A Study of DMLs as a Full Range Speaker

[homeswinghome]

Hi Christian, I do like a systematic approach with measurements to guide us. Thank you for sharing.

I agree the surface composition and its resulting stiffness is likely to determine the energy transfer to the panel from the exciter. I would have anticipated that the relatively soft surface ( compared to ply ) of XPS would need a large flat contact area. Is the surface under the soda can flat or is it a ring contact?
Composite panels typically have a stiff and hard skin to promote energy transfer so I would anticipate that the force concentrator approach would work with composites as well as ply but thats an unproven guess.

Burnt.

It is a ring contact (like the soda can on a table) glued with epoxy. The idea comes from a patent I shared some post ago. This patent claims (with no figures) the mechanical impedance increases with the length of the contact line.

 

[lenta]

Perfect Impulse


Commercial exciters need to be affordable in order to sell and so cost will define the performance of exciters to a large degree. If you pay $30 for an exciter taxation, distribution costs, retailer profit all reduce the cost available to the manufacturer for the component. DIY is free of all of those constraints so what might a clean sheet of paper design for an exciter look like if we removed cost constraints?

I have captured my initial thoughts below, but I would be very interested in any observations you might wish to make.

What are the areas of exciter design that may benefit?

1. Panel/exciter interface- can we do better than double sided tape? Can we get a benefit by increasing the attachment area to increase adhesion, or is decreasing the attachment area as a force concentrator a better way to go?
2. Voice coil design- 25mm diameter voice coils seem to suffer from less buckling strain than 32mm voice coils. But 25mm coils offer less coil length ( the L in F=BLI) and so, all other things being equal, less motor force. Can we either use better materials for 32mm coil formers- e.g Carbon fibre or Aluminium, or can we compensate for 25mm diameter coils reduced L by increasing B, the magnetising flux, or increasing I, the current delivered?
3. Suspension design. From what I can see this part of exciter design has received a lot of focus and performs quite well, however there is plenty of evidence of noise from some metal suspensions. To gain a performance enhancement perhaps we rethink this area completely?
4. Regarding the motor force F= BLI, current, the I in the motor force equation, is supplied by the user. For the manufacturer current is free, so it makes sense for the manufacturer to bias their motor design towards being a current user and, for a given F, reduce the magnet and amount of wide in the voice coil as these are a direct cost to the manufacturer. We can gain efficiency if we increase magnetic field intensity B and voice coil length L
5. Demonstrating that all design is a mater of balancing opposing constraints, increasing L increases the inductance of the voice coil which rolls off the HF response. Can we improve HF response by reducing L and increasing B to compensate for the loss in efficiency?
6. Newton tells us every force has an equal and opposite reaction. We usually absorb the reaction by mounting the exciter to a spline, or let the exciter mass provide a reaction mass which in my experience only works well at low to mid volume or where we limit the bandwidth of the exciter to omit bass frequencies. . Is it possible to harness the reaction force to drive a second panel?
7. Exciters need to be shipped worldwide and mass is expensive to ship. Freed from this constraint can we gain performance improvements by using heavier materials, e.g. metal instead of plastic for the body?
8. There are a lot elements in the panel/exciter system. Voice coil, voice coil suspension/spider, and sometimes two of them, voice coil drive plate, suspension frame or basket, the panel itself, which makes for a complex system. Each component alone is a potential source of distortion and there will be secondary effects cause by the interactions between the different components. Can we eliminate some components to simplify the system?
9. Can we improve the performance by using techniques new to exciter design e.g. the use of ferrofluid may improve heat transfer from the voicemail and potentially damp any unwanted voicemail resonances?
10. At this time I don’t know if the motor design uses a long coil/short magnet gap approach, but I suspect they probably do because magnets are much more expensive than copper wire so the manufacturer would want to reduce this cost. Displacement is not an issue for exciters as they are designed to impart an impulse to the panel to propagate surface waves rather than move the panels in a piston like mode, and so a short stroke for the voice coil is a given. But it is just possible that with such cheap components made in the millions there may be compromises here as well that could be removed. A long magnetising gap and short coil may offer a better balance?

Some possible routes to investigate. Two exotic options first.

1. Mounting the voice coil directly to the panel would eliminate the voice coil former which removes buckling strain noted in 2. above, removes the drive plate and provide a very stiff coupling between panel and voice coil. This would be a less efficient motor design than the conventional one and require a significantly stronger magnetic field to compensate.
2. Assuming a more conventional exciter design, can we provide a rear mount to allow a second panel to be attached to and driven by the exciter. This second panel could be of a size, shape, and material to extend, or augment the main panel. I note that a second full size panel would introduce interesting challenges, e.g. the gap between the two panels produced by a typical exciter would reduce the size of the wave form than can form between the two panels , i.e., the output would be bandwidth constrained.
3. Tapering the size of the drive plate to reduce the contact area could provide a force concentrator to improve the transmission of signal to the panel?
4. Freed of the cost constraints we can certainly revisit the balance of magnet size, voice coil length and current that F=BLI requires, e.g. a bigger magnet and smaller coil to improve HF response for a given force
5. Freed of the cost constraints we can consider stiffer materials for the voice coil former to reduce/eliminate voice coil buckling.

Practicalities.


Both speaker motor magnets and voice-coils can be bought online and motor cases and spiders can be printed using a 3D printer ( including carbon fibre filled materials) so experimentation is feasible, although it remains technically challenging. The biggest hurdles that I can see are more in the area of the motor design itself, simulating the magnetic circuit is a non-trivial exercise and measuring the results of each new design will be a challenge. However, all designs start off on a bench top somewhere so there must be a way. Also, some of the options can be tested by adapting existing exciters.


As usual all the above are shared with the DIY community under the copyleft principle used by the open source community. Anyone in the community is free to use the ideas discussed above and adapt them as they see fit. A small plea to the manufacturers that contribute to this website, please respect the copyleft principle and don’t be evil. Karma always wins in the end.

Burnt.

Tectonic in their dml panels uses exciters with copper winding on inside and outside of voice coil former. It is probably dificoult to manufacture, but gives several advantages, main are increased motor strenght(Bl) without increasing inductance.
Borresen uses it aswell.

New Tectonic TEBB5430-8 has unusualy large Bl with very low inductance, it can be used as a exciter.

Scan speak illuminator 12mu has one of the best motors in the world.
Dayton audio ES140Ti has great motor.

 

[BurntCoil]

It is a ring contact (like the soda can on a table) glued with epoxy. The idea comes from a patent I shared some post ago. This patent claims (with no figures) the mechanical impedance increases with the length of the contact line.

Hi Christian,

I think I am going to have to dig into the literature to find the parameters that determine the transmission of impulses to a solid. Theres a lot out there but its all very abstract. I will report back if I find anything useful to the group that we can use.

Burnt

 

[BurntCoil]

Tectonic in their dml panels uses exciters with copper winding on inside and outside of voice coil former. It is probably dificoult to manufacture, but gives several advantages, main are increased motor strenght(Bl) without increasing inductance.
Borresen uses it aswell.

New Tectonic TEBB5430-8 has unusualy large Bl with very low inductance, it can be used as a exciter.

Scan speak illuminator 12mu has one of the best motors in the world.
Dayton audio ES140Ti has great motor.

Hi lenta,

Thank you for these images and the references, there is a lot of useful information here. I have managed to find the website designer of some of the Tectonic exciters http://www.grahamlandick.com/ and one of his patents is full of information on his approach which is very inventive. https://uspto.report/patent/app/20190261092
I wonder if Mr Landick posts on here, quite a few professionals do.

Burnt

 

[spedge]

Lenta.
I could not find the TEBB5430-8 you mentioned?
Steve

 

[spedge]

Burnt.
Most of the problems I have with the exciter design can be reduced by controlling what goes on in the exciter area.
The use of a phase plug , fabric dome, or even something like the tannoy duel concentric with horn in the centre, could be an option.
The panel material and thickness would dictate which options are best for those panels.
Although I am very happy with the results of my exciters , I know they are not perfect.
The design is flawed , that is why piezoelectric transducers interest me, but until something better comes along I will keep trying to tweak the exciters to produce better sounds.
Steve.

 

[BurntCoil]

Hi Steve,

Your experimentation has been little short of heroic to date.

Burnt

 

[spedge]

Burnt.
I don't know about heroic, I'm still trying to figure out why I'm trying to make a CD sound good 😳 , as a panel that is 😄
Originally the CD sounded and behaved very badly.
But it's doing a lot better now, surprisingly.
I now have a small doped fabric dome on one CD and a undoped fabric dome on the other, they sound very different.
I'm trying to figure out which is right?
Acrylic is not one of my favourite materials, it has a bright Acrylic sound, so the doped tweeter is overdoing it a little.
There are a lot of problems to sort out with the CD, but it can work as a novelty speaker with a half decent sound ,if wanted?
I'm not going anywhere near an LP 🙄
Steve.

 

[BurntCoil]

It would be interesting and rewarding to learn how far you could push the best materials and exciters Steve

Burnt

 

[Veleric]

DML material chart

In previous posts, I shared a DML material chart. At this time, some aspects of the DML theory where not linked to it. Links with experience were also missing. The picture is better now.
Please find in the attached document this chart and the elements of introduction.
Thank you for the feedback to correct, improve it. After that, I will add it to the documents in the github page with the original Python script.

Christian

Christian,
Very interesting presentation. I like the magic triangle concept. I certainly agree that T>10 is a good target, although lately I'm feeling that 5 may be good enough for home applications. And fc>5 kHz likewise seems like a good target. But I'm not familiar with the cutoff frequency calculation based on the panel mechanical impedance. Is there a literature reference for equation 10 in section 2.4?

In 7.1 I understand the reasoning for the minimum values for fc and fhigh, but why the maximums? I don't see any harm for fc >7500, for example.

Also, in section 2.1, you might want to be clearer that equation 1 applies only to homogeneous materials, i.e. not composite sandwich constructions.

Eric

 

[Eucyblues99]

Hello Burnt,
This the beginning of a full functional analysis!
I had the possibility to make some tests this week. One of the target was to measure the effect of a coupling part. As you will see, it is not full conclusive on that point but interesting on another (at least to my point of view) and related to your list.
Here is a picture of the tested panels.
All the panels are 20x30cm XPS 9mm (depron)... I had a stock of this. In fact it is the second material I made panels which have never worked and I would like to understand why...
Left is a concentrator made of wood sticks, central is a bottom of soda can (44mm) to enlarge the contact diameter with the panel. Both are on the exciter side. Right is an Eucy's dome so in front of the panel.
View attachment 1104277 The results :
Red is panel without accessory, green the concentrator (left on the picture above)

View attachment 1104281
Red, the panel without accessory, blue with the 44mm dome as coupler
View attachment 1104338
Red, the panel without accessory, orange with an Eucy's dome
View attachment 1104339
What I understand from this test is the sensitivity of light membrane to additional mass - in better words, low mechanical impedance material.
Higher mechanical impedance membrane allows more flexibility on the voice coil mass. Unfortunately this is also low efficiency material, like plywood.
The design solution for the exciter might be strongly linked to the membrane material.
Christian

Hi Christian

Thanks for trying the dome

Seems like a significant boost in the 2k to 5k range?

But is that a hole I see in the centre or a black mark?

Eucy

 

[pway]

Tectonic in their dml panels uses exciters with copper winding on inside and outside of voice coil former. It is probably dificoult to manufacture, but gives several advantages, main are increased motor strenght(Bl) without increasing inductance.
Borresen uses it aswell.

New Tectonic TEBB5430-8 has unusualy large Bl with very low inductance, it can be used as a exciter.

Scan speak illuminator 12mu has one of the best motors in the world.
Dayton audio ES140Ti has great motor.

Interesting discussion about the possibility of DIY exciters. I raised this some time ago on the forum too.

I think it's probably very difficult to build a better exciter by competing in terms of manufacturing techniques. However there may be some low-hanging fruit stemming from the fact that we don't have to build a saleable exciter product, but a complete system.

For example, exciters have to compromise on field strength because one side of the exciter has to be open, with resulting field leakage. But we have a good deal of flexibility in the geometry of the panel, and are not limited by size because we can separate the moving parts from the heavy parts carrying the magnetic flux. You could, for example, put a hole through the panel, glue the coil & former inside that, and pass a soft iron core clear through the hole, with neodymium magnets on both sides. Of course it does need to be supported on both sides, but the higher field strength & better geometry should mean that the tolerances are less important. Maybe you could also think about eliminating the coil wiring altogether by using induction coils, at least for low frequency.

 

[aagas]

I had some things sitting around my workshop so I decided to put together another set of DMLs over the weekend to sit on either side of a TV.
See attached.

Dimensions

I just set the dimensions to fit with the TV rather than anything else.
The frames measure 16" w X 34 3/4" h

Frames

For the frames, I used bits of Abstracta modular framing system that I had from an earlier project.
Steel tubes (1/2" OD) and die-cast metal-alloy connectors.
You can cut the tubes to size with a small pipe cutter.

I considered filling the tubes with sand or spray foam. Probably overkill.

I suspend the DML panels within the frames; isolated on small felt pads.
Not very stable yet. I'll try and come up with something better. Suggestions appreciated.

I thought these frames could prove useful to easily try out other materials.

Panels

I used 3/8" Gatorboard for the panels and Thrusters as exciters (I had these sitting around too).

Amps

I hooked up an old pair of Hypex FA123s (only using one of the there channels from each amp) to drive the exciters.

Measurements to come...

No measurements yet, so no DSP tuning yet.
They could use some leveling up on the low end and will likely need some help elsewhere.

First listening session,

They have air and space. They have a convincing soundstage with width and depth and pretty precise placement of performers. Great imaging. They disappear sonically. Most importantly they engage one in the music.

They have all the volume capacity I'd ever need for this system.

More to come...

 

[homeswinghome]

Hi Christian

Thanks for trying the dome

Seems like a significant boost in the 2k to 5k range?

But is that a hole I see in the centre or a black mark?

Eucy

Hello Eucy,
It is a 2mm hole in the center. The way I made them is a bit different from yours. I made test on different samples of material I had : EPS, XPS, plywood, "homemade plywood".
In few words :

• Very easy to do. As I drilled also a 2mm hole in the panel and the exciter have also a 2mm hole, it was very easy to align them.
• For low density materials (EPS, XPS) there is the boost you mention but the FR stops a bit before then
• For the high density materails (plywood or similar)... The on axis FR was not changed in an obvious way like for the PS.

The FR were "only" on the axis. No conclusion about a possible effect on the dispersion.
I have the intention to gather the pictures and the FR to share them. Sorry I put it in a second time compare to the influence of the additional mass those FR show.
Christian

 

[homeswinghome]

Christian,
Very interesting presentation. I like the magic triangle concept. I certainly agree that T>10 is a good target, although lately I'm feeling that 5 may be good enough for home applications. And fc>5 kHz likewise seems like a good target. But I'm not familiar with the cutoff frequency calculation based on the panel mechanical impedance. Is there a literature reference for equation 10 in section 2.4?

In 7.1 I understand the reasoning for the minimum values for fc and fhigh, but why the maximums? I don't see any harm for fc >7500, for example.

Also, in section 2.1, you might want to be clearer that equation 1 applies only to homogeneous materials, i.e. not composite sandwich constructions.

Eric

Hello Eric,

Thank you. It is exactly the feedback I expected

About T, I have used the value suggested by M. Heron's patent so 10 to 100 to start from something already published. I agree that 5 might be ok as it seems to be the value for poplar plywood and we are happy with it.

The cutoff frequency based on the mechanical impedance is directly or indirectly in all the papers that present a simplified electroacoustic model of a DML. You can have a look the reference 5 (see the references in the last page) and also it appears in a Tectonic document reference 4 (among the reasons I chose this format of document is the possibility to write clearly the math and to make a bibliography... I am a bit new to that, let me know if the links inside and outside the document works). The reality of this cutoff frequency became clear to me recently while testing light materials, it is not visible with plywood for exemple (as predicts in the chart). It is probably the main input at this time in my understanding of DML. The mechanical impedance seems to play an important role let say above 2k or 3kHz.

About eq 1 in 2.1 limited to plain material : yes. I will add it. You have probably understood I am not very familiar with composite materials, all help to to extend the document to this domain is welcome! In the do list is to come back to the bending of a composite you sent and to add it (needed for 8.2)

Christian

 

[spedge]

aagas.
At one stage ,ziggy used to use 8x4ft gatorboard with ,I think , about 8 exciters per panel ,similar to podiums.
This looks similar to foamcore , which is a heavily damped panel, so the HF will probably be restricted somewhat.
The 8 exciters ziggy used might have helped this a little ? But he never measured the panels.
Measurements would be interesting.
Steve.

 

[spedge]

It would be interesting and rewarding to learn how far you could push the best materials and exciters Steve

Burnt

The materials seem to be down to a matter of taste, so that's not an easy one to answer ?
The best exciter is a mystery to me, mainly because of the frequency plots I have seen, on this site and daytons own measurements?
The so called improvements I make on my exciters are in areas above the frequencies that these exciters go?
Are you intending to actually build an exciter from scratch?
Or are you just trying to get ideas of improvements to standard exciters?
Steve.

 

[spedge]

Eucy.
Probably a daft question, but did you try to glue the dome upside-down to mimic a whizzer cone .
This might? Work better, especially on ply?
I just this second, stuck this coffee machine container to the front of my aluminium panel for demonstration purposes.
I did use it on other panels , it gave a much higher output, too much.
i might cut it down in size to reduce the output.
oh no, not another project 🤪
Steve.

 

[spedge]

This guy also posts on diy audio I believe.
He has a vast youtube video collection of planar building.

His latest planar panel recordings sound very nice too.
He makes it look so simple.
Many great ideas.
Steve.

 

[BurntCoil]

DML material chart

In previous posts, I shared a DML material chart. At this time, some aspects of the DML theory where not linked to it. Links with experience were also missing. The picture is better now.
Please find in the attached document this chart and the elements of introduction.
Thank you for the feedback to correct, improve it. After that, I will add it to the documents in the github page with the original Python script.

Christian

Hi Christian,

I read this over the weekend. Its a great paper, well done. Much food for thought.

Burnt.