I come and go from the forum, busy with other things and trying to dip back into some more experiments. Im still a newby compared to many contributors here. Yes Ive read quite a bit, and have a technical background so can understand much of it, but there is an awful lot to put together. I try to keep an open mind 
Ive done some experimenting and measuring inside and outside with a calibrated mic, trying to develop a repeatable technique that's not too onerous. A lot of results were inconclusive or showed only mild effects. Progress depends upon proper technique and careful documentation, and mine has been a bit hit-and-miss as I have been learning.
A lot of measurements were quick-and-dirty 'lets just try this inside', because dragging the amplifier, panel, mounting system, the mic and computer outdoors, even when the weather is cooperative, is a chore. Ive wasted quite a bit of time
Ive been using XPS, as the simplest, cheaply available material, and Id like to stick with that for the time being (or HD EPS), perhaps eventually with a fibreglass layer.
A few of the tests may be worth sharing here, I'll have a look back and see. A lot of what I encounter just confirms what's been said before, so there's little point repeating it. I think I can provide evidence supporting my assertions above about frames boosting base for example - I thought that was pretty uncontroversial.
I did some measurements looking at reflections from edges and corners, where I would take a FR, then revisit some of the resonant peaks with a pure sin wave, then either scan the panel with the mic, or just listen close to the panel to locate the antinodes. In many/most cases, it was between the exciter and a perpendicular edge, or a corner, which acts as a retro-reflector. This led me to believe that panel shape is very important. Interestingly, you can also look at troughs in the same way. Tune to a trough frequency with a sin wave and you will find two, occasionally three antinodes on the panel that may all be quite loud individually, but will cancel at the exact point of the mic - AKA, comb filtering in action.
Again with regard to shape, I read a few articles/papers on what was called (if I remember correctly) 'billiard ball theory', where they study the areas covered by a ball moving about a shape, bouncing off the edges. Some shapes lead to chaotic behavior. The stadium shape (a rectangle with 2 semicircles on each end) is one of those shapes. Convex internal edges also help to create randomness, because they disperse the ray (ie the ball) in different directions given a small change in trajectory. I was describing something similar when I said 'cusp' in the post you mentioned.
Anyway Im still thinking that such shapes should improve the midrange in particular. This accords with something Benjamin Zenker said in one of his papers, where he was aiming to create more of a 'travelling wave' effect than a 'standing wave'. IOW, he is also trying to minimise resonance peaks whilst keeping the surface 'alive' .
Anyway, this post is way too long. I'll post a photo tomorrow of my last experiment which used a frame and played with a different panel shape, and hopefully some relevant measurements.
Ive done some experimenting and measuring inside and outside with a calibrated mic, trying to develop a repeatable technique that's not too onerous. A lot of results were inconclusive or showed only mild effects. Progress depends upon proper technique and careful documentation, and mine has been a bit hit-and-miss as I have been learning.
A lot of measurements were quick-and-dirty 'lets just try this inside', because dragging the amplifier, panel, mounting system, the mic and computer outdoors, even when the weather is cooperative, is a chore. Ive wasted quite a bit of time
Ive been using XPS, as the simplest, cheaply available material, and Id like to stick with that for the time being (or HD EPS), perhaps eventually with a fibreglass layer.
A few of the tests may be worth sharing here, I'll have a look back and see. A lot of what I encounter just confirms what's been said before, so there's little point repeating it. I think I can provide evidence supporting my assertions above about frames boosting base for example - I thought that was pretty uncontroversial.
I did some measurements looking at reflections from edges and corners, where I would take a FR, then revisit some of the resonant peaks with a pure sin wave, then either scan the panel with the mic, or just listen close to the panel to locate the antinodes. In many/most cases, it was between the exciter and a perpendicular edge, or a corner, which acts as a retro-reflector. This led me to believe that panel shape is very important. Interestingly, you can also look at troughs in the same way. Tune to a trough frequency with a sin wave and you will find two, occasionally three antinodes on the panel that may all be quite loud individually, but will cancel at the exact point of the mic - AKA, comb filtering in action.
Again with regard to shape, I read a few articles/papers on what was called (if I remember correctly) 'billiard ball theory', where they study the areas covered by a ball moving about a shape, bouncing off the edges. Some shapes lead to chaotic behavior. The stadium shape (a rectangle with 2 semicircles on each end) is one of those shapes. Convex internal edges also help to create randomness, because they disperse the ray (ie the ball) in different directions given a small change in trajectory. I was describing something similar when I said 'cusp' in the post you mentioned.
Anyway Im still thinking that such shapes should improve the midrange in particular. This accords with something Benjamin Zenker said in one of his papers, where he was aiming to create more of a 'travelling wave' effect than a 'standing wave'. IOW, he is also trying to minimise resonance peaks whilst keeping the surface 'alive' .
Anyway, this post is way too long. I'll post a photo tomorrow of my last experiment which used a frame and played with a different panel shape, and hopefully some relevant measurements.
Christian.
The problem I have with your measurements is , you start off by saying the microphone might be in a different position and other variations.
These differences can make large differences to plots ? So that I don't know what I am looking at ?
I end up looking at a lot of squiggly lines but not knowing if it is just the position of the microphone I am looking at ?
If the microphone is not in exactly the same position and the panel not in exactly the same position, it is hard to guess what is happening ?
The change in microphone position by a few cm can change the plot considerably.
This is the problem with single point microphone positions , that response will only be at that position and distance .
I don't mean to sound negative, but I hope you will understand.
Steve.
The problem I have with your measurements is , you start off by saying the microphone might be in a different position and other variations.
These differences can make large differences to plots ? So that I don't know what I am looking at ?
I end up looking at a lot of squiggly lines but not knowing if it is just the position of the microphone I am looking at ?
If the microphone is not in exactly the same position and the panel not in exactly the same position, it is hard to guess what is happening ?
The change in microphone position by a few cm can change the plot considerably.
This is the problem with single point microphone positions , that response will only be at that position and distance .
I don't mean to sound negative, but I hope you will understand.
Steve.
Eric .
I have just re-read you post about how a tectonic panel works.
I have hinted quite strongly that this is not how I believe they work.
If you ask nicely , I would be happy to enlighten you 😁
I did try this method years ago on my EPS panel , but didn't like the sound, but obviously it must work better for heavier panels ?
Steve.
I have just re-read you post about how a tectonic panel works.
I have hinted quite strongly that this is not how I believe they work.
If you ask nicely , I would be happy to enlighten you 😁
I did try this method years ago on my EPS panel , but didn't like the sound, but obviously it must work better for heavier panels ?
Steve.
I'm must admit I am curious how you think they work. You keep hinting that you think you know something about the Tectonic panel that we all don't. You might as well just tell us.
I doubt I'll be enlightened, but I'm willing to take a chance.😁
Eric
Eric .
OK.
I will give it a go .
I've just cracked a couple of beers ,so I will have to do it tomorrow 😜
I still think it's a bit of a coincidence that tectonic blocked their pdf specs not long after I mentioned what I had noticed.
unless it's just me they have blocked 😱
has anyone else tried looking to access them?
Steve.
OK.
I will give it a go .
I've just cracked a couple of beers ,so I will have to do it tomorrow 😜
I still think it's a bit of a coincidence that tectonic blocked their pdf specs not long after I mentioned what I had noticed.
unless it's just me they have blocked 😱
has anyone else tried looking to access them?
Steve.
Thank you for this quick feedback Pway.I come and go from the forum, busy with other things and trying to dip back into some more experiments. Im still a newby compared to many contributors here. Yes Ive read quite a bit, and have a technical background so can understand much of it, but there is an awful lot to put together. I try to keep an open mind
Ive done some experimenting and measuring inside and outside with a calibrated mic, trying to develop a repeatable technique that's not too onerous. A lot of results were inconclusive or showed only mild effects. Progress depends upon proper technique and careful documentation, and mine has been a bit hit-and-miss as I have been learning.
A lot of measurements were quick-and-dirty 'lets just try this inside', because dragging the amplifier, panel, mounting system, the mic and computer outdoors, even when the weather is cooperative, is a chore. Ive wasted quite a bit of time
Ive been using XPS, as the simplest, cheaply available material, and Id like to stick with that for the time being (or HD EPS), perhaps eventually with a fibreglass layer.
A few of the tests may be worth sharing here, I'll have a look back and see. A lot of what I encounter just confirms what's been said before, so there's little point repeating it. I think I can provide evidence supporting my assertions above about frames boosting base for example - I thought that was pretty uncontroversial.
I did some measurements looking at reflections from edges and corners, where I would take a FR, then revisit some of the resonant peaks with a pure sin wave, then either scan the panel with the mic, or just listen close to the panel to locate the antinodes. In many/most cases, it was between the exciter and a perpendicular edge, or a corner, which acts as a retro-reflector. This led me to believe that panel shape is very important. Interestingly, you can also look at troughs in the same way. Tune to a trough frequency with a sin wave and you will find two, occasionally three antinodes on the panel that may all be quite loud individually, but will cancel at the exact point of the mic - AKA, comb filtering in action.
Again with regard to shape, I read a few articles/papers on what was called (if I remember correctly) 'billiard ball theory', where they study the areas covered by a ball moving about a shape, bouncing off the edges. Some shapes lead to chaotic behavior. The stadium shape (a rectangle with 2 semicircles on each end) is one of those shapes. Convex internal edges also help to create randomness, because they disperse the ray (ie the ball) in different directions given a small change in trajectory. I was describing something similar when I said 'cusp' in the post you mentioned.
Anyway Im still thinking that such shapes should improve the midrange in particular. This accords with something Benjamin Zenker said in one of his papers, where he was aiming to create more of a 'travelling wave' effect than a 'standing wave'. IOW, he is also trying to minimise resonance peaks whilst keeping the surface 'alive' .
Anyway, this post is way too long. I'll post a photo tomorrow of my last experiment which used a frame and played with a different panel shape, and hopefully some relevant measurements.
ChristianProgress depends upon proper technique and careful documentation, and mine has been a bit hit-and-miss as I have been learning.
Not with honeycomb. If you are laminating on a solid surface like balsa or closed cell then yes, vacuum helps form a tight surface and remove excess resin, assuming you use a bleeder of some sort. But vacuum does nothing to remove excess resin that fills the empty spaces in open honeycomb. The only way to avoid resin filling your honeycomb is to squeeze out all the excess first, before applying the honeycomb (or, like the pros do, use prepreg so there is not excess resin to start). That video with the thin fiberglass is a great primer because he's specifically demonstrating you can do it without prepreg. He uses a low vacuum just to keep the honeycomb pressed firmly to the skin, not to remove excess resin. This objective can be achieved instead with careful application of weights. You co see despite his efforts he had some excess resin in the honeycomb, but it's little enough it should be fine for our purposes. I believe if you do it in two steps instead of one then it will be even easier to keep the excess down.
Keep in mind that durability and a good bond is important.to this purpose but really we shouldn't be pushing the material anywhere close to it's stress point in any one spot. Unbonded areas might rattle and expand over time but if the whole surface is bonded, even very weakly then there should be no faults to propagate and it's unlikely to delaminate from DML use (but don't drop it on the floor and resist the urge to see how far you can flex it.)
Steve,
I agree it would be an improvement to add arrows or boxes to help to focus on the area I point... I hoped the text will replace it but seems no. This needs additional time I don't necessary have... but seems the condition for a more useful sharing. It goes also in the way to build a better documentation.
About the positions, my experience currently is from one day to an other one; placing the panel and the mic at "about" the same position, the changes are not very important. I remember we don't have perhaps the same reading; I am not looking for change in the range let say 2dB for now but more 10dB as it happens with suspension.
Each day I start, a first FR is done to check something not link to what I want to test has not changed.
Have a look to the FR below : 3 days (25, 28, 31 march), same panel. Mic at 1m. The panel ant the mic are put away after. I would say not too bad for measurment in a living room (litteraly where we live!). The FR shown in the previous posts are from those days.
To avoid a question... I don't focus so much above 10k.
Steve,
I am curious to know practically how we can determine objectively a panel is in one the 3 modes you list.
Pistonic is a model used in acoustic to allow some calculations based on the hypothesis all the points of the surface move from the same quantity at the same time (like the piston of a gasoline engine).
When some points of the surface are fixed, it is no more possible. The panel works in flexion.
The only clear possible case is the free floating panel but here the 1st mode of flexion is so low in frequency that, if there is, the frequency range for a pistonic mode which occurs before the 1st mode is very low in frequency, if it can produce enough SPL It is exactly for that cones are cones : the shape gives enough rigidity to reject the flexion at the end of the target frequency range. A flat membrane can't do that.
For the art panel as the edges are fixed in position with frame, it is a typical (1,1) mode that occurs first.
My understanding is that sometimes, "pistonic" is used to say all the points of the surface go in the same direction, not necessary the same value, which is not pistonic but the 1st flexion mode. The word "pistonic" is in this case perhaps more efficient in a general communication even if not completely true... which is I think you do speaking of the Tall Blond.
The area within the coil diameter is also submitted to modes. The best proof of that is the change in the FR when adding a mass at the center to stop the resonance which occurs in the 10k area.
Christian
Last edited:
Steve,
I feel pretty sure that pway has it right. When a DML is operating near the fundamental (1,1) frequency, it oscillates between the two positions below. And while this motion is not truly pistonic, at least (as Christian notes) the whole plate is moving in the same direction at the same time, so it could be reasonably considered "quasi-pistonic". But this motion is a motion of the entire plate, not just the coil area.
In the Tectonic video where they discuss multiple exciters, they do refer expanding the central radiation zone, but they never talk about that zone as being "pistonic". Of course, if you pick any small enough region of the plate, you could think of that small region as pistonic, but that's true of any region, not just the coil area.
I'm curious too, how you differentiate between "DML" and "Bending Wave". To me, bending wave and dml are just different terms for flat panel speakers. Gobel call his speakers "bending wave" and makes a big deal about his special damping features. Are you making the same distinction? That is, that "bending wave" means damped reflections, while "dml" means undamped reflections (and hence standing waves)? I can't evaluate your claim that panels go from bending wave to dml without understanding what you think the difference is. And do you mean to say that panels go from bending wave to dml as a function of frequency, or as a function of location on the panel? Sometimes you seem to be implying the former, and other times the latter. Which is it?
Eric
Thanks NaRenaud,
I wasn't really going for looks on this one, I was just looking for one step up from "prototype", to use in my basement workshop. The frame material is leftovers from a work project, and was prefinished, so I did not chose the color. But the contrast between the frame and the light birch panel does look nice, I think.
Sadly, so far there is only one, not two. I planned two but the panel I got for the second speaker turned out to be different from the first panel. It was from the same bin at Home Depot and identically labelled as "birch plywood" but one panel was 5 ply and the other was 3 ply! I didn't even notice until I went to build the second speaker and noticed a difference in weight, and then, upon closer inspection, the difference in the number of plies. I've since gone to four different store locations looking for a panel that matches the first one, but no luck so far! The first one is more efficient than any of the plywood panels I've used before, so I'm really bummed that I can't find a matching panel.
Eric
I didn't think that the vacuum was meant to remove any resin. But when gluing something, the tighter you can join the parts, the less glue you need. Of course that is the same for honeycomb as for everything else. If it is a little bit uneven, a little bit extra glue can fill in so you get a bond everywhere.
We are talking about a plate that will be subject to a lot o stress over long periods of time, where you want to really minimize weight and where slightest delamination will mean that the plate becomes useless. If vacuum would not be needed for this application, I really wonder in what applications it would be useful?
Especially since materials are quite costly and issues can show up after extended use, to me it seems just too risky to rely on my skills in "careful application of weights". And we are talking a skin with a thickness of 0.064mm that should have even pressure over about 1/4 m2 area. I know from 3d printers that when you are talking about that kind of precision, a mirror is not as flat as you think.
But of course, a pump is not cheap, and adds a lot to the cost if you are only making two plates, and it would be great if it works without. I might try it myself with glass, but especially since I'm planning to make a bunch of plates that should handle a lot of power, I guess it is worth investing in a pump anyway.
It doesn't help to use general terms like 'bending mode' or 'dml' to describe regimes of behaviour. But I can at least agree with Steve that there seem to be three.
(1) The low frequency regime, presumably lower than coincidence frequency, where cancellation between high and low pressure regions takes place more and more as the frequency drops, producing a declining response in the far field, or low frequency rolloff. As frequency decreases you have firstly edge and corner modes dominating, plus perhaps some remnant uncancelled 'pistonic' mode from the larger displacement near the exciter, culminating finally in the lone fundamental, most pistonic (1, 1) mode. In this regime, the antinodes are fewer and fewer, there is little overlap between resonant peaks, and you see the peaks as broad and distinct - culminating in (1, 1), the classic 'one-note bass'.
I think it could be useful to think of this regime as where the bending wave wavelength is approaching or exceeding the smallest planar dimension of the panel itself.
(2) The central regime, presumably above coincidence, where the sound waves in air are more efficiently created from panel oscillations, where the resonance peaks are many, the overlap is greater, and the far-field response reaches a peak. This is the regime that is characteristic of - the strength of - DML speakers. In this regime, the bending waves move about the panel, will reach the edges and reflect, and will find any places where standing waves are possible. This regime will have sharp, and often too-high resonance peaks, caused by specular reflections from the exciter to corners and edges, between parallel edges, or more complex combinations. It could be useful to think of this regime as being where the bending wave wavelength is smaller than the panel's smallest planar dimension, down to something like 6 times the panel thickness.
(3) The upper regime, where the bending wave wavelength is less than (say) 6 times the panel thickness, and we are approaching the limits for bending wave propagation in the panel. In this regime the energy in the wave is less, and (depending upon the material, surface, and panel thickness) it is absorbed before it reaches the edge of the panel. This has the happy coincidence that it tends to reduce beaming, and maintain the relatively more diffuse angular response that is an advantage of DML speakers.
Paul
(1) The low frequency regime, presumably lower than coincidence frequency, where cancellation between high and low pressure regions takes place more and more as the frequency drops, producing a declining response in the far field, or low frequency rolloff. As frequency decreases you have firstly edge and corner modes dominating, plus perhaps some remnant uncancelled 'pistonic' mode from the larger displacement near the exciter, culminating finally in the lone fundamental, most pistonic (1, 1) mode. In this regime, the antinodes are fewer and fewer, there is little overlap between resonant peaks, and you see the peaks as broad and distinct - culminating in (1, 1), the classic 'one-note bass'.
I think it could be useful to think of this regime as where the bending wave wavelength is approaching or exceeding the smallest planar dimension of the panel itself.
(2) The central regime, presumably above coincidence, where the sound waves in air are more efficiently created from panel oscillations, where the resonance peaks are many, the overlap is greater, and the far-field response reaches a peak. This is the regime that is characteristic of - the strength of - DML speakers. In this regime, the bending waves move about the panel, will reach the edges and reflect, and will find any places where standing waves are possible. This regime will have sharp, and often too-high resonance peaks, caused by specular reflections from the exciter to corners and edges, between parallel edges, or more complex combinations. It could be useful to think of this regime as being where the bending wave wavelength is smaller than the panel's smallest planar dimension, down to something like 6 times the panel thickness.
(3) The upper regime, where the bending wave wavelength is less than (say) 6 times the panel thickness, and we are approaching the limits for bending wave propagation in the panel. In this regime the energy in the wave is less, and (depending upon the material, surface, and panel thickness) it is absorbed before it reaches the edge of the panel. This has the happy coincidence that it tends to reduce beaming, and maintain the relatively more diffuse angular response that is an advantage of DML speakers.
- In all three regimes, the mode of wave propagation is the bending wave.
- Strong peaks in the upper regime could be resonance in the panel itself within the central area, with a large exciter, it could be Helmholtz resonance in the exciter cavity, or it could be break-up in the exciter mechanics.
- Membranes like art panels are definitely worth pursuing, but let's remember that they are not bending wave devices. Membranes are transverse-wave devices, they cannot support a bending moment.
Paul
Christian.
I know it is rather trivial , but I do not remember at any time using string to mount my panels.
the closest I've come, is using net curtain spring wire (whatever it is called) for my small but very heavy rigid ply panel.
for quick mounting I use masking tape, or for a more permanent mounting on my EPS panels I use a small sponge material which is quick and easy to hang, as easy as hanging a sock on the washing line😃
usually with panels that are lighter than the exciter I usually ,but not always, mount the exciter only ,and leave the panel free floating.
Steve.
I know it is rather trivial , but I do not remember at any time using string to mount my panels.
the closest I've come, is using net curtain spring wire (whatever it is called) for my small but very heavy rigid ply panel.
for quick mounting I use masking tape, or for a more permanent mounting on my EPS panels I use a small sponge material which is quick and easy to hang, as easy as hanging a sock on the washing line😃
usually with panels that are lighter than the exciter I usually ,but not always, mount the exciter only ,and leave the panel free floating.
Steve.
As promised, an image of my first experiment with shapes - in this case, ellipses, which I did last year. The exciter location (the dot at the centre left) is at the focus of all four ellipses. The 'baffle' is 1200x600 7mm ply (not rigid enough!) and the panel is 20mm XPS. At this point, the panel is attached to the frame with silicone sealant.
With an ellipse, any ray leaving the exciter would reflect once from the side of one of the ellipses, then through the other focus, then from the other side, or else (when the side is missing) move through to an adjacent ellipse. The ideas were to maximise randomness and asymmetry, maximise ray travel distance before reflecting back towards the exciter, minimise parallel edges and any edges perpendicular to a ray from the exciter. I also hoped that the four lobes would each resonate at a different low frequency, providing a (potentially designable/tunable) boost at particular frequencies in the bass, where modes are sparse. At the time, Id also read about acoustic black holes, and though that if you could machine an ABH at each focus, it should improve the impulse response and further eliminate reflections. Or just kill the sound completely :\
I used the term 'cusp' a couple of times above as a means to disperse a reflected ray in different directions, and people asked what I meant. The cusps are what I'm calling those pointy bits pointing inwards.
I believe that some of these ideas were supported by the tests and listening. I'll show some measurements tomorrow, not sure they show anything convincingly. Measurements were done inside. The best sound was with the panel supported as lightly as possible with little strips of foam squished into the 5mm gap between the baffle and the panel. Until it fell out, that is There was a clear audible bass boost from the baffle.
Main problem was I didn't notice just how close the exciter came to the baffle edge, and being the 40W thruster, resonance from there was sort of inevitable. I will return to this one I think, smooth out those cusps (I think now a smooth curve is better anyway, as waves beyond a few cm would not 'see' the point anyhow), and mount the exciter more toward the centre.
With an ellipse, any ray leaving the exciter would reflect once from the side of one of the ellipses, then through the other focus, then from the other side, or else (when the side is missing) move through to an adjacent ellipse. The ideas were to maximise randomness and asymmetry, maximise ray travel distance before reflecting back towards the exciter, minimise parallel edges and any edges perpendicular to a ray from the exciter. I also hoped that the four lobes would each resonate at a different low frequency, providing a (potentially designable/tunable) boost at particular frequencies in the bass, where modes are sparse. At the time, Id also read about acoustic black holes, and though that if you could machine an ABH at each focus, it should improve the impulse response and further eliminate reflections. Or just kill the sound completely :\
I used the term 'cusp' a couple of times above as a means to disperse a reflected ray in different directions, and people asked what I meant. The cusps are what I'm calling those pointy bits pointing inwards.
I believe that some of these ideas were supported by the tests and listening. I'll show some measurements tomorrow, not sure they show anything convincingly. Measurements were done inside. The best sound was with the panel supported as lightly as possible with little strips of foam squished into the 5mm gap between the baffle and the panel. Until it fell out, that is
There was a clear audible bass boost from the baffle.Main problem was I didn't notice just how close the exciter came to the baffle edge, and being the 40W thruster, resonance from there was sort of inevitable. I will return to this one I think, smooth out those cusps (I think now a smooth curve is better anyway, as waves beyond a few cm would not 'see' the point anyhow), and mount the exciter more toward the centre.
Last edited:
Eric.
when is a DML panel not a DML panel ?
when it is a bending wave panel 😂
I think goebel, explains it quite well.
And it saves me from getting shot down in flames again😱
He describes dml as being a resounding ,and diffused and washed out sound 😬
If there is any flack, can you point in the direction of goebel 😃
http://www.goebel-highend.de
Steve
when is a DML panel not a DML panel ?
when it is a bending wave panel 😂
I think goebel, explains it quite well.
And it saves me from getting shot down in flames again😱
He describes dml as being a resounding ,and diffused and washed out sound 😬
If there is any flack, can you point in the direction of goebel 😃
http://www.goebel-highend.de
Steve