No surprise there, at least with regard to the relative volume (booming) of PS vs glass.
The "T" factor gives a pretty good prediction of relative efficiency of materials where
T=E/(12*rho^3)
and E=Elastic Modulus (Pa) and
rho=density (kg/m3)
For glass, with E=70 GPa and rho =2500 kg/m3, the T factor is only about 0.37.
In comparison, for PS foam, T is on the order of 50-100.
You could do worse than glass, however. For example, here is T for some other materials:
Aluminum: 0.3
Acrylic (i.e. PMMA): 0.14
Steel: 0.03
Eric
I have only used 6mm thick glass and 1mm thick glass. I don't have 4mm glass yet. 6mm glass sheet gave very loud sound, as much as the 20mm EPS sheet, only my ears didn't like it. One of these days, I'll get 3mm and 4mm pieces of glass, then I'll know the difference. Now, that the scientists at the Delft University had helped those students to win the Dyson award says that the 4mm glass would be good enough, but I have no idea what actuators they used. They might have made them at the University labs. I don't think these commercially available "DIY" exciters are of that quality, they are that, DIY.
Anyway, what might be the T factor of paper? The standard drivers use very thin paper membranes and very little of it. By the way, tightened paper on a frame give very good sound with an exciter on face. @spedge has done lot of experiments with paper DMLs, I believe.
I have only subjective experience to relate.
These "DIY" exciters I have used and those, to which I have had the opportunity to listen have produced the most compelling home audio experiences I have had.
I have had access to listen to some of the most celebrated (and expensive systems) available to reproduce music in a home setting over the past 50 years. Many of these because I have participated in the DIY audio community (long before diyAudio.com launched) and because I have the good fortune of living near some of the US's premier stereo shops and the audiophiles that have frequented them.
This discussion thread certainly explores strategies for optimizing | hacking the use of exciters - much do to the great attention, thought, and experimentation Steve (@spedge) has reported.
Certainly, this thread has speculated about what might comprise an even better exciter.
Nothing I have previously heard comes close to the experience of well crafted DMLs driven by these inexpensive exciters.
I've certainly wondered about what additional quality or design would or could improve the exciters we have readily available, but stepping back and looking at how much progress DMLs have made compared to everything else:
What else do they need to do?
What additional quality or qualities would they need to have?
These small devices have made such a profound contribution to my experience of music in my home, I find myself wanting to protect them and their reputation ;-)
Would such scientists really have access to information or construction techniques to which this community doesn't?I guess all of this goes to the question of what comes next?
Have we approached some limit in the technology?
Do the laws of physics still allow movement forward?
What stands in the way of even better exciters?
Or... should we focus on all the other bits? Material? Suspension | mounting. Dampening? Proportion? Placement?
(Going back now to assembling my rear channel surround DMLs and the amp to drive them).
Hello Chdsl,
As Veleric explained just before, the "T factor" known from for example the piano soundboard reflects the efficiency. Some measurements might be interesting to have a better understanding between glass and EPS. We have almost nothing about the role of the damping that change probably the peak at resonances. The thickness plays low or even no role in the efficiency.
1mm glass is usable for frame, 4mm is (or was?) standard for windows.
I wonder what is the glass behavior at low frequency? Any risk of destruction? Is it possible to go down to 300Hz or even below?
About the T factor of paper, I read some where (but where?), paper has one of the best rigidity, density compromise and probably cost for cone application. We'll see if I can find again this source.
What are the facts or the parameters that lead you to say the standard exciter used might be improved?
Christian
Experimenting with round objects found at home, all sitting on base this time.
The ceramic tall vase gave a bit more bass, but sound was tubular. Small drinking glass gave a nice sound sitting on its base and on edge. The pinkish half full glass candleholder was also good, but the closed glass candleholder gave a fuller sound. The smallest coloured one was just as good. Most interesting sound came from the crown like one, maybe the spikes added something extra. On all these I could use only one exciter.
Also tried all the cutting boards at home, on face and on edge. Plastic boards were not that good either way, but the timber ones were really good, the bamboo one being the best. It was 1cm thick, 28cm x 22 cm in size. On edge was much better than on face. Both stereo channels were used.
Then I found an old DKNY cosmetic box/tube without the lid, made of paper, and covered with some glossy foil. Wife didn't care what I do with it, so used it for experimenting.
Tried to place it on the open edge, but the edge was not in a good shape, so hard to keep it strait, the paper just bent. So, placed it on the bottom edge, which protrudes about 5-6mm. Pasted it over both exciters with double-sided tape. Sound was quite good for an old paper tube. It was even better, after I closed the top with few Ikea mats. They also gave some weight to the paper tube. By the way, it was omnidirectional.
I suppose the paper tube is much better than a glass tube, but the paper is much weaker than glass. I don't have a timber or bamboo tube at home to try. Actually, covering the tube from above gave a fuller better sound, even with the ceramic vase. Can understand why Sony covered their glass tube omnidirectional speakers from top, the cheaper LSPX -S1, 2 and 3 and the very expensive Sountina NSA-PF1. Of course, a paper tube won't be marketable gadget, but a glass tube would. Sony LSPX-S3 is now selling at 250USD, but who would buy it to dismantle it to check which actuators Sony had used? Those 3 actuators must be pretty tiny.
Anyway, actuators can be up-fired and through the edge. Closed empty objects would give better sound output than flat 2-d ones. These are my subjective feeling. My ears are not the same as few decades ago, though. As this kind of stuff is available in every home, you can do the same (and better) experiments.
The ceramic tall vase gave a bit more bass, but sound was tubular. Small drinking glass gave a nice sound sitting on its base and on edge. The pinkish half full glass candleholder was also good, but the closed glass candleholder gave a fuller sound. The smallest coloured one was just as good. Most interesting sound came from the crown like one, maybe the spikes added something extra. On all these I could use only one exciter.
Also tried all the cutting boards at home, on face and on edge. Plastic boards were not that good either way, but the timber ones were really good, the bamboo one being the best. It was 1cm thick, 28cm x 22 cm in size. On edge was much better than on face. Both stereo channels were used.
Then I found an old DKNY cosmetic box/tube without the lid, made of paper, and covered with some glossy foil. Wife didn't care what I do with it, so used it for experimenting.
Tried to place it on the open edge, but the edge was not in a good shape, so hard to keep it strait, the paper just bent. So, placed it on the bottom edge, which protrudes about 5-6mm. Pasted it over both exciters with double-sided tape. Sound was quite good for an old paper tube. It was even better, after I closed the top with few Ikea mats. They also gave some weight to the paper tube. By the way, it was omnidirectional.
I suppose the paper tube is much better than a glass tube, but the paper is much weaker than glass. I don't have a timber or bamboo tube at home to try. Actually, covering the tube from above gave a fuller better sound, even with the ceramic vase. Can understand why Sony covered their glass tube omnidirectional speakers from top, the cheaper LSPX -S1, 2 and 3 and the very expensive Sountina NSA-PF1. Of course, a paper tube won't be marketable gadget, but a glass tube would. Sony LSPX-S3 is now selling at 250USD, but who would buy it to dismantle it to check which actuators Sony had used? Those 3 actuators must be pretty tiny.
Anyway, actuators can be up-fired and through the edge. Closed empty objects would give better sound output than flat 2-d ones. These are my subjective feeling. My ears are not the same as few decades ago, though. As this kind of stuff is available in every home, you can do the same (and better) experiments.
By the way, here's an interesting article how our inner ear becomes an electrical station that transfers the sound energy into electrical energy. There are few interesting articles at the end of that article about how music acts on our mind. Some of this was taught at school, but we've usually had forgotten that.
Watching a cone of a speaker driver, or even the vibrations of the flat panel, we might think that the sound vibrations are pushing the air. They don't. Sound simply travels through air. It doesn't push it. There's no wind in the room. Air is just the medium, which is much lighter, so it travels further, than in steel, even though speed of sound is very much higher in steel than in air.
There's a certain distance from the voice coil edge to the surround in a paper cone speaker driver. When the voice coil is working we see the cone moving forward and back, but we don't see any vibrations going through it from VC to the surround. We only see the surface vibrations, sort of to and fro ones. There should be sound travelling through the paper, even though the distance is too little, which in turn sends its own sound vibrations through air. The cone is not held tight between the VC and the surround, as the surround is flexible.
I held a very thin card vertically with its edge touching the exciter/transducer, the music got amplified. The card's width was 30cm, while the edge that touched was just 3.5cm. I tried this with an A4 paper too. Fingers too felt the vibrations.
Watching a cone of a speaker driver, or even the vibrations of the flat panel, we might think that the sound vibrations are pushing the air. They don't. Sound simply travels through air. It doesn't push it. There's no wind in the room. Air is just the medium, which is much lighter, so it travels further, than in steel, even though speed of sound is very much higher in steel than in air.
There's a certain distance from the voice coil edge to the surround in a paper cone speaker driver. When the voice coil is working we see the cone moving forward and back, but we don't see any vibrations going through it from VC to the surround. We only see the surface vibrations, sort of to and fro ones. There should be sound travelling through the paper, even though the distance is too little, which in turn sends its own sound vibrations through air. The cone is not held tight between the VC and the surround, as the surround is flexible.
I held a very thin card vertically with its edge touching the exciter/transducer, the music got amplified. The card's width was 30cm, while the edge that touched was just 3.5cm. I tried this with an A4 paper too. Fingers too felt the vibrations.
Hello Chdsl
Thank you for the link. The video is a good summary. The hearing mechanism is quite clear in my mind up to the output of the cochlea which operates like a filter bank. Many things about that are available including the model with gammatone filters. The spectrograms shown in posts here are close to that. The difficulties come after in the operations of the brain... mainly in the way the brain integrates the SPL over the time.
About vibrations in cones or DML, I think there were already many exchanges or papers about them. In a standard cone speaker, the main working mode search is the pistonic mode (all the points moving the same direction, same quantity at the same time). When the frequency incrediyAudioases, because of the mass and the rigidity of the membrane, it is not more possible; the behavior of the cone changes. A DML works in this mode.
For now, the way a membrane excited at the edge enters works, is not clear for me... when the exciter pushes to the membrane, I can imagine the membrane being blocked at the top or even by inertia with its own mass is then bending. But what happens when the exciter is pulling?
Christian
We just assume that all the points move in the same direction, same quantity when the voice coil moves in the pistonic mode. It is easy to assume it that way. But does all points in the cone move that way, really? The points just next to the voice coil on the paper cone move maybe at the same speed as the voice coil, but what about the points few mms away, few cms away, or near the surround? They move at different speeds, don't they? The speed of sound in few times more in solids than in air. In paper, it could be ~3000m/s. Speed of sound in normal air is 343 m/s. So, the vibrations go faster through the paper, before it transfers to air.
I didn't think of edge transfer of sound in solid objects before reading about Ammos speaker. Then I found about the Sony product, which intrigued me more. So, the experiments on edge, rather than on face of materials. Even, the small cork mats I use to isolate the exciter from the table, amplifies sound, when placed on edge on top of the exciter, tiny but clear. I tried everything normally available in a home, even the plastic flower pot covers, round ones, square ones, flared ones, on the bottom and on edge. Anyone can experiment, for these are items one has at home.
What happens when the exciter is pulling is a good question, for which I don't have an answer. I didn't think of that before. I usually use a double sided tape to fix things that I can't tightly hold over the exciter by hand. Your question suggests that the any surface, large or quite thin must always be in contact with the voice coil to transfer the vibrations from it. This is true.
Chdsl
To show you loudspeakers close of the driving method you are experimenting: in post 7366 a DIY inspired by the Walsh speaker and the one below inspired by the Lineum. Both have an initial driving direction almost in the plane of the membrane. I made them years ago... what I remember is a lack of high frequency. The yellow membrane is a simple A4 80g sheet of paper. The exciter is full range for car without membrane, the frame was cut. A more rigid material would have probably get a better bandwidth. My trial was inspired by an article from Audioxpress which used a membrane made in the material of drum membrane.
I made those tentative long time ago before crossing the threads about DML. I stopped despite the not too bad results because of a lack of information on how to improve them, how they work. DML have the advantage of quick better results (wider bandwidth), a large documentation (not always easy to understand), a dynamic community of DIYer (like our here). This open a question : why the DML knows such a development when other solutions (Walsh, Lineum, should I include Manger?) are more limited?
About DML, one characteristic to keep in mind is their diffusive characteristic. The physical aspect of that is the low correlation of impulse response in different directions; The consequence is I think the low sensitivity to the boundaries, in other words, the high quality of the image which is the most remarkable quality of DML A loudspeaker like a vertical tube or cone won't have this property. The lineum probably.
Your experiment is pretty good. Did you paste the bent edge over the voice coil top plate? Why did you bent the paper like an half a cylinder to both sides? Did the paper gave a better sound than it was standing straight up? Whether the transducer plate/voice coil plate transferring the vibrations to a larger surface on face, or to a very thin surface (on edge), it is still a distributed mode loudspeaker, if the panel, sheet flat or curved amplifies the sound. Like to know what you did and why you did that experiment.
The Linaeum idea, I suppose.
My estimate of the T factor of paper is that it's about 1, assuming an elastic modulus of about 1 GPa and a density of about 1000 kg/m^3. So likely a bit higher than glass but lower than plywood (about 2 to 10).
Yes, for sure standard pistonic drivers commonly use paper cones. But not because of it's high T factor, I would guess. Perhaps more likely because of it's relatively low T factor.
Eric
I have not done a lot of experiments with paper, only thin card.
And usually sandwiched between layers of pva, a ply basically.
The performance matches the larger panels well, and in some respects improves on them especially in smaller rooms.
Sometimes smaller is better.
Steve.
Can someone please explain to me based on their own experience how to connect the vibrating plate to its frame, such as whether it is better to make it integral around the entire perimeter or only the short or long side or maybe just some points and if you need rubber pads or screwed directly
Thank you
Thank you
@pixel1 --
I don't think anyone can give you one answer. Rather, we can begin to think about paths to explore. Some thoughts (at least from my perspective) follow.
The connection of a DML panel ("vibrating plate") to a "support" (a broader idea than a "frame") must | may do a couple of things...
Isolate the DML panel from interacting with the support. Lots of ways exist to do this:
Hang the top of panels by threads or fishing line from a support (ceiling, brackets extended from a wall, top of a frame). The tiny dimension of the threads shouldn't transmit human perceptible audio frequencies to the support. A similar idea to this uses spikes under conventional speaker cabinets or equipment that won't transmit low frequencies. hanging from the top, for practical purposes, positions a panel as if floating in space (with the caveat that something different may go on in the corners that support the panel than across the rest of the panel when so suspended). You can find lots of examples of this with EPS panels early on in DML explorations. I've used this method with a number of different materials including foam core, acrylic, baltic birch, thin aluminum sheet.
Hang (generally rectangular) panels by threads or fishing line from their 4 corners - I've thought about this, but haven't got around to trying it yet. Capturing all the corners might affect the vibration across the panel.
Hanging notes:
Hanging panels has the weight of the panels extending down from the supports. Any such hanging | suspension of the panels will generally position the panels parallel to reflective walls behind them. I've started wondering about standing waves (especially when positioning panels close to a wall). One could "hang" panels off parallel, but that might introduce other stresses on the panels that won affect their sound.
Minimalist support - I supported my first DML panels on the points of spikes and balanced them from the top with a single cantilevered arm with some foam between it and the panel. This created a minimalist 3-point support. It sounded good, but proved treacherous when vacuum cleaners appeared in their proximity. I eventually replaced the spikes with 1/2" wide channels lined with foam. Supporting panels at their bottom edge with spikes or resting panels on relatively small points isolated by foam, has the entire weight of the panel passing through (if mostly vibrationally isolated from) very small supports|connections.
What does this do to the vibration of the the panels? Worth exploration.
I tend to like minimalist solutions. I currently have panels, each panel sitting on a pair of 4" L brackets (screwed into a brick wall) with some foam between the brackets and the panels. I've then stuck a piece of the same foam on the wall where the center of the top edge will lean against it. This creates 3-point support with long "ports" (approximately 3/8" across the top and nearly 4" across the bottom) the width of the panel and (theoretically) should mitigate standing waves. Pure speculation this.
4 or more connections - With respect to isolating the panel from the support, one can move from 3 "points" of support to supporting the entire edge of panels. Participants in this thread have reported lost of different ways of doing this. Once can clamp the panels between foam (isolating it from the larger frame) or attach panels to a frame with double sided foam tape of some kind. Some have secured a pair of parallel edges in this way leaving the other edges (e.g., to and bottom) free.
Cantilever - Some recent discussion has explored catching rectangular panels at 2 points at various fractions of their vertical height or even along the entire bottom of a panel. This certainly deserves exploration.
Contact of panel with support - Material stretched over a frame has full contact with the support with no isolation from the support. Steve (@spedge) should opine of this. Others appear to have had success by directly screwing acrylic or 3 mm birch panels to a frame without isolating the panels from it. Instrument makers (guitars, cellos, ...) probably have relevant experience or explanations of the benefits and limitations of doing this.
Cantilevering from a hard connection (e.g., across the bottom or maybe even the top) provides an interesting idea. My experiments with using an OLED TV as a center speaker has the DML portion of the panel "hard" connected at its bottom without isolation from its support. Sounds pretty good.
The reported performance of DMLs under all of these various ways of supporting them continues to astonish me. DMLs represent an uncannily versatile and forgiving technology.- Andreas
There's always a problem of baffle edge audio distortion in normal box speakers. Some manufacturers try to rid the problem by curving the baffle/front panel. DML is practically always a rectangle panel. Interesting how we'd get rid of the edge distortion.
The known reviewers such as 6moons don't review mass produced stereo systems and their speakers. Mass produced here means, Sony, LG and such like. They have massive R&D budget, and as they produce a lot, and for a short time scale, they can sell such systems very much cheaper than all those high-end (translated as terribly expensive), so they are sort of excluded from reviewing. Anyway, these companies are using methods for ridding the edge distortion. They place speakers at different planes, different angles, and so on. Here's two examples,
The baffle is a specially designed 3-d plastic enclosure, while the back volume is a rectangular box, most times MDF. They also have 'strange' looking bass-reflex holes, rarely round, or even horizontal.
This post appears maybe somewhat off topic, but maybe someone here might think of a non-flat DML panel, or maybe with some edges like in the 2nd image? Or, curvy like in the 1st?
The known reviewers such as 6moons don't review mass produced stereo systems and their speakers. Mass produced here means, Sony, LG and such like. They have massive R&D budget, and as they produce a lot, and for a short time scale, they can sell such systems very much cheaper than all those high-end (translated as terribly expensive), so they are sort of excluded from reviewing. Anyway, these companies are using methods for ridding the edge distortion. They place speakers at different planes, different angles, and so on. Here's two examples,
The baffle is a specially designed 3-d plastic enclosure, while the back volume is a rectangular box, most times MDF. They also have 'strange' looking bass-reflex holes, rarely round, or even horizontal.
This post appears maybe somewhat off topic, but maybe someone here might think of a non-flat DML panel, or maybe with some edges like in the 2nd image? Or, curvy like in the 1st?
Here's a video where a Russian is dismantling 1986 Sony DML acoustic system, APM-66ES. It is interesting how Sony had transferred voice coil vibrations to that small aluminium honeycomb square flat panel. You can see that at 2.44mins into the video. In another video, you hear them playing on a Luxman l-570 from 1989.
Interestingly, the DMLs doesn't have to be big, and can be boxed, and bass-reflexed. Even the tweeter is a dml.
Interestingly, the DMLs doesn't have to be big, and can be boxed, and bass-reflexed. Even the tweeter is a dml.
A Soviet DML speaker restoration project. The LF DML speaker driver in this video. The USSR stopped existing in 1991, so this 35AS-021 Estonia speakers must have been made sometime in the '80s. I've seen all kinds of Soviet speakers, but never saw this one before. This too looks like an aluminium honeycomb sheet. Too bad this guy is not dismantling the driver, or even take the plastic cover off from the back.
And here's guy showing the dismantled Technics SB-X500 honeycomb speakers 13 years ago. Technics transferred the sound vibrations to the DML through a stiff aluminium (or plastic) cone. Here's another guy showing how they work and sound. The video was done 10 years ago.
Any thoughts?
And here's guy showing the dismantled Technics SB-X500 honeycomb speakers 13 years ago. Technics transferred the sound vibrations to the DML through a stiff aluminium (or plastic) cone. Here's another guy showing how they work and sound. The video was done 10 years ago.
Any thoughts?
chdsl,
Cool, I especially liked the last one. It clearly shows that it's a BMR (balanced mode radiator) which combines pistonic motion for the low end with dml response for higher frequencies.
Thanks for sharing those.
Eric
Hello Chdsl,
Yes the I pasted the edge (the folding of the paper) over the voice coil plate.
The half cylinder is part of the original design like in the link about Linaeum (sorry for the missing "a" in the previous post).
I didn't test with paper in another position. At his time, I had few inputs on bending waves speakers; It was more a first trial than a real design.
Here are some key words from the same family : rubanoide, Janus50, F Deminiere who made one speaker like that. See A DIY Ribbon Speaker of a different Kind or here. The intriguing point is the high efficiency of those speaker. a big difference with our DML is the very long coil