A Study of DMLs as a Full Range Speaker

[homeswinghome]

What mass values did you add, (or did I miss that detail??)

Hello Eucy,
I used what I had... coins of 0.5, 1 and 2€ which are 7.5g to 8.5g... nothing highly scientific.

So far, I have completely failed to get anyone to try the coke dome....It's now your turn to be encouraged, even harassed to try it 🤞🙏... Particularly with the PP material, as it does have a 'soft' sound.

It will make a difference to the brightness, and it may also mitigate your 2.5k hump

Next planned step is to test solutions do give an escape way to the air trapped between the membrane and the magnet. Air that flows through the voice coil and then the back. I would like to see the impact on this 2.5k hump and on the HF extension. For now, I don't see the dome as a solution to the 2.5k hump. For the HF extension, I don't see other solution with PP than a tweeter. To follow

Another idea I had for PP and have commenced a trial of (suspended until I return home) is this:

At the location of the exciter, I drilled small holes in the rear face into each flute, moving the location of each hole so they don't line up. I then injected PU foam into each hole, and was surprised at how easily and far it spread along the internal flutes (see attached photos) It made a mess on the injection face, but it trimmed off easily when cured.

Good to know. The connection of the external face of the corrugated PP to internal face which is driven by the exciter seems "weak". So why not. Next time in a DIY store, I will have a look to PU foam.
Christian

 

[homeswinghome]

Christian,
I'm curious if the impedance peaks correlate with panel natural frequencies. Did you (or would you) do some tap tasting to find the natural frequencies?
Eric

Hello Eric,
I made some tap tests on samples of PP ( a square and rectangular shape) but not on the shape of the panels... I was wondering how to do it... As the exciter was only temporary attached (double side tape), I can quite easily make a tap test with the membrane with the suspension. As the mass and the stiffness of the exciter change the modes, I can make a test with it in place. What conditions would you like to see?
Christian

 

[homeswinghome]

From the first page of the tutorial for ARTA

How this works ? We are sending an impulse through the speaker, and take a reading using the ARTA software. We can determine after how many milliseconds the first sound wave reached the microphone, and after how many milliseconds the 2nd wave have reached the microphone. The second wave is sound reflected from the nearest boundary (floor or ceiling etc), and it will be picked up by the microphone few milliseconds later.
Between these two, there is a time window (that’s why it is called gated), where the microphone picks up only the sound generated by the loudspeaker, therefore ignoring room reflections. This is all fine and dandy, but there is a catch. Depending on the size of the time window, the frequency response will be accurate until a certain point. The larger the distance, between the loudspeaker and the first boundary, the lower in frequency you can plot the response curve.

If REW does have it I hope some one tells me. The initial impulse test allows the time to the 1st reflection to be measured. The frequency content of the impulse is determined by it's rise time. The Dirac pulse is a mathematical type where the rise time is zero so contains all frequencies, infinity stuff. Conversely there is a simple constant that off hand I can't remember that relates bandwidth of an amp to it's rise time. Also square wave testing when the wave is viewed. It can be over damped, under damped or critically damped.

Near field testing is just using distance to increase the difference between the sound power of the measurement compared with the reflection. The mic is placed as close to the speaker as it can be. It can't be placed so close on DML's but ..... pass I have no idea if there is any gain to be had.

LOL Practicalities - arrange to measure the response accurately accounting for reflections and what happens in typical actual use - there will be reflections.

Hello AJohnL,

What seems not said here is there is a trap behind gating or windowing a signal... The window is set just before the arrival of the 1st reflection. Sound velocity is 340m/s so 0.34m for 1ms. A 10ms window as shown suppose the reflection path being 3.4m?
In our rooms, the first reflections occurs in the 1 to 3ms?
The trap is, according to M Fourier, if the window is T (ie T=3ms), the lowest frequency you can observe is 1/T (ie 1/3ms = 330Hz).
Even worse, it is the next information are at n/T (n=2,3,4... ie 660Hz, 990Hz).
The big illusion using tools based on FFT like REW is that by extending the observation on the right of the window by adding 0 value signal, the calculation window is extended leading to possibly think T increases and 1/T decreases. this is an illusion. Out of the window the information is irremediably lost!
In addition, DML are objects with "long tail" IR. So cutting the signal with the window rejects also information from the panel.
So my current (little) experience is : deal with the reflections. Free reflection measurements are not for DIYer. You should find pictures of DIYer elavating their speaker at some meters above the floor in outdoor conditions to get the best reflection free conditions.
Christian

 

[Veleric]

I can quite easily make a tap test with the membrane with the suspension. As the mass and the stiffness of the exciter change the modes, I can make a test with it in place. What conditions would you like to see?
Christian

Hi Christian,
I was thinking of the tap test with everything exactly as you did the impedance test. With the membrane on the suspension and the exciter in position on the plate. I would suggest tapping and mic-ing at the exciter position first, to see if any of the peaks you saw in the impedance test correspond with the frequency peaks you observe in the tap test. I think your impedance test showed five strong peaks below 200 Hz. Does tapping show the same five peaks, or at least the strongest of them?
Eric

 

[Veleric]

Yes with a stepped sweep.

REW does not have that, as far as I know. Only gated measurements with a continuous frequency sweep.

In addition, DML are objects with "long tail" IR. So cutting the signal with the window rejects also information from the panel.

I fear Christian is correct about this.
Eric

 

[AjohnL]

Hello AJohnL,

What seems not said here is there is a trap behind gating or windowing a signal.

Actually I appreciate that and the time to first reflection problem but what is the technique intended to do - provide a reflectance free frequency plot. A further piece of information. A standard sweep can still be done.

Is it of any use though, That depends on what it shows. That leaves a bit of a problem if what is needed isn't available.
.

 

[Eucyblues99]

Next planned step is to test solutions do give an escape way to the air trapped between the membrane and the magnet. Air that flows through the voice coil and then the back. I would like to see the impact on this 2.5k hump and on the HF extension.

Are they not vented already?

For now, I don't see the dome as a solution to the 2.5k hump.

Steve thought so...

For the HF extension, I don't see other solution with PP than a tweeter. To follow

Ah me oh mi.. Not even worth a try....que sera 🥺

 

[Audiofrenzy]

 

[Eucyblues99]

And not a skerrick of carpet to be seen 🤗

 

[toddincabo]

And not a skerrick of carpet to be seen

aaawww shucks

 

[Veleric]

One is with aluminum honeycomb which has fish paper on one side.

Jamie,
Did you try the fish paper on both sides? I think that could be a much more efficient panel.
Eric

 

[jamienelson]

I wonder if filling the spring slots with silicone or similar would be as effective as the neoprene?

i have been looking into this further. I think i have found out what is happening when i make the ortho-planar spring, and then damp it with neoprene.

In this publication, the authors have found a way to create and calculate a frequency band gap in a metamaterial, by cutting parts of a 1cm panel away and replacing it with a viscous material.
https://arxiv.org/pdf/2108.06171.pdf

What I found by accident, they were already modelling mathematically.

so to answer your question above, yes it will be effective, but the sound may be different. as in, the neoprene is a viscoelastic foam, and a silicone is an elastic sheet based on the research foams have a non-homogenous structure and thus transfer vibrations in a different way to homogenous structures like a silicone sheet or gel.

Briefly (from what i understand) it describes how the shape of the inner vibrating plate dictates frequency that will be attenuated, and the viscosity of the material around it dictates the width of the band gap it produces.

With the ortho-planar spring i cut into the panel:

1. When i make the arms of the spring flexible, it is behaving like the viscous material does.
2. When the arms are of the spring are quite stiff still, and the rubber is added to them, then it is also dictating the band gap.

When considering the SPL graph of a commonly used exciter like the dayton ex32u2-4 as an example (30cmx30cm foam core board, infinite baffle) It shows high spl for lower frequencies, and it slowly reduces as the frequencies get higher.

So I assume the reason i am getting a flatter response on my panels with ortho-planar springs is two reasons:

1. The higher frequencies are maintained while the lower frequencies are attenuated by a few decibels.
2. The higher frequencies are easier to produce for the exciter on this smaller more mobile disc at the center of the spring.

Some new terms:​

1. The DML panel that we put the exciter on is called a phonotic resonator
   A novel metamaterial configuration is presented that combines hybrid attributes from acoustic metamaterials and phononic crystals. The dispersion characteristics of the system, referred to as the phononic resonator (PR), is shown to vary across a wide spectrum of behaviors that can, via optimal selection of inertial and stiffness parameters, be tuned to resemble a locally resonant mechanism, a phononic system, as well as a uniform homogeneous lattice. When coupled with damping elements, the emergent dissipative effect, known as the metadamping phenomenon, of the PR is shown to exceed that of a statically equivalent acoustic metamaterial under certain conditions which are highlighted here. The metadamping amplification is verified in finite PR systems via a power flow approach that depicts the spatial rate of energy dissipation along the length of a 100 cells phononic resonator.
2. The addition of a cut away damped section is called metadamping
   The first notions on the beneficial effects of viscous damping in acoustic metamaterials were reported by Hussein and Frazier (Metadamping: An emergent phenomenon in dissipative metamaterials), who introduced the concept of metadamping to refer to the damping emergence phenomenon produced as a result of combining the effects of local resonance with viscous dissipation

as a side note: this same research paper (page 147) actually points out how dips in the frequency graph are from a coincidence phenomenon, where overlapping modes cancel each other out, which is a problem I had recently.

 

[spedge]

Another nice sounding recording from Audiofrenzy, but I do miss the carpet 😄
Why aren't others posting recordings of their panels ?

Today I visited my friend that I have not seen for about 3 years, what with one thing and another.
I said I would post a picture of his panels with open baffles, for low end duties .
They sounded very nice and we listened to various types of music.
The panels are about 2mm thick cc.
I took a photo of the panels to post here, it wasn't till I got home that I realised I could have recorded them playing 🙄
We were talking too much , three years is too long.
Steve.

 

[jamienelson]

Jamie,
Did you try the fish paper on both sides? I think that could be a much more efficient panel.
Eric

i haven't, but i will.
i'be been testing out a few other things in the mean time, my short term goal is finding a really good tweeter material.
thin aluminium, thin acrylic, glass all have good SPL in the 14khz-17khz area, but it's like a huge spike, and i am looking for the right sandwich composite , and panel size to get there. (i.e: needs to be about 16cm of glass if i want a quarter of a wavelength of 7khz to fit within it, according to https://www.omnicalculator.com/physics/sound-wavelength)

 

[jamienelson]

It was 5.6 mm 5 ply poplar. I'd have been happy to use thinner if I could get it in 5 ply. This is not to say you can't use 3 ply, but it doesn't work well for my particular design, for reasons I explained in this post:

https://www.diyaudio.com/community/...s-as-a-full-range-speaker.272576/post-7326424

For the surround I used 3M Extreme double sided mounting tape. I chose it because it had good damping properties, or at least I thought so at the time. But now I think I have a better handle on how to assess damping ability, and now I see it as decent at damping but far from the best.

So I'm heading in the opposite direction. I think energy dissipation at the perimeter is a good thing, and that the real challenge is to find materials that provide sufficient dissipation, rather than avoiding those that cause too much.

Eric

Yes! damping the sides smooths the frequency response out, see images below.

 

[homeswinghome]

Are they not vented already?

Hi Eucy,
Yes they are. From my different tests (plywood or PP), it is clear the hump comes from the rear. Some resonance in the venting path (volume, section change...) is a good candidate for explanation. A specific test should prove it (or discard) the hypothesis.
Christian

 

[Andre Bellwood]

Yes with a stepped sweep. I started looking around after thinking about linear sweeps and thought if a pulse of a frequency was produce there would be a delay before the mic received reflections so a delay can allow a reflection free measurement to be made. Maybe i have missed something but that seems to be the general idea used. REW's - I've seen no examples of it being used this way. The other packages are pretty clear about this use. From memory going on the video reflections need to be >25db down, hence the advantage of use of near field for LF..

Mic still hasn't arrived so no measurements yet. Also a PC problem I wont go into. That is part resolved. I'll see how it goes,

I think REW uses an input notch filter that tracks the output signal without having to step it. But I suspect this works only if you're using a loop-back so it can calculate the expected time delay from source to mic.
I would assume it then notches out everything except the fundamental and its harmonics (to do distortion calculations. )

 

[Andre Bellwood]

My second and more favorite option for dml panels has been by cutting ortho-planar springs into the board, and then covering it 3mm neoprene.

One of my very first experiments was a polycarb disc (100mm dia) mounted by "legs" of hot glue, in a 20mm EPS-paper-skinned panel.

Got a very nice result too.

I think it might be working on the same principles as your ortho-planar design?

 

[homeswinghome]

I think REW uses an input notch filter that tracks the output signal without having to step it. But I suspect this works only if you're using a loop-back so it can calculate the expected time delay from source to mic.
I would assume it then notches out everything except the fundamental and its harmonics (to do distortion calculations. )

Hello André,
My understanding of how REW works is a bit different. The basis is from the work of M Farina who shown that a recording of a logsweep which is then convoluted with a specific complementary signal gives the IR with a high signal to noise ratio (ratio depending of the duration of the sweep). In a loop back, the electrical signal is directly sent to a second input. This IR is used as time reference.
Christian

 

[Andre Bellwood]

That's a sheet of 10mm-thick EPS rapidly becoming 2 sheets of EPS just-under 5mm each.
I cannot find 5mm EPS anywhere in this wonderful country of ours. And 24V/100VA transformers and nichrome wire are out of the question for the time being.
What to do...
I had a few guitar strings and an old stage monitor amp lying around in a gig bag. So a guitar 'B' string (nickel-plated steel, 0.3mm thick, 3.5ohms per metre) seemed just the right thing, and the amplifier could deliver a clean 100W (which was way too much anyway) into 4 ohms. So I fired up REW's sig gen with a 400hz sine wave (why not?) and got me some 5mm EPS panels from a 10mm panel.

Now the next problem is to laminate 0.5mm veneer onto EPS. That's a very serious PIA: Cold glue will not cure in a laminate press (no airflow). Solvent glue eats the EPS. Water-based PVA makes the veneer expand and wrinkle. Epoxy resin leaks through the veneer and sticks to the press itself, but it's too heavy anyway and doubles the weight of the panel. Contact cement either eats the EPS or it makes bubbles under the veneer.

Back to the drawing board.