I think that the sum of what is of no importance is, precisely what is important to explain the nature of the dml diffuseness.
In the sound design/mixing the best results came from the judicious blend of “non important” teaks per se, although the goal is the final product, so the sum of them.
And we can`t really think of sound reproduction coherence without considering:
https://www.pnas.org/doi/10.1073/pnas.1612524113
So the “light colour” comes from a source that due to it`s nature seems to be the result of a very competent sound engineer which we strive to know the details of his knowledge, i,e, the judicious effects applied (blended as such which seems as no important).
Just for fun, try to insert a reverberation tool in your system and anulate the dml reproduction nature.
Good auditions
Rui
Hello Eric.
I think you are referring to the in air measurements at 1m ?
Which is a well-known fact of two point sources causing comb filtering in the air.
The comb filtering I was talking about , and Audiofrenzy for that matter, was the comb filtering on or should I say in the panel itself.
This happens even before the sound has been radiated into the air.
If you throw two stones onto a very large pond, the ripples will clash and collide long before the ripples reach the pond edge.
So no DML action has even started, these are all bending waves .
The mishmash between the two exciters ( stone splashes) causing problem peaks and dips (comb filtering or whatever you wish to call it)is very audible.
But I must stress that this is separate from the in air response of two cone drivers (point sources ) placed close together on a baffle.
Yes the two exciters placed on a dml panel are also point sources but the vibrating surfaces ,unlike the cones, are joined together .
I could make some very poor quality drawings ( Christian knows what I mean 😀) which would also show the similarities within the coil area.
I hope this finally clears up this problem.
It would be interesting to see if this can be measured(duplicated )using your various softwares ?
Steve.
Tanks! Reading trough 678 pages to find answers will take me a lot of time, which I don't have. Unfortunately everything I need to know is somewhere in that thread...but where...? I am a beginner and feel like a kindergarden boy in a high school reading your technical posts. AI can help me to understand some basics and I am a bit in a hurry since my genelec's recently burned and I have no sound system in my house. I don't need a Rolls Royce of a DML right now, but something which sounds good until I find someone who can repair my monitor speakers professionally here in Goa. I will post my idea a little later (still working on it) and hope for some constructive feedback. shido
Chat GPT: "
The thread "A Study of DMLs as a Full Range Speaker" on diyAudio is a valuable resource where enthusiasts share their experiences and findings on Distributed Mode Loudspeakers (DMLs). Participants discuss various aspects, including panel materials, exciter placement, and design enhancements. For instance, users have explored different panel sizes and materials to optimize sound quality.
While I aim to incorporate insights from such discussions, I may not capture every update or individual experiment. Therefore, for the most current and practical advice, especially for hands-on projects like building DML speakers, I recommend consulting active forums and communities where enthusiasts regularly share their latest findings."
Yes, 1 meter. And yes, of course two ideal point sources cause comb filtering in the air. But for me, the question was always this: do exciters on a panel act like ideal point sources? Or is there something different about the way a flat panel produces sound waves in the air such that they don't act like ideal point sources? But for me now the answer is clear that (at least at high frequencies) they do act like ideal point sources.
I think it's confusing to talk about this effect as comb filtering. But absolutely there is constructive and destructive interference between the separate groups of bending waves on a panel with multiple exciters. But I don't believe that this is really any different from the interference caused by bending waves from a single exciter bouncing off the perimeter of the panel and interfering with the original wave. Is there some reason it would be? For even a single exciter, the interaction of the reflected waves with the original wave (and with each other) equally creates a "mishmash" of peaks and dips in the panels velocity profile. These dips and peaks are in fact the nodes and antinodes of whichever modes (natural frequencies) are most effectively excited by a particular exciter location. Combining multiple exciters on a single panel mainly excites a different set of modes, in much the same way that moving a single exciter from one location to another location would. Except that multiple exciters provide the possibility of even better control over multiple modes than is possible with a single exciter, as described in this paper:
https://www.researchgate.net/public...ment_for_Array-Driven_Flat-Panel_Loudspeakers
The Surface Velocity Map in Dave's PETTaLS program should show pretty much all the effects of the interactions of multiple exciters on the bending waves in the panel. But from what I have seen so far, there is nothing particularly alarming in the velocity map when two exciters are used that would indicates a violent clash of waves on a panel that might lead to particularly bad audible effects (other than the comb filtering in air that was demonstrated in his most recent post).
I'm never sure what you mean when you say this. Can you clarify? For me, the term DML simply means a speaker using bending waves (i.e. a bending wave speaker) designed using the NXT principles. So DML action and bending wave action have no real distinction in my mind.
But I suspect in this context you may be thinking of the instant at which there is a sudden change, like a new note is played, and the new waveform is traveling across the plate but has not yet reflected from the perimeters, and hence the new pattern of standing waves is not yet established. Is that what you mean when you say "these are all bending waves"? Or do you mean something different?
Eric