A Study of DMLs as a Full Range Speaker

Stiffness
I did some reading and regarding the question about stiffness and material choice; from what i understand mainly from this¹ paper I understand that material choice is always related to size and distance from the exciter from the edges. Again from what i understand is that this comes from two different effects; 1: the absorption/dampening from the material itself and 2; the transfer from panel to air. So if the question about material is always answered with CF or CF-sandwich as the best material to go to, the second question should be how big a distance is needed from the exciter to the edge to fully absorb or transfer all the energy for all frequencies? Because if there is energy left on the edges and gets reflected back onto the panel this can modulate other newly introduced frequencies.
Related to this question of size is the effect of waves traveling in the DML is not the same for every material, more important the speed is not the same for each frequency². It appears that lower frequency travel more slowly then 340 m/s, higher frequencies travel faster then 340 m/s.

If i interpreted this correctly it the following goals should be targeted: Balans between stiffness, absorption and size should be matching.
So in some cases it is better to choose carboard above CF. Correct or not?

Reflections/material impedance
Although problems related to reflections within the panel could also be solved by adding dampening on selective places, most probably on the outer edges. A second option could be to add or remove weight a few cm's/inches from the edge to induce a impedance matching towards the edges so the transfer to the air is optimized. (like an RF antenna that is engineered to transfer all energy into the air) The last options has the advantage with increasing the efficiency, the first option lowers the efficiency.
Simulate
To simulate, I asked some college's of mine specialized in FEM, the answer was simple, this can not be done by any free FEM software package on the marked. You need to use a professional tool with some expensive add-ons.
Experiment
So that leaves me with experimenting. The only thing i could think of is measuring the impedance of the transducer, unrelated to the material and size used, this should be resistive for each frequency. I also ordered some accelometers with the hope the don't add to much mass into the measurements. Maybe something optic should have been better..

¹ https://riunet.upv.es/bitstream/handle/10251/3347/tesisUPV2873.pdf
² https://publications.lib.chalmers.se/records/fulltext/154618.pdf
Hello Hibren,

Very good questions and wording! Your questions sound close to mine. I appreciate reading about mechanical impedance and impedance adaptation. This is something in my knowledge in electronics, for the panel, I understand the principle but I have no idea about the possibility to make some measurements/tests to approach values.
The information that free FEM software can't model properly the edge termination is not a good news.
About measurements, any input to approach the peripheral suspension characteristics is welcome.
About the panel impedance, it is approximated in many paper as resistive but seems not true on all the frequency range. Here also more information welcome

Christian
 
Stiffness
I did some reading and regarding the question about stiffness and material choice; from what i understand mainly from this¹ paper I understand that material choice is always related to size and distance from the exciter from the edges. Again from what i understand is that this comes from two different effects; 1: the absorption/dampening from the material itself and 2; the transfer from panel to air. So if the question about material is always answered with CF or CF-sandwich as the best material to go to, the second question should be how big a distance is needed from the exciter to the edge to fully absorb or transfer all the energy for all frequencies? Because if there is energy left on the edges and gets reflected back onto the panel this can modulate other newly introduced frequencies.
Related to this question of size is the effect of waves traveling in the DML is not the same for every material, more important the speed is not the same for each frequency². It appears that lower frequency travel more slowly then 340 m/s, higher frequencies travel faster then 340 m/s.

If i interpreted this correctly it the following goals should be targeted: Balans between stiffness, absorption and size should be matching.
So in some cases it is better to choose carboard above CF. Correct or not?

Reflections/material impedance
Although problems related to reflections within the panel could also be solved by adding dampening on selective places, most probably on the outer edges. A second option could be to add or remove weight a few cm's/inches from the edge to induce a impedance matching towards the edges so the transfer to the air is optimized. (like an RF antenna that is engineered to transfer all energy into the air) The last options has the advantage with increasing the efficiency, the first option lowers the efficiency.
Simulate
To simulate, I asked some college's of mine specialized in FEM, the answer was simple, this can not be done by any free FEM software package on the marked. You need to use a professional tool with some expensive add-ons.
Experiment
So that leaves me with experimenting. The only thing i could think of is measuring the impedance of the transducer, unrelated to the material and size used, this should be resistive for each frequency. I also ordered some accelometers with the hope the don't add to much mass into the measurements. Maybe something optic should have been better..

¹ https://riunet.upv.es/bitstream/handle/10251/3347/tesisUPV2873.pdf
² https://publications.lib.chalmers.se/records/fulltext/154618.pdf
"Professional tool with expensive add-ons", can you state what tools these might have been? As an engineering student I have access to a fair few programs for research purposes, definitely something I'd like to research, at least in my free time
 
Hello Eucy

1/ good trick. here I made some tooling in 2 parts to improve the precision while gluing the exciter. Not fully satisfied. Your trick sounds more simple.

4/ For my applications, I am in the opinion of only one HF source due to small wavelength. So if multiple exciters, I would suggest one dome and a cross over to keep only one HF source. Answer for a PA application might be a bit different due to the power to handle and the more limited bandwidth (need of domes?)

7/ I am also on this idea
Christian

One other odd way of centring a vented exciter is to mark the centre spot on the panel with a black marker 'spot' - then light it up from the side with a torch/phone, and having applied the glue or whatever, you can actually sight through the vent hole quite well, follow the exciter down with your eye (ala spyglass) and you will be able to get the hole on the marked dot.

Cheers
Eucy
 
And some measurements:

Here are the frequency responses. For each speaker individually, and with a sub added (purple)
View attachment 1090249

And especially for Christian, here is the Spectrogram (wavelet), for the right speaker alone, which I think shows that the resonances are reasonably well controlled, but not completely eliminated.

Eric

View attachment 1090252
Hi Eric.
At what distance and positions we're these measurements taken ?
Steve.
 
Eucy.
With the vented exciters I would center them by using a thin wooden stick through the hole ( or anything none magnetic).
After applying the glue to the exciter foot You simply slide exciter down the stick into position, after obviously placing the stick on the centre dot.
Steve.
 
  • Like
Reactions: 1 user
Eucy.
With the vented exciters I would center them by using a thin wooden stick through the hole ( or anything none magnetic).
After applying the glue to the exciter foot You simply slide exciter down the stick into position, after obviously placing the stick on the centre dot.
Steve.
Hi Steve
Yep... Good idea... Are you up to trying a dome?... I'd like to hear your feedback 😁
Eucy
 
The information that free FEM software can't model properly the edge termination is not a good news.
Christian,
Be careful not to draw such a broad conclusion.
The LISA FEM, for example, can model some types of edge terminations perfectly well, such as simple (hinged), clamped, or free edges, and also can model the addition of masses around the perimeter (or anywhere else). But there is indeed no facility for introducing and evaluating the addition of damping either at the perimeter or within the panel itself.
Eric
 
"Professional tool with expensive add-ons", can you state what tools these might have been? As an engineering student I have access to a fair few programs for research purposes, definitely something I'd like to research, at least in my free time
Perhaps something like this (no idea what it costs):

https://www.3ds.com/products-services/simulia/products/wave6/

as described here:

https://www.researchgate.net/public...zing_the_Off-Axis_Radiation_below_Coincidence

1663423247016.png


Eric
 
I tried fibreglass on nomex core. I had some demoulding issues, so only got a small plate to test, but that was not impressive at all and had low efficiency. Still learning the process and could be that it is possible to make good glassfiber plates, but my impression was that it is not worth pursuing.
It's hard to do a good job making a nomex core with a wet layup instead of prepreg. I tried the method from the EC video myself but failed miserably. I think I used too little epoxy and the skins didn't really stick to the nomex.
But I also wonder if the cells in nomex are perhaps too large for a very thin layer of either CF or FG to avoid buckling of the skin on the compression side. That would lead to much lower than expected stiffness, and hence disappointing efficiency.
Eric
 
It's hard to do a good job making a nomex core with a wet layup instead of prepreg. I tried the method from the EC video myself but failed miserably. I think I used too little epoxy and the skins didn't really stick to the nomex.
But I also wonder if the cells in nomex are perhaps too large for a very thin layer of either CF or FG to avoid buckling of the skin on the compression side. That would lead to much lower than expected stiffness, and hence disappointing efficiency.
Eric
I did actually get the skins to bond, using a very minimal amount of epoxy as in the ES video. Stiffness seemed similar as when he shows trying to bend his plate.

It is very possible my plate could have been better, and it will be interesting to try the carbon to see how much it is the material or technique. But I think I will get some more fibreglass to practice with first since I haven't managed to make a whole complete plate yet.
 
I did actually get the skins to bond, using a very minimal amount of epoxy as in the ES video. Stiffness seemed similar as when he shows trying to bend his plate.

It is very possible my plate could have been better, and it will be interesting to try the carbon to see how much it is the material or technique. But I think I will get some more fibreglass to practice with first since I haven't managed to make a whole complete plate yet.
practice with FG is a good idea.
 
could you recommend me a method or material to attach the exciters without using its adhesive so i can try before deciding.
Thank you
pixel1,
You can actually just gently press the exciter (with the release film still covering the adhesive) to the panel with your hand. Another thing I've done is to orient the panel horizontally and place the exciter on top. Adding a bit weight on top (with a buffer of foam between) can help keep the exciter from bouncing.
What I've been doing lately, however, it to use an extra exciter as a test exciter. The test exciter is one that's been used before and no longer has the adhesive on it. I apply a thin double sided tape to the panel at the test position, and then attach the test exciter to the tape. It's easy to gently pry the exciter off the tape with a small putty knife. Often I apply the double side tape in a strip, and that allows me to test any position I want along the length of the strip.
Eric
 
could you recommend me a method or material to attach the exciters without using its adhesive so i can try before deciding.
Thank you
pixel1

Somewhere back in my posts I describe my process
A strip of wide masking tape is applied to the panel, the exciter is then glued to the tape. The tape can be peeled off with the exciter and even repositioned without re-gluing (if you're careful)
Read back for a more detailed account
Eucy
 
  • Like
Reactions: 1 user
pixel1

Somewhere back in my posts I describe my process
A strip of wide masking tape is applied to the panel, the exciter is then glued to the tape. The tape can be peeled off with the exciter and even repositioned without re-gluing (if you're careful)
Read back for a more detailed account
Eucy
I use double sided tape as it is very quick and easy for instant measuring.
Not for long term use though.
The sound will be pretty much the same as using glue, but the HF tends to roll off above 10k.
I would be very careful about using this method on the lower grades of EPS ,as it would pull off the surface beads when removed .
The double sided tape will also slowly peel off with time, with the weight of the exciter pulling it down.
You might get a better HF response with the masking tape and glue method?
But with a longer waiting time for the glue to dry.
I do sometimes wonder whether the manufacturers use the double sided tape to dampen some of the noises in the coil area?
Some panel materials suffer more than others in this area.
Steve.
 
  • Like
Reactions: 1 user