I think it is very doable with quite easy means to make a design that holds up to handling but still is very competitive when it comes to SPL per kg, also when using EPS. Regardless of material used you need to have some kind of grilles if you want it to be safe for road abuse, but even with steel grilles it will be lightweight compared to a regular speaker.
Last edited:
Hello Andre
Your results are very impressive! Excellent!
I have chosen this posts among the different you wrote because you describe here what is the "magic" of those panels.
After about 2 years playing with such panels, I see them as wide range speakers with an excellent dispersion almost all over their range but with a perfectible (at least!) FR when pistonic speakers shows good on axis FR with a perfectible dispersion.
The characteristic of your compound panel is its FR. It reminds me the good results of canvas panel which are not so far in their principle.
I have some questions (of course!) :
- Could you detail the paper coating? which paper? the gluing method? on both side?... I made some tests about that but with no success
- The polycarbonate disc seems made from those sheets with flutes from DIY store. Correct? Thickness?
- Could you confirm my understanding the EPS is clamped (glued) on the long side in the left and right side and free on the upper and lower side? Among the DIY DML the 2 main techniques are suspended panels or suspension that try to mimic simply supported edges. Clamping technique tend to increase the frequency of the 1st mode. What leads you to this solution? Seing your pictures, there is at least a nice simplicity of construction.
- The exciter is at the center of the disc which seems of center of the panel. Could you precise the "off center"? Classical 2/5?
- Efficiency is a key point for DML. If it is not too difficult to get low to medium efficiency (to make a comparison with pistonic speaker), good to high efficiency is challenging (need a stiff and light material). EPS is good in that (if not too weight is added by the coating), plastics like polycarbonate are not good (density plays by its cube in the efficiency). Have you estimated the efficiency of your panel? I do that by setting the volume of my amp as usual, doing a first measurement with REW with a known speaker and then with the panel. Not highly scientific but gives a good idea.
- Could you post the spectrogram (ie wavelet 1/6 octave) from REW of the FR you shown?
- Last, sorry to be long... could you detail a little bit your calculation that links wave speed in the panel, in the air and panel dimension to the accident in the FR. As you wrote, the wave speed in the panel is frequency dependant (low in the bass, high in HF with some where what is called coincidence frequency where it equals the speed in the air)?
Christian
Okay, but if you decreased the panel it drives would you not potentially get the FR you want? Sure, two exciters, one as Andre has implemented and another more powerful one driving a larger detached disc bonded to the same panel in a similar way. Say a 6-8" disc. Now you have a lot more power driving the bottom end too.
Hi Christian,
Thanks for your questions and comments.
- The paper coating is simple Kraft paper. Density 0.98g/cm3. Thickness is approx 0.065mm. Yes, glued to both sides of the panel. I used Alcolin Ultra wood glue to adhere it to the EPS. I did notice that the amount of glue used does make a non-trivial difference to the FR. It's worth experimenting with this.
I didn't like the floppiness of the EPS, which is only 15mm thick, so at first I used aluminium foil to stiffen it. But because both the foil and the EPS are non porous, the glue was not exposed to air and it would not dry, and I could not get around to testing. The paper is nicely porous, and the glue dried in a few hours. Maybe in another iteration I'll use an epoxy-based resin and try alu foil again.
- The polycarb is 4mm TwinWall. It's used for awnings and roofing. But you don't have to use Twinwall. I wanted a thin, stiffer disc for the higher frequencies, because I suspected that the EPS in front of the driver was absorbing or damping too much HF. So even something like a CD should work. I think the hole in the CD might get rid of the enclosed cavity in front of the driver VC and flatten out that notch at 10k.
- Correct, the panels are clamped left/right, and unsupported top/bottom. There was no fore-thought about this, it was merely for convenience. I have tried all the variations on other panels, and there seem to be very arbitrary pros and cons to each arrangement.
- Yes, the exciter is centred on the disc, and the positioning is indeed the classic 2-5ths/3-5ths arrangement.
- I have not measured efficiency yet. I need to solder up a rig to measure 1VRMS signals with REW, and calibrate SPL somehow. Cell phone SPL meters are not too bad. Maybe I'll use that as a basic reference.
Having said that, the EPS panels I've tested so far are definitely louder than any other material that I've tried, but not as loud as the LD MAU11 PA column that I'm using as a rough audio reference.
- This spectrogram is 1/12th Oct.
- Panel size vs wavelength... this is tricky because there are so many interrelated variables. One can only make very rough correlations. For example,
Acoustic SS through polycarb is approx 2250m/s. At 10kHz this gives you a 112mm half-wave, which is exactly where I see the notch in my FR curve. But correlation is not causation, so I need to check this out more carefully. There are notches at the sub-harmonic frequencies too, but there are so many panel resonances and reflections, I think time would be better spent looking at other issues.
AIR SS is 340m/s and at 10kHz the wavelengths are in the region of 34mm. Maybe this correlates with the driver VC cavity? I don't know. It needs closer inspection.
I haven't spent too much time on panel size vs notch frequencies except to notice in passing that there are correlations.
I also have not spent any time at all working out coincidence frequencies. But maybe I should.
Cheers,
Andre.
Hello Andre,
Thank you for this quick and detailed answer.
Your kraft paper is 64g/m². It is said to be a strong paper. You can have a look to this post 5594 or 5599 from Eucy about the use of a "yellow trace paper" which is only 27g/m². Not tested on my side but I made the same experimentation than you with non porous material...
About the role of the disc, you might have a look to this patent US20020094095A1 (see my post 6714). It shows the presence of a low pass filter between the voice coil mass and the panel mechanical impedance. Materials like EPS which have a low density have also a low mechanical impedance so a lower capacity to go in the treble. By enlarging the diameter of the which the exciter force is applied, the mechanical impedance seen by the panel increases. I wonder if your dick is working in fact as a coupler or a kind of mechanical impedance transformer, in addition to be by itself a vibrating element.
This patent shows also the role of the central area which becomes a source when the panel is going to stop its emission. It explains the 10k peak (or about) we have on almost all the panel. If there is a hole in the central area, this emission can't exist; Those 2 ranges of frequencies seem needing different solution. Some calculations of the drum behavior of the central area might be interesting. A paper (don't remember which one) suggest to increase the stiffness in the area inside the voice coil diameter.
You are I think right also saying having the hole will stop what Spedge calls the exciter noise which is the air pushed to the rear by the panel inside of this area. See 6712.
What is remarkable in the FR of your panel is there is no areas where we can say this is the panel, this the disc. Like if both blend perfectly (blend, is it the right word... some limitations in English...); I see few chance that by luck, the 2 areas match perfectly... Like if we can have a good chance to associate a mid-range and a tweeter choosing them almost randomly.
Thank you for the spectrogram.
Here is below the spectrogram of a perfect IR (flat FR). Probably no speaker will give that. It is just to see in comparison the "ringing" frequencies which is typical of DML. We see clearly in your spectrogram that the low frequency modes remain a long time. In this other thread there are some exchanges about the possible consequences (audibility) of such "long tail". Some EQ or LF filtering might change that (if needed...)
About the wave length, you makes the evaluation with a constant value of the speed (2250m/s). Is it an approximation? A limitation of the wave speed due to the shear effect? The wave speed in a plate changes with the frequency no?
I hope you will find better weather condition to go further.
Christian
Thanks Christian!
The acoustic S/S for polycarb is around 2260m/s. It's a standard material property. But this is the value of the material itself, and not of a surface wave. S/S of water, for example is 1,480m/s. But the waves over the top of the water move at much much lower speeds, also depending on wavelength. So I think it's a given that bending waves (surface waves) over a panel will move at a different speed to the internal speeds, and at a different speed to S/S in air, and at different speeds according to wavelength...
I'm afraid it feels like my brain oozes out of my ears when I try to analyse all of the various variables. It would take me weeks to work this all out with whatever models and simulations I use, often incorrectly. But it takes me only a day or two to build the stuff and to blindly try different things in the hopes of building a reservoir of experience and 'feelings' for how it all works together.
I did try out a CD disc instead of the polycarb twin-wall disc. The results were disappointing. This is a balancing act between the properties of the disc material, the number of legs connecting to the panel, the panel material itself, and obviously also the characteristics of the driver. All of these interact with each other.
This is a massively complex subject.
Did the first trial with DML myself.
Plates 48x60 cm, exciter places on 2/5 both ways.
The panels are 10mm EPS with thin wall fiberglass on both sides, and glued with wall glue which is predominantly PVA.
Putting the two thin laminates on the panel made them quite stiff.
Put them into mu listening room and played them in stead of my main speakers.
First experience:
Very efficient, so a lot of sound.
Sound quite thin, meaning a lot of mid frequencies ... consistent with all the measurements I have seen.
I measured the impedance and the panels with the exciter have a lower resonance at 40 Hz
I crossed over to a sub for most of the listening.
I'm sure the freq response can be eq'ed to ruler fit and this would of course have a tremendous impact on the sound.
But what really strikes me is the "stereo perspective or sound stage" .... it just sounds strange!!
I have watched a lot of videos and got the exact same feeling, but really wanted it to be down to recording, but I have a feeling it is inherent in the DML principle.
In a way it sounds like the channels are out of phase ... this strange feeling you get if you play swap the phase on near field speakers (have tried it with my PC speakers as this is super easy as they use DSP.
It's a bit like that.
Or like the sound comes through a tunnel .... but not really the sound you get through a smaller tube (which results in comb filter effect)
And yes I have checked the phase of the panels ... they are correctly connected
Does anyone else have the same experience??
PS. the picture is taken in my office, just for showing the buildup. In the listening room they where also on the floor on a carpet and resting up the wall with a sock in between to make for a compliant coupling
Plates 48x60 cm, exciter places on 2/5 both ways.
The panels are 10mm EPS with thin wall fiberglass on both sides, and glued with wall glue which is predominantly PVA.
Putting the two thin laminates on the panel made them quite stiff.
Put them into mu listening room and played them in stead of my main speakers.
First experience:
Very efficient, so a lot of sound.
Sound quite thin, meaning a lot of mid frequencies ... consistent with all the measurements I have seen.
I measured the impedance and the panels with the exciter have a lower resonance at 40 Hz
I crossed over to a sub for most of the listening.
I'm sure the freq response can be eq'ed to ruler fit and this would of course have a tremendous impact on the sound.
But what really strikes me is the "stereo perspective or sound stage" .... it just sounds strange!!
I have watched a lot of videos and got the exact same feeling, but really wanted it to be down to recording, but I have a feeling it is inherent in the DML principle.
In a way it sounds like the channels are out of phase ... this strange feeling you get if you play swap the phase on near field speakers (have tried it with my PC speakers as this is super easy as they use DSP.
It's a bit like that.
Or like the sound comes through a tunnel .... but not really the sound you get through a smaller tube (which results in comb filter effect)
And yes I have checked the phase of the panels ... they are correctly connected
Does anyone else have the same experience??
PS. the picture is taken in my office, just for showing the buildup. In the listening room they where also on the floor on a carpet and resting up the wall with a sock in between to make for a compliant coupling
Attachments
Yes a complex subject! The challenge is to find the right level of theoretical approach to avoid being in an endless trial and error. We can see also loops in the long threads... ideas or questions appear again so all the tentative of of design rules or guide lines are welcome, even empirical.
Thank you for the test of the CD. This is really the difficulty of those panels. When a result is encouraging, what can be the next step, which variation is possible.
Christian
HelloDid the first trial with DML myself.
Plates 48x60 cm, exciter places on 2/5 both ways.
The panels are 10mm EPS with thin wall fiberglass on both sides, and glued with wall glue which is predominantly PVA.
Putting the two thin laminates on the panel made them quite stiff.
Put them into mu listening room and played them in stead of my main speakers.
First experience:
Very efficient, so a lot of sound.
Sound quite thin, meaning a lot of mid frequencies ... consistent with all the measurements I have seen.
I measured the impedance and the panels with the exciter have a lower resonance at 40 Hz
I crossed over to a sub for most of the listening.
I'm sure the freq response can be eq'ed to ruler fit and this would of course have a tremendous impact on the sound.
But what really strikes me is the "stereo perspective or sound stage" .... it just sounds strange!!
I have watched a lot of videos and got the exact same feeling, but really wanted it to be down to recording, but I have a feeling it is inherent in the DML principle.
In a way it sounds like the channels are out of phase ... this strange feeling you get if you play swap the phase on near field speakers (have tried it with my PC speakers as this is super easy as they use DSP.
It's a bit like that.
Or like the sound comes through a tunnel .... but not really the sound you get through a smaller tube (which results in comb filter effect)
And yes I have checked the phase of the panels ... they are correctly connected
Does anyone else have the same experience??
PS. the picture is taken in my office, just for showing the buildup. In the listening room they where also on the floor on a carpet and resting up the wall with a sock in between to make for a compliant coupling
I would say that a good and stable sound stage is one characteristic of DML. In our living room, the panel being just laying on the previous my 3 way, that's ok. Better than my 3 ways. The strange think is the sound is a bit behind the panel plan but not far. I think this was already mentioned in the different thread.
Due to the rear wave, the panels need a certain distance from the wall (often said at least 30cm).
Have you tested the panels away from the walls. Like on a chair with some means to avoid noise at the contact points.
Yes they will sound a little thin and 'phasey' when placed against a wall. Give them some breathing room, and then you will hear that there's no pistonic speaker that can match them for dynamics, transient response or sound stage.Did the first trial with DML myself.
Plates 48x60 cm, exciter places on 2/5 both ways.
The panels are 10mm EPS with thin wall fiberglass on both sides, and glued with wall glue which is predominantly PVA.
Putting the two thin laminates on the panel made them quite stiff.
Put them into mu listening room and played them in stead of my main speakers.
First experience:
Very efficient, so a lot of sound.
Sound quite thin, meaning a lot of mid frequencies ... consistent with all the measurements I have seen.
I measured the impedance and the panels with the exciter have a lower resonance at 40 Hz
I crossed over to a sub for most of the listening.
I'm sure the freq response can be eq'ed to ruler fit and this would of course have a tremendous impact on the sound.
But what really strikes me is the "stereo perspective or sound stage" .... it just sounds strange!!
I have watched a lot of videos and got the exact same feeling, but really wanted it to be down to recording, but I have a feeling it is inherent in the DML principle.
In a way it sounds like the channels are out of phase ... this strange feeling you get if you play swap the phase on near field speakers (have tried it with my PC speakers as this is super easy as they use DSP.
It's a bit like that.
Or like the sound comes through a tunnel .... but not really the sound you get through a smaller tube (which results in comb filter effect)
And yes I have checked the phase of the panels ... they are correctly connected
Does anyone else have the same experience??
PS. the picture is taken in my office, just for showing the buildup. In the listening room they where also on the floor on a carpet and resting up the wall with a sock in between to make for a compliant coupling
There's also the lack of sweet spot. Since they are almost omni-directional, everywhere in the room becomes a sweet spot.
Have fun!
https://www.easycomposites.eu/foam-cored-carbon-fibre-panel
Have you looked into easycomposites range of foams? I’ve ordered stuff from them to Sweden thru their Eu site.
https://www.easycomposites.eu/rigid-foams
And have anybody tried to arrange their actuators in a straight line like in this paper?
https://publications.lib.chalmers.se/records/fulltext/61709.pdf
Thank you Nissep! This is very good info.
I'm trying to find better info on the PolyEther foam mentioned in that thesis by Redondo. It does not seem like the best material to me from what I've seen. But maybe the paper is a bit dated? Or maybe he was looking only at frequency response flatness at the expense of efficiency?
Apparently, actuators in a straight line like that are not a good idea according to Dayton/Tectonic. The actuators should be staggered in such a way as to activate the maximum number of modes across the panel. But it seems that Redondo is using piezo drivers. Not magnetics. And he's using 20 of them to increase SPL's. But the bandwidth he's getting is impressive.
I would rather use a normal actuator and then add maybe three or four piezo's if I wanted to boost the HF. I'd run the piezo's in parallel, and drive that combined load through a low-Q, lossy inductor to resonate at the target frequency. The piezo's present a capacitive load in the region of 50nF. You can connect as many as you like in parallel whilst keeping an eye on amplifier stability. One might have to add a Zobel network to mop up any stray reactances that would otherwise send the amp into orbital oscillation.
EasyComposites is a very nice site. Thank you!
Sure they used piezo speakers, but the part about damping the edges and material choice is interesting.
But if one like to replicate the exact speaker in the paper there is nice options.
https://product.tdk.com/en/products/sw_piezo/speaker/piezolisten/index.html
Wow those piezos have massive capacitances! That series resistor that they recommend is a very good idea.
Hello André
Is Rohacell in the familly of the PolyEther? I have pointed to it in previous posts in the "main" thread as it appears in different papers. I don't know this material, even how it looks like. If I remember from other papers, the material surface as to be smooth no porous (to be checked)
If it can help : The efficiency can be evaluated by E/rho³ or D/µ³. E Young modulus, rho density, D stiffness (=Eh³/12), µ areal mass kg/m² (=rho.h) h thickness.
Easycomposites was pointed in the "main" thread being a good source yes.
Christian
Dang it! Does this mean I need to read the whole “original” thread to find out if anybody tried the carbon fiber skinned foam from easycomposites?
Like most of the ready made carbon panels I fear the skins are a little bit thick. They say 0.5 mm, which might mean not that efficient. Could be worth a try, but not much cheaper than nomex/carbon composites which are available with a bit thinner skin (some Italian site linked in the full-range thread), and I would probably go with that first since it is a proven combination that Tectonic use.
My conclusion was that you probably need to make custom plates to get the right thickness of skin, and ordered equipment from easycomposites as well as thin carbon and nomex to do that. Since I was not successful with my e-glass experiments I haven't gotten around to trying out the carbon yet, but hope to get around to it soon.
Tectonic explains the reasoning for their placement in one of the videos on their youtube. Basically it helps enlarge the area working on piston principle for high frequencies, so becomes like one large exciter with a large central area rather than cancelling each other out....or something like that
Hello Nissep
2 possibilities before that :
- advanced research of this forum ?
- have a look to the history file. With the small arrow under my logo on the left of the post, follow the link to github.