And here’s a jolly Brit having a go at DML with a cannibalised 'normal' speaker from a few years back, and getting omnidirectional sound...
So simple...👍😀
Last edited:
I went mad and ordered one of these just to test how well it drives a stud wall.
Also to try on my heavier panels.
It was going cheap for a one off, so I thought, why not 🤗
Steve.
https://www.aliexpress.com/item/32860847091.html
Also to try on my heavier panels.
It was going cheap for a one off, so I thought, why not 🤗
Steve.
https://www.aliexpress.com/item/32860847091.html
Hmm... Low activity here... Here is some reading...
Directivity measurements applied to DML (how to do it?)
2 years ago, I posted a paper (see link in post #1 or the pdf in post #6900 DML Material chart) that makes a synthesis of the main relations from the physics that are supposed to describe the behavior of a DML.
One important conclusion of the paper is the efficiency of a DML, its high frequency cutoff and its coincidence frequency are linked and can't be adjusted independently. Those figures change according to the plate characteristics not all in the desired way so that only a compromise is possible.
This is purely theoretical and if there are clues about what plays a role in the efficiency, there are almost no evidence in our DIYer world about the coincidence frequency and the high frequency cut off.
The directivity aspect and its relation to the coincidence frequency were probably not explained enough in the paper even if it was shared in other posts of this thread.
The coincidence frequency is the frequency at which the speed of the waves in the panel are equal to the speed of the sound in the air.
According to the theory, below this frequency, the panel is supposed to have enough lobes to be considered omnidirectional.
At the coincidence frequency, the emission is supposed to jump to 90°.
Above, as the frequency increase, the main lobe angle decreases.
A directivity diagram, a spinorama of the DML should show that.
The few directivity diagrams shared here (see for exemple post #8160 from Eric, @Veleric ) are the one from Tectonic or those from M Zenker's papers.
Some months ago, a discussion here about the content of the 1st reflections seen in IRs and the fact I have a panel with such a behavior that sounds not good let me to think again about the directivity and the possibility of some HF emissions at high angles (ie post #11896).
More recently, I was reading documents about VituixCAD where how to make directivity measurements with REW is explained... and it is so simple, at least for the measurements that I am still wondering not thinking about that before!
So after this very long introduction, the goal of this post is to share how to do a directivity measurement of a DML.
Before going to details, I would like to share 2 tips from Steve (thank you! @spedge ) that make for me the DML tests much easier and relialable (compared to what I did before):
So directivity measurement is no more than a collection of REW measurements where the direction of the loudspeaker is changed at each sweep. I experimented from -90° to +90° by 10° steps. Think to make also at least one at 180° on the rear side to check about a possible exciter noise (emission in a specific frequency band not visible from the front side).
The first key point is to adjust the "IR window" of REW so that only the signal before the 1st reflexion is kept in order to make a quasi-anechoic measurement.
When the mic is at 1m, the first reflection occurs in the range 3 to 4ms. Note that with a large panel, due to the proximity with the floor, it can be lower like 2ms.
Here is an example with a Visaton FRS8 full range speaker(with a DML which has less directivity than a full range, the amplitude of the 1st reflection is higher):
The IR going to bei gated let say at 2 to 3ms, that means only the frequencies above 330Hz or even 500Hz are measured. In addition, even if REW shows more, only the multiple of these frequencies are "real", the other points are interpolated. This makes a kind of smoothing.
To adjust the IR window, check the IR for 0°.
The IR window can be set :
* After measurements, click on "IR windows" then fill the Left and Right values with the desired value
* Before measurements, for all with the same values in "Preferences" > "Preferences" > "Analysis" tab > unchek the "Set window width automaticaly" box and enter the values just below. It supposes a first measurement was done.
In addition set the measurement window, choose for Timing "Set t=0 at IR start"... (I didn't do it in my first tests). Adjust also the file name options to add automatically some angle info or index to the measurement name. The angle value will have to be in the file name. VituixCAD ask also for the information "hor" or "ver" for the plane of the rotation.
When all the measurements are collected, they are visible in the "All SPL" graph but not easy to examine from there.
Check each IR to be sure the window is correctly adjusted.
When the angle is high, depending of the environment, with a USB mic, it is possible that REW does't apply the window on the right peak if the option t=0 at start above was not selected. A manual adjustment is needed (see in the "Tools" > "IR windows" "ref time" or IR window icon). It may happen also that at high angle some reflections come earlier.
Then is the second key point which is to export them as a set of txt files to a tool that will build the "spinorama", the pseudo 3D view with the frequency and the angle as axis and the SPL as a color.
Here are the settings I used for the export :
It is a functionality of VituixCAD to build a spinorama. VituixCAD has some requirements in the naming of the files to operate.
Being under Linux, it is not the way I have followed so I can't explain it fully. You can refer to the VituixCAD site or to the following ASR thread How to make quasi-anechoic speaker measurements/spinoramas with REW and VituixCAD
It was quicker for me in a first time to write a Python script that gathers the measurements. The advantage is the possibility to add on the graph some information specific to DML like an hypothesis of coincidence frequency or a possible frequency of exciter noise seen on the rear side. I intend to dive in VituixCAD installation under Linux later.
For standard loudspeaker, the gated FR are extended by a near field measurement (the mic is very close to the membrane). For a DML because of modes below the lower frequency of the gated response, it is not possible to proceed in the same way.
Note that VituixCAD recommend not adding smoothing to FR before export, it applies it after. For my script which is much less sophisticated, I applied a 1/6 octave smoothing before export.
Some correction of the names of the txt files may be needed to show clearly the angle and the plan.
Here is the directivity diagram for the FRS8 with this method:
A standard approach is to normalize the diagram to the on axis values which gives:
Our topic is DMLL... but it will be for a future post, this one being already very long!
If those measurements were made in my kitchen or in my living room, it will work from a garage or any other room. Joke aside, any location or arrangement that increases the time for the 1st reflection is better.
Christian
Directivity measurements applied to DML (how to do it?)
2 years ago, I posted a paper (see link in post #1 or the pdf in post #6900 DML Material chart) that makes a synthesis of the main relations from the physics that are supposed to describe the behavior of a DML.
One important conclusion of the paper is the efficiency of a DML, its high frequency cutoff and its coincidence frequency are linked and can't be adjusted independently. Those figures change according to the plate characteristics not all in the desired way so that only a compromise is possible.
This is purely theoretical and if there are clues about what plays a role in the efficiency, there are almost no evidence in our DIYer world about the coincidence frequency and the high frequency cut off.
The directivity aspect and its relation to the coincidence frequency were probably not explained enough in the paper even if it was shared in other posts of this thread.
The coincidence frequency is the frequency at which the speed of the waves in the panel are equal to the speed of the sound in the air.
According to the theory, below this frequency, the panel is supposed to have enough lobes to be considered omnidirectional.
At the coincidence frequency, the emission is supposed to jump to 90°.
Above, as the frequency increase, the main lobe angle decreases.
A directivity diagram, a spinorama of the DML should show that.
The few directivity diagrams shared here (see for exemple post #8160 from Eric, @Veleric ) are the one from Tectonic or those from M Zenker's papers.
Some months ago, a discussion here about the content of the 1st reflections seen in IRs and the fact I have a panel with such a behavior that sounds not good let me to think again about the directivity and the possibility of some HF emissions at high angles (ie post #11896).
More recently, I was reading documents about VituixCAD where how to make directivity measurements with REW is explained... and it is so simple, at least for the measurements that I am still wondering not thinking about that before!
So after this very long introduction, the goal of this post is to share how to do a directivity measurement of a DML.
Before going to details, I would like to share 2 tips from Steve (thank you! @spedge ) that make for me the DML tests much easier and relialable (compared to what I did before):
- the exciters are glued to the panel with PVA. With PVA it is strong enough to have a good HF transmission, there is no risk to brake the exciter when changing for a new test. The central area remains in addition free of any tape (what I used before).
- a panel without suspension is suspended with a double ribbon of large tape at the center of the upper side, each end of the ribbon sticked on one face of the plate. A sponge is at the top of the panel to damp the possible tape noise. It is a dish washing sponge aged by 2 or 3 dish washer cycles ;-)
So directivity measurement is no more than a collection of REW measurements where the direction of the loudspeaker is changed at each sweep. I experimented from -90° to +90° by 10° steps. Think to make also at least one at 180° on the rear side to check about a possible exciter noise (emission in a specific frequency band not visible from the front side).
The first key point is to adjust the "IR window" of REW so that only the signal before the 1st reflexion is kept in order to make a quasi-anechoic measurement.
When the mic is at 1m, the first reflection occurs in the range 3 to 4ms. Note that with a large panel, due to the proximity with the floor, it can be lower like 2ms.
Here is an example with a Visaton FRS8 full range speaker(with a DML which has less directivity than a full range, the amplitude of the 1st reflection is higher):
The IR going to bei gated let say at 2 to 3ms, that means only the frequencies above 330Hz or even 500Hz are measured. In addition, even if REW shows more, only the multiple of these frequencies are "real", the other points are interpolated. This makes a kind of smoothing.
To adjust the IR window, check the IR for 0°.
The IR window can be set :
* After measurements, click on "IR windows" then fill the Left and Right values with the desired value
* Before measurements, for all with the same values in "Preferences" > "Preferences" > "Analysis" tab > unchek the "Set window width automaticaly" box and enter the values just below. It supposes a first measurement was done.
In addition set the measurement window, choose for Timing "Set t=0 at IR start"... (I didn't do it in my first tests). Adjust also the file name options to add automatically some angle info or index to the measurement name. The angle value will have to be in the file name. VituixCAD ask also for the information "hor" or "ver" for the plane of the rotation.
When all the measurements are collected, they are visible in the "All SPL" graph but not easy to examine from there.
Check each IR to be sure the window is correctly adjusted.
When the angle is high, depending of the environment, with a USB mic, it is possible that REW does't apply the window on the right peak if the option t=0 at start above was not selected. A manual adjustment is needed (see in the "Tools" > "IR windows" "ref time" or IR window icon). It may happen also that at high angle some reflections come earlier.
Then is the second key point which is to export them as a set of txt files to a tool that will build the "spinorama", the pseudo 3D view with the frequency and the angle as axis and the SPL as a color.
Here are the settings I used for the export :
It is a functionality of VituixCAD to build a spinorama. VituixCAD has some requirements in the naming of the files to operate.
Being under Linux, it is not the way I have followed so I can't explain it fully. You can refer to the VituixCAD site or to the following ASR thread How to make quasi-anechoic speaker measurements/spinoramas with REW and VituixCAD
It was quicker for me in a first time to write a Python script that gathers the measurements. The advantage is the possibility to add on the graph some information specific to DML like an hypothesis of coincidence frequency or a possible frequency of exciter noise seen on the rear side. I intend to dive in VituixCAD installation under Linux later.
For standard loudspeaker, the gated FR are extended by a near field measurement (the mic is very close to the membrane). For a DML because of modes below the lower frequency of the gated response, it is not possible to proceed in the same way.
Note that VituixCAD recommend not adding smoothing to FR before export, it applies it after. For my script which is much less sophisticated, I applied a 1/6 octave smoothing before export.
Some correction of the names of the txt files may be needed to show clearly the angle and the plan.
Here is the directivity diagram for the FRS8 with this method:
A standard approach is to normalize the diagram to the on axis values which gives:
Our topic is DMLL... but it will be for a future post, this one being already very long!
If those measurements were made in my kitchen or in my living room, it will work from a garage or any other room. Joke aside, any location or arrangement that increases the time for the 1st reflection is better.
Christian
Attachments
Directivity measurements applied to DML (how to do it?)... continued
Here is the content of the previous post in a pdf file (see attachement) and the python script.
The pdf file was also uploaded in Github with the other documents about DML : https://github.com/Homeswinghome/AudioDIY_DML
Here is the content of the previous post in a pdf file (see attachement) and the python script.
The pdf file was also uploaded in Github with the other documents about DML : https://github.com/Homeswinghome/AudioDIY_DML
Hi aagas,
I tried whipping the PVA, which worked, but putting it between two layers of aluminum was so air tight, it was like putting the glue back in the bottle. It wouldn't cure. Any suggestions?
Thanks,
Bruce
Directivity measurements applied to DML (first spinoramas)
For all the spinoramas after
As my questions about directivity started with tests of an ESP panel (30x40cm, 10mm thick, PVA coated only on the exciter side), it is the first spinorama I did.
No spine, no peripheral suspension (central tape + sponge tip)
Note that the central area, inside the voice coil perimeter, was already tweaked in order to try to remove a peak at 11kHz (a small wood tube was inserted at the center to add a weight), so this panel is maybe not fully representative in HF.
Anyway, here it is :
It shows on the left that the panel is a bit short in HF but more interesting and more visible on the right that the SPL from about 2k to 8k are higher in the direction 40 to 70° than on-axis... Probably not good.
Writing that, I wonder if it is something we could detect listening to the panel hand held at different angle...
From this spinorama, I can make no assumption about the coincidence frequency.
Note that for this panel and as others about the same dimensions after, the high SPL in the 500Hz area on the left might not be a big issue. My assumption is a DML being a diffuse dipole... it is a dipole so it should behave as an open baffle. At certain frequencies the front and rear waves combines in phase and increase the level. This is another topic.
The results of this panel lead me to consider panels with a possible higher coincidence frequency. In short to increase the coincidence frequency:
So the next test is with a clear polystyrene, the one that makes the protective window of frames. I had one sample of that (30x40cm, 1.3mm thick), no spine but a peripheral suspension made of a D shape wheather strip. This panel was used also for other tests before making a directivity measurement.
Here no evidence of the coincidence frequency, no more higher SPL in the 40° to 80° as the EPS one but it appears a too high level in the 2kHz/40° (x mark and dotted line) and a lack of SPL in a kind of oblique channel in HF.
The high level at 2kHz/40° in the normalized view is caused by the dip seen at 2kHz/0° which ask for an important gain correction. So the question would be first why this dip like if in this frequency range only odd modes occur?
On this panel, I changed the DAEX25FHE by a small DAEX13CT. The idea behind was to see what happened whith an exciter having a lower moving mass and a lower inductance which are suppose to limit the highest frequency.
Here it is the PS panel with a DAEX13CT :
No more "channel" in the HF area but a high level remains; slightly lower in frequency at 1400Hz. As the difference compare to the previous test is the exciter (same 2/5 location for both), the "channel" seems purely linked to the exciter and the 1.5k to 2KHz/0° dip to the panel including the suspension and the exciter... or an other variation in my measurement environment!
The next test is with a piece of 5mm standard plywood. Its dimensions are a bit larger than the previous panel: 34x42cm. It is also much stiffer than the previous ones.
This one seems suffer of all what was seen before : the high SPL in the 40° to 80° direction, the HF "channel", the 2kHz/0° dip and many other dips. In addition the dotted line and the cross points make an assumption about a possible coincidence frequency which is evaluated here at 8000Hz.
The evaluation of the coincidence frequency is based on an information from the theory that tells us there is a relation between the coincidence frequency, the considered frequency above the coincidence and the angle of the beam in the form:
f/fc = 1/sin²(alpha)
There is no parameter from the panel here so this relation was coded in the python script. Just by changing the coincidence frequency fc value, we can see what fits the best.
The difficulty here is that the value of 8kHz is not the value expected for a 5mm plywood which is more around 3.5kHz supposing a 700kg/m³ density and a 9GPa Young modulus. Possibly something is wrong in this data.
There are papers about the directivity characteristics of the DML; among them the ones from B Zenker. In those papers, the directivity of the DML is described as something to work on.
The general requirement about the directivity of a loudspeaker is to have a smooth decreasing diagram. A DML below the coincidence frequency and because of the modes as an uneven directivity. Depending of the angle, the contribution of each anti-node can be constructive or destructive.
In his papers, B Zenkers proposes 2 ways to improve the panel directivity :
The first panel shown here (EPS) was more or less in free edge conditions when the second (PS) is with a full peripheral suspension, the plywood 5mm is also in almost free edge conditions.
A possible next step of test would be to try to remove the suspension of the PS panel to see if it changes the behavior or if this material is simply more damped than the EPS.
The EPS panel as already now a new type of suspension which should be tested in directivity too.
The 5mm plywood seems to stiff (to low coincidence frequency) and not damped enough (material damping or edge damping?).
To continue with the available samples, 2 poplar plywood panel were tested; one being waiting for next panels, the second being the large plywood panels I made maybe 2 years ago.
One is 25x125cm, no suspension, no spine while the existing panels are 42x120cm, with a spine and a peripheral weather foam.
Those panels show the same characteristics as the previous one.
With the same method as for the 5mm plywood, the coincidence frequency evaluation is 12500Hz for both when a calculation based on a 9GPa Young modulus and a 550kg/m³ leads to 5kHz.
A dotted line at the frequency of a hump in the rear side FR is added to see if it as some relation to the front behavior. It is at 2.5kHz for the 25cm raw panel and 1.9kHz for the 42cm plywood panel. Not clear at this moment but this rear side hump is visible at +90° and -90° as a kind of leakage at the limit of the spinorama.
So this measurement technique seems really promising.
The test of a sample of acrylic or even clear polystyrene in the 5 to 6mm thickness might make the link with the spinoramas from B Zenker's paper.
The tap tests Eric initiates might be also a way to get more accurate characteristics of each sample.
It is interesting to see mainly from the 3mm plywood on the "as measured" spinorama how the density of lobes increases with the frequency.
More simply, it seems to be a method to select material, to choose the thinkness of a given material.
To be continued!
For all the spinoramas after
- on the left is "as measured" with a global gain applied to each point. The gain is set to give an average 80dB SPL over the -10° to 10° measurements. The idea is to try to give the same color on that area for each test.
- on the right is the normalized version. For this one, the SPL of one frequency for all the angle is compensate by the gain giving 80dB for 0° (= all the points of the on axis FR are at 80dB). After testing it, it might be this view is a bit "extreme" as if there is a dip, it is compensate by a gain that won't be used in reallity.
As my questions about directivity started with tests of an ESP panel (30x40cm, 10mm thick, PVA coated only on the exciter side), it is the first spinorama I did.
No spine, no peripheral suspension (central tape + sponge tip)
Note that the central area, inside the voice coil perimeter, was already tweaked in order to try to remove a peak at 11kHz (a small wood tube was inserted at the center to add a weight), so this panel is maybe not fully representative in HF.
Anyway, here it is :
It shows on the left that the panel is a bit short in HF but more interesting and more visible on the right that the SPL from about 2k to 8k are higher in the direction 40 to 70° than on-axis... Probably not good.
Writing that, I wonder if it is something we could detect listening to the panel hand held at different angle...
From this spinorama, I can make no assumption about the coincidence frequency.
Note that for this panel and as others about the same dimensions after, the high SPL in the 500Hz area on the left might not be a big issue. My assumption is a DML being a diffuse dipole... it is a dipole so it should behave as an open baffle. At certain frequencies the front and rear waves combines in phase and increase the level. This is another topic.
The results of this panel lead me to consider panels with a possible higher coincidence frequency. In short to increase the coincidence frequency:
- For a given plain material reduce the thickness
- Change the material to an heavier or softer one (less stiff).
So the next test is with a clear polystyrene, the one that makes the protective window of frames. I had one sample of that (30x40cm, 1.3mm thick), no spine but a peripheral suspension made of a D shape wheather strip. This panel was used also for other tests before making a directivity measurement.
Here no evidence of the coincidence frequency, no more higher SPL in the 40° to 80° as the EPS one but it appears a too high level in the 2kHz/40° (x mark and dotted line) and a lack of SPL in a kind of oblique channel in HF.
The high level at 2kHz/40° in the normalized view is caused by the dip seen at 2kHz/0° which ask for an important gain correction. So the question would be first why this dip like if in this frequency range only odd modes occur?
On this panel, I changed the DAEX25FHE by a small DAEX13CT. The idea behind was to see what happened whith an exciter having a lower moving mass and a lower inductance which are suppose to limit the highest frequency.
Here it is the PS panel with a DAEX13CT :
No more "channel" in the HF area but a high level remains; slightly lower in frequency at 1400Hz. As the difference compare to the previous test is the exciter (same 2/5 location for both), the "channel" seems purely linked to the exciter and the 1.5k to 2KHz/0° dip to the panel including the suspension and the exciter... or an other variation in my measurement environment!
The next test is with a piece of 5mm standard plywood. Its dimensions are a bit larger than the previous panel: 34x42cm. It is also much stiffer than the previous ones.
This one seems suffer of all what was seen before : the high SPL in the 40° to 80° direction, the HF "channel", the 2kHz/0° dip and many other dips. In addition the dotted line and the cross points make an assumption about a possible coincidence frequency which is evaluated here at 8000Hz.
The evaluation of the coincidence frequency is based on an information from the theory that tells us there is a relation between the coincidence frequency, the considered frequency above the coincidence and the angle of the beam in the form:
f/fc = 1/sin²(alpha)
There is no parameter from the panel here so this relation was coded in the python script. Just by changing the coincidence frequency fc value, we can see what fits the best.
The difficulty here is that the value of 8kHz is not the value expected for a 5mm plywood which is more around 3.5kHz supposing a 700kg/m³ density and a 9GPa Young modulus. Possibly something is wrong in this data.
There are papers about the directivity characteristics of the DML; among them the ones from B Zenker. In those papers, the directivity of the DML is described as something to work on.
The general requirement about the directivity of a loudspeaker is to have a smooth decreasing diagram. A DML below the coincidence frequency and because of the modes as an uneven directivity. Depending of the angle, the contribution of each anti-node can be constructive or destructive.
In his papers, B Zenkers proposes 2 ways to improve the panel directivity :
- increasing the panel material damping
- adding some edge damping
The first panel shown here (EPS) was more or less in free edge conditions when the second (PS) is with a full peripheral suspension, the plywood 5mm is also in almost free edge conditions.
A possible next step of test would be to try to remove the suspension of the PS panel to see if it changes the behavior or if this material is simply more damped than the EPS.
The EPS panel as already now a new type of suspension which should be tested in directivity too.
The 5mm plywood seems to stiff (to low coincidence frequency) and not damped enough (material damping or edge damping?).
To continue with the available samples, 2 poplar plywood panel were tested; one being waiting for next panels, the second being the large plywood panels I made maybe 2 years ago.
One is 25x125cm, no suspension, no spine while the existing panels are 42x120cm, with a spine and a peripheral weather foam.
Those panels show the same characteristics as the previous one.
With the same method as for the 5mm plywood, the coincidence frequency evaluation is 12500Hz for both when a calculation based on a 9GPa Young modulus and a 550kg/m³ leads to 5kHz.
A dotted line at the frequency of a hump in the rear side FR is added to see if it as some relation to the front behavior. It is at 2.5kHz for the 25cm raw panel and 1.9kHz for the 42cm plywood panel. Not clear at this moment but this rear side hump is visible at +90° and -90° as a kind of leakage at the limit of the spinorama.
So this measurement technique seems really promising.
The test of a sample of acrylic or even clear polystyrene in the 5 to 6mm thickness might make the link with the spinoramas from B Zenker's paper.
The tap tests Eric initiates might be also a way to get more accurate characteristics of each sample.
It is interesting to see mainly from the 3mm plywood on the "as measured" spinorama how the density of lobes increases with the frequency.
More simply, it seems to be a method to select material, to choose the thinkness of a given material.
To be continued!
Yes, they don't work very well. Although they're quite lightweight the efficiency is not good (plastic too dead material?), not much high and low freq output. "plastiqy" kind of sound - but that's just my observation.
It would have been great if they worked...... I also tried this : https://igepa.nl/06-0008-0066-akyliter-wit-600-g-m-2050-mm-x-3050-mm-2900 (polyprpolene honeycomb sheet use for printing outboard signs) , but alas to no avail.....
Directivity measurements applied to DML (first spinoramas)
pdf version of the post #12708 with add-on (directivity index DI)
pdf version of the post #12708 with add-on (directivity index DI)
Christian,
This is so cool! Thanks for sharing this. I have not gotten through your posts in detail, but this looks like a very good tool.
Those panels show the same characteristics as the previous one.
With the same method as for the 5mm plywood, the coincidence frequency evaluation is 12500Hz for both when a calculation based on a 9GPa Young modulus and a 550kg/m³ leads to 5kHz.
And you have found clear evidence of the coincidence frequency! Nice! Did the peak in the SPL at coincidence evident in the original SPL measurements at 60 to 90 degrees? Or did you only see it in the final spinoramas? As you may recall, I have tried many times myself to find it at high angles in the SPL response, but it has never been apparent in any of my attempts. Does windowing help in that regard? I can't say for sure what panels or suspensions I tried, but I feel sure I tried plywood at least once. Perhaps I will have to try again.
Regarding the difference between the measured and calculated values, there may be a simple explanation. What is the basis of your calculated value? If I use the "thin plate" approximation, I get a value of 3.2 kHz, rather than 5 kHz (assuming I have not made a simple error). But in any event, at high frequencies the thin plate approximation is no longer valid, as shear deformation and rotary inertia become significant. As a result, the wave speed is slower (and the coincidence frequency is higher) than the "thin plate" approximation would suggest. Or have you already accounted for that in your calculation?
Also, concerning the attachment of your exciters, you said you used Steve's method of using PVA, instead of tape. And did you find that with this method you were able to later remove the glued exciter from plywood panels without damaging the plywood and/or exciter?
Thanks,
Eric
I was sure of your interest for such a post!
Seems like yes. It is something we have been looking for for a long time. In addition to give evidence to the coincidence frequency, it answers to one of our old question which is "where to have the coincidence frequency in the panel bandwidth?". Well most probably outside. That means if the target is a wide range panel, it should be somewhere above 20kHz with in consequence to have to admit a lower efficiency. In other words, as light and as stiff as possible for a quality home application might not be the way.
The spinorama doesn't modify the data. It is more a graphical representation. So you can already see things in REW. The help of the spinorama is to show how those peaks are organized in angle and frequency like the beam angle that occurs above the coincidence.
Windowing is the key point. The goal is to keep only the direct sound without all the reflections sent by the room which are a delayed view of sound coming from many other angles from the panel.
I just used the formula for thin plate... and it possible I made a mistake. If you remember previous posts, the statistic say : I post a result, you check... and your result is the right one!
It is interesting to re-enter the shear deformation in the discussion because yes it is a good reason that the plate reach the coincidence frequency much later. If I remember, we already found papers (not dedicated to DML) about that. It might be also an explanation for material like EPS doesn't have a coincidence frequency signature. Its shear modulus might be too low to allow it to create supersonic waves. The consequence is when I was not happy with my EPS test panel because of peaks off axis, it was not because of a too low coincidence frequency but because a too low damping. In his paper, B Zenker mentions the needed balance between the traveling waves and the standing waves. This might make the link to what Steve introduces in how a panel is a bending wave one (= traveling waves only?) or a DML (=standing waves fully reflected?).
For now, I have had no damage to my exciters. Some were remove maybe 5 or 6 times, maybe more. This for DAEX25FHE. It worked also for DAEX13CT. It is more risky for the panel. For a plywood one some glue remains on the panel. Not possible then to come fully to the prior state but you can remove the excess of glue and sand the panel to limit and glue again. For EPS, there is a risk to damage the surface. So ok for the exciter, not so for the panel. I reused some of those panels with the risk to create an unexpected behavior. Anyway, much better in result than my previous double side tape.
In the post or the pdf, you will see my assumption that the peak in the 500Hz area is caused by the dipole nature of the panel like in an openbaffle and I opened the question of the origine of a dip in the 1.5 to 2k range... It would be simply also written in the dipole nature of a DML.
I have some other material like cardboard or XPS that I am going to test, probably in 2 weeks from now, whith 2mm basswood plywood I just received.
One other important question opened with those measurements is the choice of the coating or protective layer of the membrane. If the external layer increases the stiffness, the coincidence frequency will decrease which is not the goal.
So I wonder which varnish or finishing product to choose for plywood ? It should increase the material damping, probably not too much, but not the stiffness. Any suggestion welcome!
Christian
PS : the post and the pdf are similar in content. For the results, read preferably the pdf than the post. I added in the pdf the directivity index graphs.
Plywood 5mm spinorama
@Eucyblues99
Hello Eucy, among the directivity tests I made, there is the one below about 5mm plywood. I am wondering if it is something you had?
Seeing the emission in HF (16k?) in almost all the direction, the youngest hears might not be happy in such a situation! May your Eucy dome a countermeasure to such emissions?
@Eucyblues99
Hello Eucy, among the directivity tests I made, there is the one below about 5mm plywood. I am wondering if it is something you had?
Seeing the emission in HF (16k?) in almost all the direction, the youngest hears might not be happy in such a situation! May your Eucy dome a countermeasure to such emissions?
@Veleric
Eric, for the EPS there is another possible explanation. When we look at the DI graph, there is like a pattern common to the graphs showing a possible coincidence frequency which is the DI decreases up to the fc. The EPS panel DI has such a characteristic with a lower point around 8kHz. There is a peak in the 60° to 80° range.
A way to confirm this hypothesis would be to sand the panel down to 5mm.
Christian
I have made several pair of panels out of these. I buy a 2 feet by 8 feet panel and cut it into two 2'x4' sections. They ring some - especially if bare. However they were the most neutral and lively of the panel materials I have tried so far.
One pair was painted with the lightest coat of spray paint I could manage. The other was decorate in a floral paper by my wife. Both seemed to ring less after this, but of course not quite as detailed and lively- but still really good.
The ringing I mentioned I can only hear when I cut the music off. I am sure it would make a terrible waterfall chart, but it is very enjoyable to listen to in my opinion.
I really want to have another go at the meranti plywood I had used before. It was great in the midrange, but needed some help in the higher frequencies. I would like to attempt to have something like a 3-way speaker with either 3 panels of different materials or 2 panels and a super tweeter. Of course, a subwoofer to handle the lower octaves.
Damn.
Possibilities...
Go to the 3M site and look at self curing adhesives.
Something epoxy like or even contact adhesive could work.
Once can get very good adhesion to rigid foams with latex adhesive used like contact cement.
You could try the whipped pva on one sheet of aluminum. Let it dry. Then use latex or other adhesive on the other side.
@bdjohns
Or like for the veneer coming with an adhesive layer, after the PVA dried, apply by heating the metal side with an iron to make the glue adherent to the second face?
But not sure what will be the result. Adherent but not strongly glued, too damped?
Some months ago, I was in contact with an epoxy supplier with the idea to make some sandwich including aluminum. For him, aluminum is a problem due to the chemical properties of its oxide in surface. A complex treatment is needed to get a full structural strength.
I tested PVA between 2 plastic layers... never cured.
I glued also thin wood layers with PVA to make a kind of plywood, I haven't pushed the test but the first impression knotting at the panel is a too high damping as Steve reported about his trial with aluminum. I didn't go in the way because it gave what I thought to be at that time a too heavy panel.
Have a look to Steve's post like #7709] (in search > advanced search, keyword = aluminium (note the i), posted by = Spedge). @spedge
Gluing 2 layers together is the same what is called "constraint layer damping" where the goal is to get a highly damped material. About that technique, the video "World's second best loudspeaker" from Tech Ingredient on YouTube is interesting.
The glue used in plywood seems ok for a panel (damped enough?), the PVA, and I guess the neoprene too might be too damped.
Christian