A Study of DMLs as a Full Range Speaker

Christian and Eric.
So , I have two identical corrugated cardboard panels, but one has pva coating and the other epoxy.
The epoxy panel suffers from Noise from the exciter area.
A lot of the noise is from the vent hole in the middle of the magnet, which is easily heard while putting your finger over the hole and and removing it again.
The noise is there on the pva panel but at a very low level and at a lower frequency, I think ?
Is this because the epoxy is more reflective(similar to an indoor swimming pool) does it cause bad ringing in the coil area.
If I replaced the epoxy in the coil area with pva would this prevent the problem ?
I could replace the centre area with a fabric dome or phase plugs but part of the problem could be the spider and coil former ? Also reacting with the epoxy panel?
Or all of the above ?
It started to make me think, years back when I used pva to replace the skin on my eps panels, because I thought the eps skin sounded bad ?
Was it these noises that reacted with the eps skin that made the eps skin sound bad in the first place ?
Lower grade or moulded eps does not have a hard skin I've noticed , do they use a different cutting method for the heavier grades, which seem to have a harder thicker skin ?
Pva certainly improved this problem on the eps, with a few other tricks, but was I curing the wrong problem ?
I've got a lot of playing around to sort these problems out.
all very interesting.
Steve.
 
Did a little more experimenting 3d printing trying the same material in same size, but with different density and thickness.
Nothing conclusive...have a measuring mic on the way and will see if it tells me anything I don't hear already. The PLA+ is bad...if I make it really thin it sounds a bit weaker and less defined, and if I make a thicker plate with more density it sounds a bit tighter but also starts getting a tone of its own. Not the same kind of dampable ring as I had with the polycarbonate though.
Tried a slightly different material, the Extrudr BioFusion which basically is a "Silk PLA", a PLA with added copolymers.
For both materials being PLA variations the difference is surprising. Silk PLA is mostly used for its shiny visual quality, and specs are similar to regular PLA, but known to be a bit brittle in comparison. However it actually has noticeably better HF response, so might be some hope to find a decent material.

Also been reading up on wet layup with vacuum bag. It has startup cost that is a bit off-putting with pump and additional materials, but when set up the actual expoxy, carbon and nomex would not be that much per panel, and obviously it would give better results that anything I can print.

Been using PVA based glue stick to fix the exciters, and been holding my hefty 200g exciters without problem, but today I got the idea to try hairspray. Have a bottle standing here since I tried using it as adhesive for 3d printing. It is also PVA based, and seemed like it could be a nice quickfix option.
Sprayed a thin layer, let it quickly dry so it was a little sticky and placed exciter. Did it a couple of hours ago, and have it hanging without support playing music already and seems stuck fine. Well see how it holds up, and if I can remove it later, but for a quick test it is excellent it seems.
Leob,
You have probably see in the previous posts our attempts to link the observations to material characteristics. Do you have some information about the material you use? Basically the density, the Young (tensile) modulus, the panel dimensions... In your case there is perhaps an additional difficulty which is your membrane might be not fully homogeneous or with some ribs.
About the hairspray, I didn't imagine to use it as glue! I don't have that at hand for now but I will try. Let us know about the long term performance and about the possibility to use it for fast change between different membranes. The double side tape I use has a low high frequencies performance. If you want to check the performance of this glue, the plywood is a good candidate; we know it can gives easily high frequencies.
Christian
 
Christian and Eric.
So , I have two identical corrugated cardboard panels, but one has pva coating and the other epoxy.
The epoxy panel suffers from Noise from the exciter area.
A lot of the noise is from the vent hole in the middle of the magnet, which is easily heard while putting your finger over the hole and and removing it again.
The noise is there on the pva panel but at a very low level and at a lower frequency, I think ?
Is this because the epoxy is more reflective(similar to an indoor swimming pool) does it cause bad ringing in the coil area.
If I replaced the epoxy in the coil area with pva would this prevent the problem ?
I could replace the centre area with a fabric dome or phase plugs but part of the problem could be the spider and coil former ? Also reacting with the epoxy panel?
Or all of the above ?
It started to make me think, years back when I used pva to replace the skin on my eps panels, because I thought the eps skin sounded bad ?
Was it these noises that reacted with the eps skin that made the eps skin sound bad in the first place ?
Lower grade or moulded eps does not have a hard skin I've noticed , do they use a different cutting method for the heavier grades, which seem to have a harder thicker skin ?
Pva certainly improved this problem on the eps, with a few other tricks, but was I curing the wrong problem ?
I've got a lot of playing around to sort these problems out.
all very interesting.
Steve.
Steve,
I see two aspects in your observations.
The first one is the sound from the vent which is different in frequency and level between a PVA skin and an epoxy skin. Here we can imagine that the epoxy with a higher stiffness as a higher resonance frequency in the voice coil area; perhaps also a higher Q (a narrower and higher peak) because of a lower damping. Have you tried to estimate the resonance frequency of the volume in the voice coil area and the vent (Helmotz formula?)? It would be bad luck there is a resonance of this cavity just where the one of the membrane is, or some bad relation!
The second aspect is about the link with EPS. My opinion at the moment is to say it is an other phenomena. I think PS, perhaps mainly EPS has a self noise. You can hear a little characteristic noise while bending it manually.
This is opinion. Refer to my post 4656 about spectral contamination.
From our previous experiments, a mass at the exciter axis damps the membrane resonance in the voice coil area. In the way to adopt this countermeasure, I was thinking to put the mass on the voice coil side to avoid to have it visible.
Applying that should tell you more about the origin of the noise?
An other possibility is to put some absorbent at the vent output; high frequency is quite easy to attenuate no?
Last comment : do you think that a rear emission at high frequency has enough energy to have an effect on the front?
Christian
 
Today, I made a quick experiment with a 5mm XPS with 2 Kraft paper skins glued with watered PVA. No hope to find the holy grail just the possibility to collect data and see how it fits in the overall picture.
In addition, as I applied the PVA on the XPS, the paper got some wrinckle. Poor realization quality.
The panel was 20x60cm.
I will be able to use the information about the efficiency in one week. No balance at hand here...
Here are 2 other measurements.
The spectral contamination. The worst of all!
blue in background is the panel under test, red the noise floor, green a Visaton FRS8 as reference.
This test really easy to do seems good to assess objectively of the material quality
1645213829762.png

1645213890887.png

A look to the FR
red is at 1.5m, green is at proximity of the exciter, blue is at proximity of the membrane 20cm from the exciter. I can't say the distance from the panel is the same for the 2 proximity measurements.
Some quick observations :
  • Both proximity points show low frequencies that are not at distance. It was already seen. The previous panels measured outdoor show an increase in low frequency compare to indoor. So there is a mechanism of the front and rear waves recombination for which I have no theoretical model. Does somebody have?
  • The slow decrease from 1.5k in this case seems resulting of the increase of loss at distance from the exciter
  • At higher frequency, the loss at distance is even higher.
To be continued with other material and a better control of the proximity distance
1645214663765.png
 
Leob,
You have probably see in the previous posts our attempts to link the observations to material characteristics. Do you have some information about the material you use? Basically the density, the Young (tensile) modulus, the panel dimensions... In your case there is perhaps an additional difficulty which is your membrane might be not fully homogeneous or with some ribs.
I will document the panels when I get the measuring mic. My printer is limited to 300x300mm, and to have a little safety margin I made a 295x273 rectangle with 94% ratio rather than square.

The data for the better sounding material of the two I tried so far is here:
https://www.extrudr.com/filerpool/download/datei/758/
The other material is a cheap PLA+, and doesn't have a data sheet. PLA+ is not actually a type of plastic, and only means PLA with additives. Typically those additives are making the material slightly more elastic since pure PLA can be a bit brittle. But most PLA based materials have a tensile modulus a bit over 3000 MPa. I have a roll of carbon material on the way as well, which has more then double the tensile modulus of PLA: https://www.extrudr.com/filerpool/download/datei/763/
There are materials with more extreme specs, for example CA-PAHP: https://treedfilaments.com/3d-printing-filaments/pahp-carbon-3d-printing-filament/
However they can require a little bit more advanced setup than mine to print successfully.

I'm still a bit skeptical that any regular 3d printed material will be good enough for it to be worthwhile, but it is very interesting to be able to experiment with the thickness and density so easily. So far the surprise for me is that using different types (honeycomb, rectilinear, etc) and density of infill as well as varying thickness seems doesn't make as much difference in sound as the material used, even when using two relatively similar plastics.
Also, efficiency seems to vary much less than expected and didn't notice much difference between for example a 1mm thin plate with 0.2 mm thin surfaces and sparse infill and a much heavier 2.4mm plate with very dense infill. Also here the material seemed to make more difference, with the BioFusion being louder than the PLA+.
 
I will document the panels when I get the measuring mic. My printer is limited to 300x300mm, and to have a little safety margin I made a 295x273 rectangle with 94% ratio rather than square.

The data for the better sounding material of the two I tried so far is here:
https://www.extrudr.com/filerpool/download/datei/758/
The other material is a cheap PLA+, and doesn't have a data sheet. PLA+ is not actually a type of plastic, and only means PLA with additives. Typically those additives are making the material slightly more elastic since pure PLA can be a bit brittle. But most PLA based materials have a tensile modulus a bit over 3000 MPa. I have a roll of carbon material on the way as well, which has more then double the tensile modulus of PLA: https://www.extrudr.com/filerpool/download/datei/763/
There are materials with more extreme specs, for example CA-PAHP: https://treedfilaments.com/3d-printing-filaments/pahp-carbon-3d-printing-filament/
However they can require a little bit more advanced setup than mine to print successfully.

I'm still a bit skeptical that any regular 3d printed material will be good enough for it to be worthwhile, but it is very interesting to be able to experiment with the thickness and density so easily. So far the surprise for me is that using different types (honeycomb, rectilinear, etc) and density of infill as well as varying thickness seems doesn't make as much difference in sound as the material used, even when using two relatively similar plastics.
Also, efficiency seems to vary much less than expected and didn't notice much difference between for example a 1mm thin plate with 0.2 mm thin surfaces and sparse infill and a much heavier 2.4mm plate with very dense infill. Also here the material seemed to make more difference, with the BioFusion being louder than the PLA+.
Leob
The last experiments I made with DML show :
  • the efficiency is almost not or even not at all link with the thickness of the panel. Your experience goes also in this way I think. The thickness combines with the area gives the low frequency.
  • the efficiency is linked to a parameter named T in a patent. T = D/µ³ = E/12/rho³
  • without any scientific link, I make an heuristic between T and a pseudo efficiency, let say a way to compare to a standard conic loudspeaker : eff = 5.log10(T) + 83
  • Patents recommend a T=10 so 88dB/watt not too bad and better T=100 so 93dB/watt
  • Typically the BeoFusion is given for E=3200MPa and rho=1250kg/m³ so T is about 0.13. It is similar to acrylic. The evaluation of the efficiency says 78dB/watt.
  • In comparison the plywood which my best reference of affordable and "easy to find" material has a E 3 times higher and the density 2.5 times lower. T "jumps" to 5; which is better but not enough.
So I come to the same opinion than you, those materials seem not really suitable for this application.
I think there is at least an other parameter to explore which is the absorption of the waves by the membrane. As mentioned by Spedge, all the materials are not the same in the "how far" the wave propagates from the exciter to the edge. We can imagine if the absorption becomes too high with the frequency, the high frequencies will have a lower emission area so a lower level at distance. At the opposite, if the absorption is too low, the resonances will become too sharp and the frequency response not smooth enough. This is the next thing I would like to approach with the in room loss of low frequency.
Any hints about the absorption according to the frequency in materials is welcome
Christian
 
I guess that what I realized is that changing structure of the material doesn't really change material properties in any significant way.

If I print with a 50% rectilinear infill I end up with an object that seems solid but have half the density of the material used.
But of course all the other properties goes with it, so stiffness is also reduced, making the end result pretty much the same as a solid object with half the thickness. A honeycomb infill for example should have a slight advantage with more compressive strength in the direction that matters, but seems like the effect of structure, at least with plastics, is really quite marginal. At least that is my initial impression after some quick subjective experimentation.

I only had the 6mm polyacrylate to compare with so far, but the printed plate sounds both louder and better, and if it was for home use only and I didn't care that much about efficiency it does seem to be some hope of making a good sounding panel. And the steampunk copper plate I printed looks really great visually! Maybe hard to say from photo, but really looks like a shiny copper plate.
IMG-20220219-125610.jpg
 
I guess that what I realized is that changing structure of the material doesn't really change material properties in any significant way.

If I print with a 50% rectilinear infill I end up with an object that seems solid but have half the density of the material used.
But of course all the other properties goes with it, so stiffness is also reduced, making the end result pretty much the same as a solid object with half the thickness. A honeycomb infill for example should have a slight advantage with more compressive strength in the direction that matters, but seems like the effect of structure, at least with plastics, is really quite marginal. At least that is my initial impression after some quick subjective experimentation.

I only had the 6mm polyacrylate to compare with so far, but the printed plate sounds both louder and better, and if it was for home use only and I didn't care that much about efficiency it does seem to be some hope of making a good sounding panel. And the steampunk copper plate I printed looks really great visually! Maybe hard to say from photo, but really looks like a shiny copper plate.
IMG-20220219-125610.jpg
Very nice and clean realization. Congratulations
The figures from my previous post are related to plain membrane. Your structure is "airy" like a composite so the efficiency performance is improved. Do you have an idea of characteristics like its weight, area, areal density, bending stiffness or at least its first mode in order to see how it fits with other data?
 
Leob .
I like the idea of the ring mountings in the corners.
I can't tell from the photos but is the panel surface flat or are those bubbles or dips on the surface ?
as the exciter should have a flat area to mount to.
Although , thinking about it maybe printing a mounting ring for the exciter former ring to mount onto could be interesting ?
maybe making it a little higher to increase the distance of the spider from the panel surface might also help things ?
Steve.
 
Very nice and clean realization. Congratulations
The figures from my previous post are related to plain membrane. Your structure is "airy" like a composite so the efficiency performance is improved. Do you have an idea of characteristics like its weight, area, areal density, bending stiffness or at least its first mode in order to see how it fits with other data?
Thanks. Yes, the efficiency should be better with it not being 100% solid, but I think it does surprisingly little. I think it might be different in an actual composite using a core material with very high compressive strength rather than basically just filling a single material plate with extra air, even if it is in a pattern that should contribute proportionally more to strength in relation to weight in a particular direction.
No doubt worth doing more experiments with some actual measurements, but it seems to me like finding the right material to print would be more important than the structure of what you print out with it.
 
Leob .
I like the idea of the ring mountings in the corners.
I can't tell from the photos but is the panel surface flat or are those bubbles or dips on the surface ?
as the exciter should have a flat area to mount to.
Although , thinking about it maybe printing a mounting ring for the exciter former ring to mount onto could be interesting ?
maybe making it a little higher to increase the distance of the spider from the panel surface might also help things ?
Steve.
Yes, things like easily making nice mountings are neat benefits of printing.
Been thinking about mounts for exciters and weights, or even varying density or thickness in places to control modes.
Or a little cone in front of exciter like this guy (although I'm not sure about the benefit): https://www.instructables.com/Plywood-DML-Speakers/

The surface structure is on the front side only, backside is flat like a mirror. At the bottom layer, the plastic is pressed against a glass surface during printing, while on top it is just laid down line by line on top of the honeycomb infill.
With only 0.2 thick surface on top, the core pattern shows through, but don't think it affects performance.
 
Christian.
I tried to show the plot of the spike from the rear vent but it was being swamped by the wide band noise coming from the exciter.
which looked very similar to your spectral contamination in blue on the plots you posted ?
with small panels ,you do not have a large barrier(the panel) blocking the airborne radiation from the back of the exciter.
So this sound is easily heard.
with large panels you may only hear the reflections from the rear wall ?
but these noises (distortions) will also come though the panels ,how much, depending on the type of panel used.
And of course these same distortions will be transmitted through the coil former to be radiated by the panel itself.
we have to make a distinction between panel noise and exciter noise .
There are various methods I have used over the years, to try and sort these problems out.
But the spider on my exciters are a tricky one to solve.
The harder I have to drive the exciter to move a heavier panel, the more the spider becomes a problem.
Steve.
 
Leob and Christian.
The link to the gals panels ,which I have seen before, shows a video of the panels in action .
As can be heard , there is a lot of prominent hf going on , which could be the whizzer cone in action ?

Chistian.
I thought you might find this video interesting ,which I found from leobs link.


Far too much hf from this type of panel , you could use eq, but will probably still have a hard sound ?
Steve.
 
Leob and Christian.
The link to the gals panels ,which I have seen before, shows a video of the panels in action .
As can be heard , there is a lot of prominent hf going on , which could be the whizzer cone in action ?

Chistian.
I thought you might find this video interesting ,which I found from leobs link.


Far too much hf from this type of panel , you could use eq, but will probably still have a hard sound ?
Steve.
Thank you Steve
The treble level is incredibly (unbelievable?) high. Does somebody else knows this level of trebel for acrylic? Something wrong in the set up like a control on?
 
Christian.
The acrylic is easier to drive in the high frequencies than in the lower frequencies.
Some panels have a downward sloping response to the high frequencies.
This panel has an upward sloping response.
Some panels have a hump in the mid frequencies.
Steve.
 
Was watching some Tectonic videos, and found some interesting parts:

First you can see him handling one of their actual plates in the very beginning. Hard to tell if it is 2mm or 3.4mm thick nomex, but not thinner than 2mm I would say.

Also they claim 104dB sensitivity!
That is insanely high and some 10dB higher than I expected. I'm not expecting a home made panel to equal theirs in sensitivity, but with those figures, even if I can get half the sensitivity (101dB) I would be more than happy.

Also I think in part two of that video they mention challenges creating a 100w exciter, so it seems like indeed they do have some extra powerful exciters they reserved for their own products.
 
Last edited:
Was watching some Tectonic videos, and found some interesting parts:

First you can see him handling one of their actual plates in the very beginning. Hard to tell if it is 2mm or 3.4mm thick nomex, but not thinner than 2mm I would say.

Also they claim 104dB sensitivity!
That is insanely high and some 10dB higher than I expected. I'm not expecting a home made panel to equal theirs in sensitivity, but with those figures, even if I can get half the sensitivity (101dB) I would be more than happy.

Also I think in part two of that video they mention challenges creating a 100w exciter, so it seems like indeed they do have some extra powerful exciters they reserved for their own products.
In the video How DMLs Work Part 2 (see at 24sec), the thickness is said to be 3.5mm which from previous posts seems difficult to link with the bending stiffness. 2mm might be a better a assumption?
About the efficiency 104dB which make sense for this kind of application is incredibly high for a DML. In the DML 500 spec, the only figure is 91dB.
If the heuristic about efficiency is not completely wrong 91dB is reachable with a 0.5kg/m² material. This is a low sensitivity for such an application.
104dB means (again IF the heuristic make sense) a D/µ³ = 16000!
The assumption on the fc gives D/µ around 20 so µ = 35g/m². Something wrong somewhere no? Is it possible to reach so low weight?
It would be interesting to find scientifically based efficiency evaluation.
One data not used is the power handling of those panels. From your experience of sound system, does this give an other evaluation of the efficiency?
 
Ah, 3.5 or a bit over makes sense including the skin if they are using a 3.4 nomex core, which seems to be a very common thickness.

I guess what they specify in the data sheet is correct and the video is wrong 🙂
I'm not a sound pro, or at least I never worked professionally with this kind of pro audio systems. I studied them many years ago as part of my sound engineering studies, and rented and handled systems many times. So I guess most of you know more about speaker sensitivity and how to measure it. For me, if one just says sensitivity, I would say it means 1w at 1m on axis, but neither the video or data sheet mentions what actual measurement is referred to.

If thickness doesn't really affect efficiency that much, then surely kg/m2 would be a bad indicator of efficiency as well?
At least what I can hear with my ears from my printed part so far, perhaps kg/m3 matters for the sensitivity, but kg/m2 has surprisingly little correlation.
 
Ah, 3.5 or a bit over makes sense including the skin if they are using a 3.4 nomex core, which seems to be a very common thickness.

I guess what they specify in the data sheet is correct and the video is wrong 🙂
I'm not a sound pro, or at least I never worked professionally with this kind of pro audio systems. I studied them many years ago as part of my sound engineering studies, and rented and handled systems many times. So I guess most of you know more about speaker sensitivity and how to measure it. For me, if one just says sensitivity, I would say it means 1w at 1m on axis, but neither the video or data sheet mentions what actual measurement is referred to.

If thickness doesn't really affect efficiency that much, then surely kg/m2 would be a bad indicator of efficiency as well?
At least what I can hear with my ears from my printed part so far, perhaps kg/m3 matters for the sensitivity, but kg/m2 has surprisingly little correlation.
Leob
All the experiences are interesting! My approach of efficiency is also "basic" viewing it as the transfer function between a power of 1 watt as input and the pressure level at 1m.
The tests I made shown a quite good correlation to the bending stiffness (named D in previous post or B in other, the unit is N.m or Pa.m³) divided by the aerial density in the cube (named µ or in some paper m in (kg/m²)³). It is a quantity named T in some patents. The most difficult value to get is D. You will find in post 4185 page 210 2 methods for that. The aerial density is easy to have with a balance. The SPL is not too difficult to get in relative value to a known loudspeaker. REW (Room Eq Wizard) can be used as pink noise generator for that. The SPL applications of mobile phones can also help to get a plausible absolute level.
To come back to the Tectonic specification it is also mentioned the maximum rating of 120dB SPL, 200W continuous electrical power. With a sensitivity of 104dB no need to reach 200W.