A Study of DMLs as a Full Range Speaker

Jump to Latest
M

moray james

Member
Joined 2003
morayjamescables@yahoo.com
you might try removing the fabric spider and use one like this instead.

1710902303909.jpeg
 
M

moray james

Member
Joined 2003
morayjamescables@yahoo.com
homeswinghome said:
Hello Eric,
The idea is to use a phono cartridge. Have a look here for some illustrations : Equivalent circuit of the dynamic system of a phono cartridge. A phono cartridge transforms the speed of the stylus in volt. The difficulty is to have an idea of this transfer function according to the frequency.

With a mic, there is no reason to proceed with a single frequency in my opinion. My favorite test is the IR with a logsweep. Depending of the sweep duration, the frequency resolution is good. Treatment like smoothing are possible afterwards. It is more difficult with a noise where the frequency resolution is defined at the measure by the RTA or spectrum analyser. setup.

Christian
Click to expand...
a stereo phono (45 and 33 3 rpm) cartridge reads the horizontal wiggles of the groove. a 78 rpm cartridge reads up and down motion of the grove depth.
 
  • Like
Reactions: 1 user
M

moray james

Member
Joined 2003
morayjamescables@yahoo.com
spedge said:
Hi Christian.
This was the part of this patent that I was most interested in, at the time.
He found that the mass loading the panel roughly half way between the exciter point and the edge of the panel, filled in the bad dip in the response at 200hz .
This is basically the same method i use to find these points on my small panels, but i use my fingers while watching the response change ,in real time.
it is very simple.
The first part also describes the bowing in the centre of the panel , which i regard as the direct radiating part of the DML panel ( pistonic in nature 😀).
This is where most of the hf above 10k comes from on an EPS panel.
steve.

quote.
When the vibratile system 22 is thus relatively stifiiy clamped, and energized at its central point by voice coil form 18, the vibratile system bows in and out with maximum excursion at the central driven point, and the degree of excursion gradually diminishes towards the peripheral edges. It may be thus generally classified as a clamped piston bowed at the center. It is well known in the art of physical vibrations of membranes that when a clamped membrane is thus energized there are set up various nodes of vibrations with trough and peaks determined by the energizing frequency, by the shape of the diaphragm, and by the physical constants of the material of the membrane. Whereas such nodes of vibrations are relatively well known for standard circular; and homogenous membranes, the determination of such nodes and anti-nodal points can be determined only by approximate empirical methods for other than symmetrically contoured areas. For the type of fiat vibratile systems which has been reduced to practice in this invention where one side of the vibratile system is approximately one and one half times the length of the other, and where the piston overall dimension is approximately 15" X 22", the node of vibration is such that at approximately 200 cycles per second there is a pronounced anti-resonance in the acoustic response of the system as indicated by the dotted curve of FIG. 9
By exploring the vibrations of the panel with a sound probe, it was found that there were anti-nodal points of vibration very nearly equidistant between the furthermost clamped extremities of the panel and the voice coil driving point. To eliminate the anti-nodal vibration areas, it is feasible to load these points with masses of proper weight that will cause the overall modal vibration display to change. The effect of adding these masses 38 (shown in FIG. 8) is to eliminate the pronounced acoustic drop at 200 c.p.s. and to smooth out the overall response curve as a whole as indicated by the solid curve of FIG. 9. This invention thus teaches a method whereby it is possible to improve and make uniform the vibration of irregular piston shapes by probing them acoustically for anti-nodal points, and then mass loading these points to upset their anti-resonance effect.
Click to expand...
so I wonder what removing the anti node points does to the sound stage and the image of the panel? some times things which make speakers and amps measure smoother also make them sound like crap. watched that in the seventies with flat response and low RD. What a wast of ten years of R&D. A hard expensive lesson to learn what not to do.
 
M

moray james

Member
Joined 2003
morayjamescables@yahoo.com
spedge said:
Hi Christian.
This was the part of this patent that I was most interested in, at the time.
He found that the mass loading the panel roughly half way between the exciter point and the edge of the panel, filled in the bad dip in the response at 200hz .
This is basically the same method i use to find these points on my small panels, but i use my fingers while watching the response change ,in real time.
it is very simple.
The first part also describes the bowing in the centre of the panel , which i regard as the direct radiating part of the DML panel ( pistonic in nature 😀).
This is where most of the hf above 10k comes from on an EPS panel.
steve.

quote.
When the vibratile system 22 is thus relatively stifiiy clamped, and energized at its central point by voice coil form 18, the vibratile system bows in and out with maximum excursion at the central driven point, and the degree of excursion gradually diminishes towards the peripheral edges. It may be thus generally classified as a clamped piston bowed at the center. It is well known in the art of physical vibrations of membranes that when a clamped membrane is thus energized there are set up various nodes of vibrations with trough and peaks determined by the energizing frequency, by the shape of the diaphragm, and by the physical constants of the material of the membrane. Whereas such nodes of vibrations are relatively well known for standard circular; and homogenous membranes, the determination of such nodes and anti-nodal points can be determined only by approximate empirical methods for other than symmetrically contoured areas. For the type of fiat vibratile systems which has been reduced to practice in this invention where one side of the vibratile system is approximately one and one half times the length of the other, and where the piston overall dimension is approximately 15" X 22", the node of vibration is such that at approximately 200 cycles per second there is a pronounced anti-resonance in the acoustic response of the system as indicated by the dotted curve of FIG. 9
By exploring the vibrations of the panel with a sound probe, it was found that there were anti-nodal points of vibration very nearly equidistant between the furthermost clamped extremities of the panel and the voice coil driving point. To eliminate the anti-nodal vibration areas, it is feasible to load these points with masses of proper weight that will cause the overall modal vibration display to change. The effect of adding these masses 38 (shown in FIG. 8) is to eliminate the pronounced acoustic drop at 200 c.p.s. and to smooth out the overall response curve as a whole as indicated by the solid curve of FIG. 9. This invention thus teaches a method whereby it is possible to improve and make uniform the vibration of irregular piston shapes by probing them acoustically for anti-nodal points, and then mass loading these points to upset their anti-resonance effect.
Click to expand...
if damping these anti noder and it reduces the impedance peaks then I would think that would make for a better sounding loudspeaker but I would not bet on it until i had tried it first. thanks for posting this interesting patent info Steve.
 
V

Veleric

Member
Joined 2016
Leob said:
Thickness 0.025, Young's 30 MPa, Poisson's 0.3 and density 25

Perhaps I'm doing the displacement in a bad way, but since I cannot have that many nodes in the free version, I have only fixed the edge nodes and not an area. It is a very rough approximation, but due to the low resolution, if I would select a couple of complete quads the area would be too big instead:
View attachment 1283046
Click to expand...
Leob,
Sorry it took me a while to get back to this model. Probably you are beyond even thinking about this particular model anymore, but once I copied your inputs and constraints I realized I had to go back (again) on what I said about the 10th mode. So actually, once I was able to reproduce it, I saw that the 10th mode really is a "spurious" in-plane mode, which you should mainly ignore.
When you display it deformed, you can see that the displacements are all in the X and y directions, with none in the Z direction.

On the question of selecting constraints to use, it can be a little tricky. That's another reason why it's good to confirm the model with tap testing or impedance testing. Either can tell you where the natural frequencies really are.

Eric

1710980212075.png
 
M

moray james

Member
Joined 2003
morayjamescables@yahoo.com
Veleric said:
Christian,
True, the three you mentioned all have one or two peaks that are higher than any of the peaks in the aluminum. I'm not sure I'd put much weight on individual peaks, but still, it is indeed surprising to me that the they are not more differentiated from the aluminum.

The PS foam results I included where uncoated. But I did also test another XPS panel that was coated with several coats (four as I recall) of PVA/water. Unfortunately, it is not a perfect comparison with any other panels as the coated panel was much thicker (25 mm vs 14 mm) and the corners were rounded. Results are shown below with the 25 mm coated/rounded XPS panel in red, and the 14 mm uncoated panel in green.

For the thicker panel, the peaks are fewer within any given frequency band, but that would be expected as the thicker panel is much stiffer, and hence all the natural frequencies are more spread out. But as far as the sharpness of the peaks, the effect of the coating (if any) is not dramatic.

Eric

View attachment 1231475
Click to expand...
from the description of the anti nodes in the patent Spedge was just talking about if you could see where those anti nodes are then perhaps you could laminate a layer of very tin aluminum film say a half inch wide which overlapped the anti nodes and attempt to smother them? or if they were spaced far enough apart then simply spot damp them with a constrained foil layer? Just thinking out loud here.
 
A

Andre Bellwood

Member
Joined 2022
Andre Bellwood said:
Oh look!
Ben Zenker is part of that paper! I think he's closely involved with the Xcite designs
Click to expand...
Thanks @moray james for resurrecting some of the old posts.
"That Paper," by Ben Zenker et al, mentions voice coil break-up as being one of the causes of poor upper frequency response.
1711002401939.png

The VC buckles under the mechanical stress of driving the (panel-plus-air) load at higher frequencies.

If we look at the Xcite drivers more closely, it's good to see that Ben has applied exactly this paper's solution to the XT32 drivers. The solution being the reinforcement of the VC, and which explains why they get a much better HF response especially when loaded with a higher density, heavier panel.

Here's a close-up of the XT32-4 on the left, with the ubiquitous DAEX30HESF-4 for comparison:
1711003898064.png

The XT32-4 not only has the VC bonded to the spider a lot higher up and closer to the load face, but also has a nice, thick layer of epoxy all around, supporting it even further.

And the proof of the pudding...
1711004551439.png


Highlighted red is the XT32-4, the DAEX30HESF is the light blue curve. The drivers were mounted at the standard 2-5ths/3-5ths positions.
The panel is a 40 x 50cm canvas frame, with a resin-hardened Nidaplast insert which is shaped to minimise corners flapping around too much. Like this:

1711004692926.png


Because of the weight of the very stiff Nidaplast/resin combination, which is quite heavy, bonded to the stretched canvas surface, I get a VERY nice bass response! But it's a bit ragged down there and I might be tempted to either change to a taller aspect ratio, or pay some attention to the steepness of those arched cut-outs.
 
Last edited:
  • Like
Reactions: 4 users
A

Andre Bellwood

Member
Joined 2022
moray james said:
from the description of the anti nodes in the patent Spedge was just talking about if you could see where those anti nodes are then perhaps you could laminate a layer of very tin aluminum film say a half inch wide which overlapped the anti nodes and attempt to smother them? or if they were spaced far enough apart then simply spot damp them with a constrained foil layer? Just thinking out loud here.
Click to expand...
Have a look at this: https://www.diyaudio.com/community/...s-as-a-full-range-speaker.272576/post-7606194
 
  • Thank You
Reactions: 1 user
S

spedge

Member
Joined 2009
moray james said:
so I wonder what removing the anti node points does to the sound stage and the image of the panel? some times things which make speakers and amps measure smoother also make them sound like crap. watched that in the seventies with flat response and low RD. What a wast of ten years of R&D. A hard expensive lesson to learn what not to do.
Click to expand...
The panel in this patent is already quite heavy and damped using glued cross strips.
Heavy panels like ply have already lost a lot of delicate information .
These points will probably not make much difference to the sound.
But with lighter panels, such as eps ,this would make a great difference in sound.
Some panels benefit from such weight positions, some do not, that is why I treat every new or different panel as a new project with many listening tests between each stage of building.
There is no quick fix that works for all panel types.
I believe this method is probably used by tectonic, but is never shown.
Steve.
 
  • Like
Reactions: 1 user
S

SapphireSloth

Member
Joined 2023
I got this from a 10"x22" (~25cmx56cm) e-glass'd EPS panel the other day:

1711014218427.png


The orange line was my first test in a new frame, driver unsupported, with angled baffles on 2 sides. Short side on top and bottom are supported full length. I'm still working on this frame, and the panel was a crazy experiment, but this LF is pretty impressive so I thought I'd share. Approximately 10"x22", 10mm thick, 185g.

One of these days I'm going to order some Xcite drivers. I figure I'll keep using this DAEX32EP-4 for all the testing though since I add & remove it so much. I'm surprised I haven't broken it yet. It's probably approaching 100 mounts using carpet tape.
 
  • Like
Reactions: 2 users
M

moray james

Member
Joined 2003
morayjamescables@yahoo.com
Andre Bellwood said:
Thanks @moray james for resurrecting some of the old posts.
"That Paper," by Ben Zenker et al, mentions voice coil break-up as being one of the causes of poor upper frequency response.
View attachment 1288794
The VC buckles under the mechanical stress of driving the (panel-plus-air) load at higher frequencies.

If we look at the Xcite drivers more closely, it's good to see that Ben has applied exactly this paper's solution to the XT32 drivers. The solution being the reinforcement of the VC, and which explains why they get a much better HF response especially when loaded with a higher density, heavier panel.

Here's a close-up of the XT32-4 on the left, with the ubiquitous DAEX30HESF-4 for comparison:
View attachment 1288798
The XT32-4 not only has the VC bonded to the spider a lot higher up and closer to the load face, but also has a nice, thick layer of epoxy all around, supporting it even further.

And the proof of the pudding...
View attachment 1288799

Highlighted red is the XT32-4, the DAEX30HESF is the light blue curve. The drivers were mounted at the standard 2-5ths/3-5ths positions.
The panel is a 40 x 50cm canvas frame, with a resin-hardened Nidaplast insert which is shaped to minimise corners flapping around too much. Like this:

View attachment 1288800

Because of the weight of the very stiff Nidaplast/resin combination, which is quite heavy, bonded to the stretched canvas surface, I get a VERY nice bass response! But it's a bit ragged down there and I might be tempted to either change to a taller aspect ratio, or pay some attention to the steepness of those arched cut-outs.
Click to expand...
Glad that my curiosity has been helpful. The shape of the panel seems so counter intuitive to me for preventing the corners from flapping. Hard to shift gears and get into a different mindset with dml panels. So are you suggesting that more shallow curves on the panel side might be an improvement? thanks.
 
E

Eucyblues99

Member
Joined 2021
Andre Bellwood said:
The panel is a 40 x 50cm canvas frame, with a resin-hardened Nidaplast insert which is shaped to minimise corners flapping around too much. Like this:
Click to expand...
Hi Andre,
Cool idea . 👍 Similar to Steve's canvas suspended ply panels

For the corners, have you tried inserting a radiused fillet in each to 'soften' them in stress/deflection terms??. It may allow the peaks in the Nidaplast insert to be reduced
Eucy
 
  • Like
Reactions: 1 user
You must log in or register to reply here.
Top Bottom