Sure can!
I don't recall catching that. Any idea which of the videos they talk about that in?
Nor are the Gobel panels rounded. Have you seen any good technical sources that describe how (or if) rounded corners actually improve performance?
Eric
Which is very neat.
Rounded corners were introduced to solve a problem of energy being trapped in the corners. This kind of links to the Godel approach which approximates and infinite plate by trapping reflections in the 45degree 35 degree alternating slots.
I have tried with large radius and without and the difference is not significant
Rounded corners were introduced to solve a problem of energy being trapped in the corners. This kind of links to the Godel approach which approximates and infinite plate by trapping reflections in the 45degree 35 degree alternating slots.
I have tried with large radius and without and the difference is not significant
Burnt,
Have you ever seen this effect discussed in any reliable technical source? The concept is still vague to me.
Eric
Some panels do have a tendency to flap and buzz at the corners ,rounding the corners does seem to sort it out.
I only have a few panels with rounded corners, one ply panel I glued some wood to brace the corner like a splint.
Tin can effect was only briefly mentioned and he said it could be used to aid hf, which I thought was a bad idea.
I think it was on YouTube, there are a few with there 2 guy's casually talking not sure which.
Steve
I only have a few panels with rounded corners, one ply panel I glued some wood to brace the corner like a splint.
Tin can effect was only briefly mentioned and he said it could be used to aid hf, which I thought was a bad idea.
I think it was on YouTube, there are a few with there 2 guy's casually talking not sure which.
Steve
I have a very vague memory from a few years ago, reading a patent paper or tech paper or maybe it was from a forum post (AC?) that mentioned leaving the corners unrounded tended to increase the bass...Sorry but which patent/ paper/forum that was in is no longer in my brain...damn I hate getting old...
Technically this might make sense from a modal point of view as the slightly larger panel area would support more modal points...but it would seem to me to be very minor. On the other hand, the TI guy did mention in his first vid (I think) that rounding the corners definitely smoothed the response, but that rounding the edges seemed to have no effect...
geo
Last edited:
As to the "Tin Can" Effect, I believe its mentioned in one of the Tectonics Videos on youtube:
YouTube
Look at the videos which have the 2 guys in them doing a tech talk. Theres a series of 21 videos...one of them briefly mentions the tin can effect...not sure which one, but all the vids are really a great intro into how DML's work and why.
OK found it. .. its in this vid: YouTube they mention the "Oil Can" effect very briefly.
geo
YouTube
Look at the videos which have the 2 guys in them doing a tech talk. Theres a series of 21 videos...one of them briefly mentions the tin can effect...not sure which one, but all the vids are really a great intro into how DML's work and why.
OK found it. .. its in this vid: YouTube they mention the "Oil Can" effect very briefly.
geo
Last edited:
@Veleric
Here is the patent
US6031926A - Panel-form loudspeakers
- Google Patents
From the patent
"There is, however, one effect that it has been found can be beneficial to take into specific account, at least as useful refinement, namely, for a substantially rectangular said member, the resonant modes associated with is diagonal dimension. Assuring at least mainly helpful contribution to desired distribution(s) of resonant modes can be aided by cropping or trimming or curving of corners, or forming to such shapes, though see further below regarding anisotropy of bending stiffness. Sensible practical limits for consideration of this refinement as to resulting shape and transverse area reduction arise naturally from approaching some other analysable shape, e.g. super- ellipses or at least regular polygons, say preferably not reducing the diagonal beyond larger or smaller of side dimensions. Useful, though not essential, further refinement is seen in such corner cropping or trimming being differential, at least for corners of different diagonals, whether to different extents for each, say in a mutual relationship producing a similar "matching" (for diagonals- related resonant modes) to that above for the side dimensions, and perhaps further relative to or effectively continuing or complementing or usefully compounding the sides- dimension matching ratio. Application of effective shortening to one diagonal only can be enough, in practice, say to each end equally for a resulting shortening to medially between full diagonal dimension and longer side dimension, preferably to attain progression of side and diagonal dimensions at or approaching maintaining or repeating above preferential aspect ratio for successively increasing values, conveniently so approaching by substantial evenness of dimensional differences or a straight-line dimensional relationship. Satisfactory shortening of only one diagonal of a said rectangular member, at least idealised for isotropy of bending stiffness, is seen at about 15% or about 10% short for above about 13.4% and about 37% aspect ratios, respectively."
It's an interesting early patent from Azima filed March 1997 and includes a lot of material on shape, the effect of anisotropy etc. However, it reads to me as part research expose and part extrapolation so treat with a degree of scepticism.
Burnt
Here is the patent
US6031926A - Panel-form loudspeakers
- Google Patents
From the patent
"There is, however, one effect that it has been found can be beneficial to take into specific account, at least as useful refinement, namely, for a substantially rectangular said member, the resonant modes associated with is diagonal dimension. Assuring at least mainly helpful contribution to desired distribution(s) of resonant modes can be aided by cropping or trimming or curving of corners, or forming to such shapes, though see further below regarding anisotropy of bending stiffness. Sensible practical limits for consideration of this refinement as to resulting shape and transverse area reduction arise naturally from approaching some other analysable shape, e.g. super- ellipses or at least regular polygons, say preferably not reducing the diagonal beyond larger or smaller of side dimensions. Useful, though not essential, further refinement is seen in such corner cropping or trimming being differential, at least for corners of different diagonals, whether to different extents for each, say in a mutual relationship producing a similar "matching" (for diagonals- related resonant modes) to that above for the side dimensions, and perhaps further relative to or effectively continuing or complementing or usefully compounding the sides- dimension matching ratio. Application of effective shortening to one diagonal only can be enough, in practice, say to each end equally for a resulting shortening to medially between full diagonal dimension and longer side dimension, preferably to attain progression of side and diagonal dimensions at or approaching maintaining or repeating above preferential aspect ratio for successively increasing values, conveniently so approaching by substantial evenness of dimensional differences or a straight-line dimensional relationship. Satisfactory shortening of only one diagonal of a said rectangular member, at least idealised for isotropy of bending stiffness, is seen at about 15% or about 10% short for above about 13.4% and about 37% aspect ratios, respectively."
It's an interesting early patent from Azima filed March 1997 and includes a lot of material on shape, the effect of anisotropy etc. However, it reads to me as part research expose and part extrapolation so treat with a degree of scepticism.
Burnt
Last edited:
Theory Theory
So treat this as an experiment to test out something new. I make no claim for any superior performance but obviously I hope for something interesting and good.
Here are the factors influencing the design.
1. Long and thin ply panels, in this case 0.3 M by 1.2M by 3mm thick, seem to reach down to 30Hz
2. Long and thin also seems to provide better mode distribution - hat tip to Veleric for that one
3. Non-parallel sides may also provide benefits - hat tip to Spedge and it is claimed by Farad Azima in his 1997 patent
4. Edge treatment via absorbing elastomeric layer to absorb edges reflection and simulate an ideal infinite plate
5. Single exciter per panel appears to be preferable - validated by various sources
6. And finally, multiple panels to allow higher SPL at 30Hz without reaching exciter excursion limit.
Images below illustrate the array.
Normally I would build it first and then let you know what happened but I can’t build anything until the lock-in in France ends in a few weeks time.
Please throw rocks and suggestions as you see fit.
Burnt
So treat this as an experiment to test out something new. I make no claim for any superior performance but obviously I hope for something interesting and good.
Here are the factors influencing the design.
1. Long and thin ply panels, in this case 0.3 M by 1.2M by 3mm thick, seem to reach down to 30Hz
2. Long and thin also seems to provide better mode distribution - hat tip to Veleric for that one
3. Non-parallel sides may also provide benefits - hat tip to Spedge and it is claimed by Farad Azima in his 1997 patent
4. Edge treatment via absorbing elastomeric layer to absorb edges reflection and simulate an ideal infinite plate
5. Single exciter per panel appears to be preferable - validated by various sources
6. And finally, multiple panels to allow higher SPL at 30Hz without reaching exciter excursion limit.
Images below illustrate the array.
Normally I would build it first and then let you know what happened but I can’t build anything until the lock-in in France ends in a few weeks time.
Please throw rocks and suggestions as you see fit.
Burnt
Attachments
Is it 4 identical panels? Or 1 with 3 braces? Maybe place the exciters slightly differently from each other to excite different eigenmodes?
/Anton
@onni
It is 4 separate panels. I have tried with braces and without in the past and I remain agnostic on them. The reason I am using four panels here is to reduce the excursion of the exciter to levels seen at circa 100Hz so I do not anticipate needing them. For a halving of frequency you need 4X the area for an equal SPL so I am in the right ball park I think. I agree that the exciter position can be experimented with when I get a chance to build them
@Jamu - I agree you could easily try them in a 2 x 2 configuration.
Burnt
It is 4 separate panels. I have tried with braces and without in the past and I remain agnostic on them. The reason I am using four panels here is to reduce the excursion of the exciter to levels seen at circa 100Hz so I do not anticipate needing them. For a halving of frequency you need 4X the area for an equal SPL so I am in the right ball park I think. I agree that the exciter position can be experimented with when I get a chance to build them
@Jamu - I agree you could easily try them in a 2 x 2 configuration.
Burnt
Old panel
Burntcoil.
This was a 6ft tall 1inch eps panel using the same shape as one of your panels,I clamped the panel between two pieces of wood in a vice at an angle ,shown by the wood at the bottom, this tilted the panEl to the side ,which looked better and also made for no parallel sides on the panel.
The panel was destroyed by the culprit sitting next to the panel ,this was what was left,it still sounds very good, which shows how flexible dml is.
The original idea was to have the thin part of the panel at the top and tapered at the top in the sloping side to prevent any paralleled sides,but having it wide at the top saved a lot of space and also when tilted gave me no parallel sides without cutting anything!!
I like win win situations.
Steve
Burntcoil.
This was a 6ft tall 1inch eps panel using the same shape as one of your panels,I clamped the panel between two pieces of wood in a vice at an angle ,shown by the wood at the bottom, this tilted the panEl to the side ,which looked better and also made for no parallel sides on the panel.
The panel was destroyed by the culprit sitting next to the panel ,this was what was left,it still sounds very good, which shows how flexible dml is.
The original idea was to have the thin part of the panel at the top and tapered at the top in the sloping side to prevent any paralleled sides,but having it wide at the top saved a lot of space and also when tilted gave me no parallel sides without cutting anything!!
I like win win situations.
Steve
Attachments
Burnt,
Wow, that's a mouthful. Thanks. I think you actually meant to cite this patent however:
https://patentimages.storage.googleapis.com/bb/ea/e0/90906b2a128e4e/US6904154.pdf
See bottom of column 9.
I read through the one you posted and could not find the quote, but then found it it the other (later and longer!) patent.
Eric