Clear Plain Polystyrene CCCC test panel (II) :
First plot below is the FDM script output compared to PETTaLS. Note they have in common the modal decomposition. I have prepared in parallel a frequency approach (similar to a classical AC simulation in electronics) but it needs a damping model which is not implemented for now. The problem being to find the good damping model...
Probably more interesting is the next plot showing the FDM half plane SPL (blue) and the tentative of open back SPL simulation (green) compared to the measurement (orange). The good point is the model is close to the measurement in the low frequencies. There is a few dB error because the modeled back wave is not strong enough. Some analysis to have to patch it? The dipole peak seems to be at work in the 500Hz area. I can say nothing about the difference in the 70 to 300Hz because of more than possible effect of the room. Some outdoor measurements would be welcome.
Next plot is the impedance from the FDM script, similar to PETTaLS one mainly because of the mode decomposition model. A better exciter model with a variable damping should improve it.
Last and I hope not too early in regard to the youth of the open back model is the same panel with the exciter at 50/35% which puts it at 13.5cm from 3 edges the exciter. The model warns about a possible dip at 500Hz due to the acoustic feedback. I can't say it really happens at that frequency but I can say after using this position in directivity tests some months ago that I have in mind it is probably not a so good exciter placement for open back use... The problem with dips is we can't correct them. An EQ can tam the peaks but for the dips? So in a kind of function to evaluate the quality of a design, such dip should probably be severely quoted.
First plot below is the FDM script output compared to PETTaLS. Note they have in common the modal decomposition. I have prepared in parallel a frequency approach (similar to a classical AC simulation in electronics) but it needs a damping model which is not implemented for now. The problem being to find the good damping model...
Probably more interesting is the next plot showing the FDM half plane SPL (blue) and the tentative of open back SPL simulation (green) compared to the measurement (orange). The good point is the model is close to the measurement in the low frequencies. There is a few dB error because the modeled back wave is not strong enough. Some analysis to have to patch it? The dipole peak seems to be at work in the 500Hz area. I can say nothing about the difference in the 70 to 300Hz because of more than possible effect of the room. Some outdoor measurements would be welcome.
Next plot is the impedance from the FDM script, similar to PETTaLS one mainly because of the mode decomposition model. A better exciter model with a variable damping should improve it.
Last and I hope not too early in regard to the youth of the open back model is the same panel with the exciter at 50/35% which puts it at 13.5cm from 3 edges the exciter. The model warns about a possible dip at 500Hz due to the acoustic feedback. I can't say it really happens at that frequency but I can say after using this position in directivity tests some months ago that I have in mind it is probably not a so good exciter placement for open back use... The problem with dips is we can't correct them. An EQ can tam the peaks but for the dips? So in a kind of function to evaluate the quality of a design, such dip should probably be severely quoted.
Christian,
Does your model us a "Q" factor for the panel, like Pettals? Are you using the Pettals value of 10 (from acrylic)? Did you try increasing the Q in the model to see if the predicted impedance comes closer to the measured results?
Did you also measure impedance of the panel in the free (hanging) condition? I'm curious if the impedance peaks look similar, or if they are even sharper in the free condition. I only once ever built a CCCC panel, and it was before I had the impedance rig. I always assumed that a CCCC frame would add damping (even if it wasn't intended), but your measured peaks look remarkably sharp, like I might expect from a free panel.
Eric
Hello Eric
Yes currently my script is based on a constant Q value, here 10. It is the only model I know currently that gives pretty good results. A simple constant viscous damping for example doesn't give good results. Some variation with the frequency, I guess an increase at low frequency is needed. It is what comes also to simulate correctly the behavior of the exciter spider (visco-elastic damping?). See in my previous posts here the one about the reduction of the peak impedance when we had some mass on the exciter voice coil for parameter extraction.
See the plots below
The correlation between simulation and measure seems better for the impedance...
But for the FR, I would say it is more difficult. Maybe it is already to late here for me?
I tested different combination of not smoothed/smoothed curves. What gives the best correlation is to introduce smoothing in the simulation output and in the measure. Like if the higher initial damping helped in the reading by a natural smoothing
To be clear, I understood just after the pretty good results of the FDM to get the modes that the damping is as much difficult or even a more difficult topic. This impression may come from the fact I have collected papers about FDM for a long time now, many are available; compared to the papers about damping.
No i didn't measure this panel in FFFF. Unfortunately it is now glued on the frame and I have no more enough material. The IR or the spectrogram show long time ringing in CCCC. So in my opinion it is not a damped design.
FR. greem simulation (OB), orange : REW FR 5ms windowed IR no smoothing
Out of curiosity I did an experiment yesterday where I tested the impedance of the same panel/mounting but with 9 different exciters. I had no particular objective in mind other than to see how similar or different they might be.
The panel in all cases was a 584 x 406 x 5.3 mm three layer balsa "plywood", mounted to a frame using four 18 mm square blocks of Poron 92 foam located at the center of each of the four sides. The exciter in all cases was mounted at the center of the panel.
The results are shown in the plots below. The DAEX25VT-4 is shown as the top trace in both plots, as a kind of reference. Generally, I think it's interesting to see that for the most part the results are pretty similar, especially for all the exciters in the first group, especially between about 100 Hz and about 8kHz. I consider this to be a light to moderately damped panel as many of the impedance peaks are pretty sharp, maybe the differences between exciters would stand out more with an undamped construction.
The second group is a little more varied, especially the Tectonic one (TEAX...).
Any peaks below 30 Hz primarily reflect the exciter "magnet" resonance. the peaks above that are all panel resonances, thought lowest two can be shifted (higher) more or less by the exciter itself. The stuff above 8 kHz perhaps reflects voice coil breakup, at least that is what I have been suspecting.
Eric
The panel in all cases was a 584 x 406 x 5.3 mm three layer balsa "plywood", mounted to a frame using four 18 mm square blocks of Poron 92 foam located at the center of each of the four sides. The exciter in all cases was mounted at the center of the panel.
The results are shown in the plots below. The DAEX25VT-4 is shown as the top trace in both plots, as a kind of reference. Generally, I think it's interesting to see that for the most part the results are pretty similar, especially for all the exciters in the first group, especially between about 100 Hz and about 8kHz. I consider this to be a light to moderately damped panel as many of the impedance peaks are pretty sharp, maybe the differences between exciters would stand out more with an undamped construction.
The second group is a little more varied, especially the Tectonic one (TEAX...).
Any peaks below 30 Hz primarily reflect the exciter "magnet" resonance. the peaks above that are all panel resonances, thought lowest two can be shifted (higher) more or less by the exciter itself. The stuff above 8 kHz perhaps reflects voice coil breakup, at least that is what I have been suspecting.
Eric
It seems the exciters change the first modes, let say below 150Hz probably by their mass and stiffness. Above as you point it there are almost no differences. The frequency of the modes are the same.
For the damping, I would look more at the peak values and here the differences are visible above 150Hz.
The TEAX32C30 is an unusual exciter : 10.2g as voice coil mass and 100Hz as blocked coil resonance so a very heavy voice coil (even if the inductance remains classical 0.1mH) and and a light magnet. Its damping is very high also at 2kg/s when usually is more 0.2...
Exactly. And adding the exciter always increases the frequency of those low frequency modes, never decreases. I noticed that first when I was doing tap testing. If I did a tap test with and without an exciter the frequency of the fundamental mode was always higher with the exciter than without, while higher frequency modes were virtually unchanged in frequency. I recall being a little surprised, as I thought the added mass of the exciter would tend to decrease the modal frequency, rather than increase it. But I guess it's more like the exciter spring acts like a stiffener with the mass to back it up.
Eric