I came across this old BBC video, when the video got to 13 minutes, I had to laugh, as it reminded me of the good old days before DML'S 🤣
Steve.
Christian,
I have also been thinking about and exploring ways to support a panel in a way that "mimics" a free panel, but is more robust and practical than just hanging it, plus avoids the problem of vibrations of the hanging members, and also provides the opportunity to provide support for the exciter. As you know, I have doubts about any true benefit of the "free" panel (other than simplicity of construction), over a well damped perimeter mounted panel, but it is an interesting technical exploration, in any event.
Originally, my thought was to use only three mounting points (the minimum to define a plane), using very soft foam to minimize any restraint. Also, I thought that the mounting points should avoid the perimeter, since virtually all the modes of a free plate have antinodes along the perimeter. Also, I thought the mounting points should be in an irregular pattern, to minimize the impact on any given mode. And finally to avoid mounting points near the exciter. But after trying many such mountings, they all suffered from ringing at one frequency or another. So I abandoned that approach.
But I recalled that hanging the panel from the centers of two adjacent sides helped minimize ringing. For example, these two spectrograms compare hanging from two points along the same edge (first images), vs. hanging from the center points (second images). Clearly, there is less ringing in the second mounting, presumably because it inhibits some of the modes most prone to ringing.
Taking inspiration from the second hanging method, and playing around a bit, I found that mounting the panel at these five points (using 18 mm squares of Poron foam) provided interesting results. Ringing isn't eliminated (still some at 60 Hz), but it's much better than the first "hanging" case. I have to admit that it is about the best sounding of any "free-ish" panel that I have tested.
And here's a comparison of a few impedance curves to provide some perspective. On top is the case of hanging from two wires on the same side, probably the worst case of ringing. The middle curve is with the "five points" mounting, and the last is for mounting with Poron foam around the entire perimeter. The "five points mounting" seems to provide very strong damping at low frequencies, but at high frequencies, the "five point" looks a lot more like the "free" (hanging) panel than the damped (Poron surround) panel.
I should note for the record that all these results are using a 2.1x584x406 mm cf/balsa sandwich panel, and a VT25-4 exciter mounted at 0.4/0.4.
Eric
Hello Eric,
Thank you for your post. It is fully in what I am considering now.
Where to start..?
Maybe by this picture:
It is from the current status of my FDM simulation script. Frankly, I don't know up to which level this script is able to simulate a panel. What we know currently is its outputs about modes make sense and are closed of FEM simulation. I have implemented the electrical impedance, the SPL in a half space (like the SPL from PETTaLLS) and just few days ago the SPL of an open back panel but also a tentative to find possible suspension points.
On the left of the picture is a view of the panel simulated. Here a free edges 300x420mm with something suppose to be a polystyrene and a DAEX25VT at 0.4/0.4.
The 2 other plots are a tentative to see what are the possible suspension points. It is the sum of the speed at each points (blue dots on the left) above a certain frequency (200Hz). The points with the lower movements (or better to say lower contribution to the SPL) are the darker one.
Why 200Hz? It is because if there is a compromise to be done, better is to alter the response below this threshold than above.
As you mentioned, the edges are anti-nodes (light blue). There are even more movements at the corners (yellow to red). It is maybe the reason of the advice to round them.
There is no point of zero speed so any tentative to hang the panel should change the behavior somewhere in the FR.
But what appear from this picture is some points not far from the edges have the minimum speed. I haven't tried for now to confirm it with PETaLLS...
The plot on the right is the same with the minimum speed points set to 0 (black dot) and the points just above set to give white dots.
2 small areas appear at about 10% from the corners far from the exciter.
As I am more working on the script currently, I haven't tested them. From a simulation with those points fixed, the FR was changed more than expected. If I think those areas are a good information, nothing says the simulation is right about the FR.
If you are interested, fill free to add this to your tests.
In addition to be elements in the research of suspension points, your test cases are excellent inputs for simulation.
Is the CF/balsa poron of this panel the same for the directivity tests? I have in the paper about directivity 2 coincidences frequency: 5300 and 15000Hz.
Could you remind me its areal mass, how the stiffness are oriented according the edge lengths?
A bit more challenging would be to evaluate the stiffness of your poron pads. I am in the same idea than you that fixed points will probably not do the job. Better would be to have some soft points able to maintain the panel without restricting to much the movement.
This leads me to an other idea... Those days, I have been laying slabs in my garden. Below I use geotextile (polypropylene fabric?). It might be a possible material to make custom noiseless (? to be tested) hanging solution.
Christian
Great progress Christian.
Yes, that's good evidence that your FDM model is working. The corners are antinodes for virtually every "free" mode, so it makes sense for them to be "yellow to red".
That is interesting. I was wondering myself about where the minimums would be. I ws inclined originally to try areas closer to 20% from the corners, rather than 10%, but I was really just guessing.
Yes, still that panel, although I plan to make a similar one soon but about half the thickness, so the 2 coincidences should be about doubled then.
But for the current panel, the moduli are about 25 GPa and 23 GPa in the length (584 mm) and width (406 mm) directions, respectively, with an areal mass of 0.945 kg/m2 (450kg/m3 x 2.1 mm).
True. However, despite my intention that they not be "fixed" points, I think they turn out to be nearly fixed. If I model the plate with FEA and assume them to be fixed points, the model predicts the modal frequencies pretty well. So, while the pads feel very soft, they still don't move much.
Eric