Hi Eric
900x300x3mm poplar 3ply..don't have the properties to hand...apart from SG which is 0.4, Et/El .05 approx
PS..found this data
Thanks
Eucy
Thought I'd have another go at trying to describe what I'm getting at wrt the effects of anisotropy
The trouble is, despite veritable reams of papers on plates, there's never enough to isolate a specific directly applicable case
Never mind...let's proceed
The basic eqn for f0 is:
π/A*√{S/m}
Where S= bending stiffness per unit width, m=mass/unit area, A = panel area
Ok...Now consider Eric's graphs on the previous page:
The area of the panels remained the same, so all of the above basic parameters remained static, hence f0 should not change with increasing aspect ratio...but it does, with approx linearity.
Why?..because to keep the area the same, the width must be reduced as the height is increased. This is affecting the relative flexibility ratio of the panel (X/Y)
Even though the 'specific' stiffness of the panel material per unit width remains constant, the bending flexibility reduces in the transverse direction as the width is reduced, and it is my contention that this raises f0.
Now let's change the transverse S value by changing E in the transverse direction by using Et/El = 0.05 (typical for panel timbers, and not too far wrong for 3 ply. 5 ply is a different case)
f0 is a function of the √ of S, so reducing Et to 0.05 of the El value gives a √ value of 0.22
Hence, all things being equal, reducing Et should drop f0 by a factor of approx 4.
All things are not equal however, as the X/Y stiffnesses are unequal, and this will affect the final result, and it can't be predicted by the above formula
On top of all this is the effect on output power of the reduced displacement in low mode cases caused by narrowing the width of an isotropic plate.
That's why I believe that rectangular plates should be anisotropic.
It would be interesting to see Eric's graphs with a constant height and steadily reducing widths. Hint, hint...
Also, I have no doubt that evening modal spreads is a good aim, esp if they overlap, but practically, based on my Chladni testing, really low modes are very unlikely to overlap on a practically sized panel.
On the cedar panel previously, the 25hz pattern took up a major portion of the panel, and only the edge of a second one appeared at the end of the panel
Higher frequency patterns do jam more closely together but they're really not the subject of this discussion
Further testing required....
Cheers
Eucy
PS. Phew!! Typing on a phone is no fun!!!
The trouble is, despite veritable reams of papers on plates, there's never enough to isolate a specific directly applicable case
Never mind...let's proceed
The basic eqn for f0 is:
π/A*√{S/m}
Where S= bending stiffness per unit width, m=mass/unit area, A = panel area
Ok...Now consider Eric's graphs on the previous page:
The area of the panels remained the same, so all of the above basic parameters remained static, hence f0 should not change with increasing aspect ratio...but it does, with approx linearity.
Why?..because to keep the area the same, the width must be reduced as the height is increased. This is affecting the relative flexibility ratio of the panel (X/Y)
Even though the 'specific' stiffness of the panel material per unit width remains constant, the bending flexibility reduces in the transverse direction as the width is reduced, and it is my contention that this raises f0.
Now let's change the transverse S value by changing E in the transverse direction by using Et/El = 0.05 (typical for panel timbers, and not too far wrong for 3 ply. 5 ply is a different case)
f0 is a function of the √ of S, so reducing Et to 0.05 of the El value gives a √ value of 0.22
Hence, all things being equal, reducing Et should drop f0 by a factor of approx 4.
All things are not equal however, as the X/Y stiffnesses are unequal, and this will affect the final result, and it can't be predicted by the above formula
On top of all this is the effect on output power of the reduced displacement in low mode cases caused by narrowing the width of an isotropic plate.
That's why I believe that rectangular plates should be anisotropic.
It would be interesting to see Eric's graphs with a constant height and steadily reducing widths. Hint, hint...
Also, I have no doubt that evening modal spreads is a good aim, esp if they overlap, but practically, based on my Chladni testing, really low modes are very unlikely to overlap on a practically sized panel.
On the cedar panel previously, the 25hz pattern took up a major portion of the panel, and only the edge of a second one appeared at the end of the panel
Higher frequency patterns do jam more closely together but they're really not the subject of this discussion
Further testing required....
Cheers
Eucy
PS. Phew!! Typing on a phone is no fun!!!
Extra to the above for someone with the time and patience to try it is:
1/. Testing the effect/s of bias cutting the ply
2/ selecting a panel width, say 300, finding the size of the f0 pattern, and making a panel 2*f0 pattern length, then placing a second exciter in the second pattern position, maybe (but preferably not) with a low pass filter on one to boost the lows and not muddy the mids/ highs.
3/ Couple 1 and 2 above for maybe a more compact result ( by distorting the f0 pattern) and more fun
Hmmm
Eucy
1/. Testing the effect/s of bias cutting the ply
2/ selecting a panel width, say 300, finding the size of the f0 pattern, and making a panel 2*f0 pattern length, then placing a second exciter in the second pattern position, maybe (but preferably not) with a low pass filter on one to boost the lows and not muddy the mids/ highs.
3/ Couple 1 and 2 above for maybe a more compact result ( by distorting the f0 pattern) and more fun
Hmmm
Eucy
A strange concept in 2/ above...Mode 1 would be expected to fill the full panel area, but with offset driver excitation, anisotropy, and high(ish) aspect ratio, it doesn't, at least in my test results. A large oval shape was formed and as mentioned, a second zone started at the far end of the panel.
Eucy
Eucy
Eucy.
I f you took johnnoGs 2ft square proplex panel, which has a good full frequency response down to 40hz.
Then compared the response to the same panel material, but 4ft long and 1ft wide, with the exciter in the top half of the panel.
what would you expect the panel performance and response to be ?
and what would you expect the effect on sound to be ?
I measured a similar panel to the tall blonds some time ago and found some of these problems you mention, and more.
measuring real panels in real rooms, shows up all these problems.
steve.
I f you took johnnoGs 2ft square proplex panel, which has a good full frequency response down to 40hz.
Then compared the response to the same panel material, but 4ft long and 1ft wide, with the exciter in the top half of the panel.
what would you expect the panel performance and response to be ?
and what would you expect the effect on sound to be ?
I measured a similar panel to the tall blonds some time ago and found some of these problems you mention, and more.
measuring real panels in real rooms, shows up all these problems.
steve.
Hi Steve
I hesitate to answer as I'm unfamiliar with Proplex but I assume it's the same type of product as that known as Corflute in Aus.
If it is the same, it's quite soft, and certainly anisotropic. It must be reasonably thick to be stable over 4ft in the transverse direction?? ( I'll have to look up JohnnoG's panel details)
I have some 3mm Corflute which I quickly tested with an exciter but I didn't like the result (it sounded dead from memory) so I went no further. Maybe I should look again.
As for your specific questions, I'd say the 4:1 aspect one would lack some low frequency output compared to the 1:1, (although exactly 1:1 is not a good ratio either).
I don't believe the quality of the mids and highs will be worse that the 1:1 example.
Overall, the sound will probably appear 'brighter'.
The exciter at the top half is tricky ...where exactly in the top half?
3 of my four 3:1 panels have exciters at the 'std' 2/5, 3/5 locations, one slightly different, and they are fine...I shuffle the exciters around and listen, and so far I haven't found that the location is hyper critical...There would be a best spot of course, but within reason moving around a bit shouldn't wreck results. I have listen tested the 3:1 ply panels with exciters closer to the top/bottom edges and in some spots found good low frequency responses but compromises at other frequencies. Whether that holds true with Proplex I can't say.
Ultimately though, with very high aspect ratios, I would expect some loss of performance as the exciter is moved further and further away from a quasi central location...in effect, depending on the 'speed' of the material, it might figuratively 'run out of puff' reaching the lower panel extremities. That's what made me ponder (above) the use of a second exciter on a really tall panel.
Pure supposition on my part though at this stage.
You're probably now going to tell me the opposite is in fact the result LOL 😄
Cheers
Eucy
Eucy,
You can do it yourself I bet. Here's the equation:
This snip comes from here:
https://www.comsol.com/multiphysics/eigenfrequency-analysis
The f0=Pi/Area... version of the equation which you mentioned actually applies stricly only to a square isotropic plate and is a fair approximation for a "squarish" plate but as you get farther from square the approximation gets worse.
Note if you set m=n=1 (fundamental), and set a=b (square plate), the expression above simplifies to the "Pi/Area..." expression that you referenced.
Eric
Hi Eucy.
I'm not sure if corflute is the same, I asked the same question about correx , but got no answers.
I made some recordings of this material ,and I certainly hope they didn't sound dead ?
The thinnest proplex johnnoG sent me was 4mm and the largest was 6mm.
I tried the centre position , and the NXT 4/9 and 3/7 positions which I use mostly for a more central position.
I have no problem with lengthening a panel as long as it is useful.
But if narrowing the panel Too much you can start to run into problems.
The thinner panel , let's say the 300mm panel , will start to act like a small 300mm panel from the sides, reducing the lower frequency response radiated to the sides.
This will give the panel a thin bright sound , especially if placed close to a wall or reflecting objects.
I had a 2ftx1ft with this very problem and posted measurements, further back.
So from thinking it was the panel beaming heavily at me , it was the wall reflecting hi frequencies only at me (because the lower frequencies had dropped away), the opposite to a standard speaker which reflects the lower frequencies .
This is easily measured and heard.
No I am not going to say the opposite, moving the exciter up a long narrow panel like the tall blonds does put the panel way out of balance, in many ways.
But it did sort of work.
I have said many times before that I am only interested in obtaining the best possible sound from my panels, hence recording them to show the sort of sound which can be obtained using many different materials.
I did this by listening to many panel materials and experiments with shapes and sizes and weights and the rest.
Patents and white papers have helped very little I'm afraid as none of them mention the sound quality, only the effects of doing this or that on measurements, not how it affects the sound.
Steve.
I'm not sure if corflute is the same, I asked the same question about correx , but got no answers.
I made some recordings of this material ,and I certainly hope they didn't sound dead ?
The thinnest proplex johnnoG sent me was 4mm and the largest was 6mm.
I tried the centre position , and the NXT 4/9 and 3/7 positions which I use mostly for a more central position.
I have no problem with lengthening a panel as long as it is useful.
But if narrowing the panel Too much you can start to run into problems.
The thinner panel , let's say the 300mm panel , will start to act like a small 300mm panel from the sides, reducing the lower frequency response radiated to the sides.
This will give the panel a thin bright sound , especially if placed close to a wall or reflecting objects.
I had a 2ftx1ft with this very problem and posted measurements, further back.
So from thinking it was the panel beaming heavily at me , it was the wall reflecting hi frequencies only at me (because the lower frequencies had dropped away), the opposite to a standard speaker which reflects the lower frequencies .
This is easily measured and heard.
No I am not going to say the opposite, moving the exciter up a long narrow panel like the tall blonds does put the panel way out of balance, in many ways.
But it did sort of work.
I have said many times before that I am only interested in obtaining the best possible sound from my panels, hence recording them to show the sort of sound which can be obtained using many different materials.
I did this by listening to many panel materials and experiments with shapes and sizes and weights and the rest.
Patents and white papers have helped very little I'm afraid as none of them mention the sound quality, only the effects of doing this or that on measurements, not how it affects the sound.
Steve.
Eucy,
Here are the lowest eigenfrequencies predicted by LISA based on your numbers and assuming El is 10 GPa.
Highlighted in yellow are the predicted odd,odd modes. But without having strong confidence in the elastic properties (especially the transverse modulus), I would not put much stock in the model. And to be honest, I would guess that Et/El is closer to 0.1 than .05 with the 3 ply.
Do you have another sheet of the same or similar that's not mounted? You could do some tap tasting and we could probably come up with pretty good estimates of El, Et, and Glt.
Eric
Eucy,
Since we are orienting panels differently (wrt anisotropy), I'm not sure if my experience will apply to your panels. That said, I have found the best exciter location for my high aspect ratio, simply supported, panels, is at 0.33 across the short direction, and anywhere from about .3 to .45 along the length.
The 0.33 position widthwise avoids driving the 3,1 and 3.3 modes that can stack up on top of other modes and cause spikes.
Lengthwise anywhere between .3 and .45 seems to work, I just try several positions until I find the best one.
And yes, as you get further from the center, the less effectively the exciter drives the panel.
As for a second exciter: I get the best results when I put the second exciter in a mirror image position across the width (not length) or the panel. The farther away the exciters are from each other, the worse the impulse response gets. But with two exciters at .33 and .66 across the width, and the same position (say .4) along the length, the frequency response is typically virtually identical to that of a single exciter, and the impulse response may be slightly improved. For this type of panel (high aspect ratio) I have never found any other pair of positions that work better than a single exciter.
YMMV,
Eric
JohnnoG.
Thanks, I look forward to receiving them.
To be honest, when I received your first samples of proplex ,I was not expecting much.
As I had heard not so good reports of similar products ?
I have been tempted to buy a 2mm sample from wickes for comparison with my various small panels(6x9inch).
Could be interesting, or not?
Also I am tempted to add a second exciter to increase power handling of the 6mm proplex.
I will use the standard NXT positions.
The proplex has a very flat response from 40hz to 20k from the get go.
I'm not saying it is perfect, but it is so quick and easy to set up.
And maybe can be improved on ?
We shall see.
Steve.
Thanks, I look forward to receiving them.
To be honest, when I received your first samples of proplex ,I was not expecting much.
As I had heard not so good reports of similar products ?
I have been tempted to buy a 2mm sample from wickes for comparison with my various small panels(6x9inch).
Could be interesting, or not?
Also I am tempted to add a second exciter to increase power handling of the 6mm proplex.
I will use the standard NXT positions.
The proplex has a very flat response from 40hz to 20k from the get go.
I'm not saying it is perfect, but it is so quick and easy to set up.
And maybe can be improved on ?
We shall see.
Steve.
Eucy.
i have just made some new recordings using the 6mm and 4mm proplex panels .
i have tried using different sounding tracks.
the 6mm is on the left, i had to roll them off at 43hz as the exciters were bouncing off the back out of balance .
at the end of the mike and the mechanics track the exciters were very hot especially the 6mm.
so not ideal conditions , and a very basic setup.
listen using headphones.
hopefully the tracks do not sound dead ?
recorded as usual at 1m with the panels firing up the room, not pointing at the mic.
steve.
i have just made some new recordings using the 6mm and 4mm proplex panels .
i have tried using different sounding tracks.
the 6mm is on the left, i had to roll them off at 43hz as the exciters were bouncing off the back out of balance .
at the end of the mike and the mechanics track the exciters were very hot especially the 6mm.
so not ideal conditions , and a very basic setup.
listen using headphones.
hopefully the tracks do not sound dead ?
recorded as usual at 1m with the panels firing up the room, not pointing at the mic.
steve.
Hi Eric ..
Thanks for running the model
I can do static deflection testing on some ply offcuts... Won't be for a while as I'm 1500km from home
I suspect you're right about Et, esp after some coats of shellac.
This is a good discussion with some useful feedback from all involved
Cheers
Eucy
Hi Steve
I agree that hours of critical listening and adjustment is at least as important as analysis and mic testing...all parts of the mix.
After all of the materials you have tried, in which order would you rank them in terms of quality?
Eucy
Eucy.
I just get annoyed when a patent or white paper uses inferior materials and exciters, then compares the dml panel to a 2inch computer speaker, and the computer speaker outperforms the dml panel in every way.
Why not go the whole hog and stick the exciter on a brick.😀
I have only found a couple of patents that I have found useful, one of which I have lost.
I use the nxt exciter positions because supposedly this is or was , for the best performance, but is it the best sound ?
It just saves me from moving the exciter all around every new panel I make.
I have not noticed much difference in mounting the exciter in the middle and using the nxt positions on the proplex panel(not that I have made comparisons yet)
Maybe it is because the proplex panel is quite heavily damped ?
Steve.
I just get annoyed when a patent or white paper uses inferior materials and exciters, then compares the dml panel to a 2inch computer speaker, and the computer speaker outperforms the dml panel in every way.
Why not go the whole hog and stick the exciter on a brick.😀
I have only found a couple of patents that I have found useful, one of which I have lost.
I use the nxt exciter positions because supposedly this is or was , for the best performance, but is it the best sound ?
It just saves me from moving the exciter all around every new panel I make.
I have not noticed much difference in mounting the exciter in the middle and using the nxt positions on the proplex panel(not that I have made comparisons yet)
Maybe it is because the proplex panel is quite heavily damped ?
Steve.
+ @Veleric
Some months ago, I made tap tests (in a learning idea).I did it with a plywood sample : Plywood tap test
2.8mm thick, 440kg/m³, E1 = 6.2GPa, E2 = 1.15GPa.
From that Eric, you adjusted the following model to fit to the tap test results (Plywood model )
E1=6.9 GPa
E2=1.25 GPa
G=0.72 GPa
nu12=0.1
So a ratio 1:5.
I don't know which it is really. It was a top of a box of potatoes
Eucy,
Before Eric's posts I even didn't know what was this kind of tests. After some experiments, I found the method quite easy and really efficient to have an idea of the material characteristics.
Christian
@Veleric , @Eucyblues99
As the discussion about modes, models are going well, here is the first paper I have found speaking about damping in wood : Sound quality assessment of wood for xylophone bars .
Some ideas to keep to compare damping by the tap test?
Christian
As the discussion about modes, models are going well, here is the first paper I have found speaking about damping in wood : Sound quality assessment of wood for xylophone bars .
Some ideas to keep to compare damping by the tap test?
Christian
Yes, I bet the elastic properties of Eucy's plywood are in reality pretty close to your 3 ply plywood.
Eucy, Christian is right, the tapping method is a pretty easy way to estimate the elastic constants. Do you have a microphone and REW or other RTA capability?
Eric
Hi Eric
I have REW and a Dayton IMM6 calibrated mic.
I'll look into it some more, maybe compare methods, and even maybe work backwards from Chladni f0 testing to see which values of El/Et fit the result
Thanks
Eucy