FWIW I've made exactly one spherical speaker and it was the worst speaker I've ever made. The coloration that thing made was just horrible to listen to, and it was chopped up pretty quickly. While a sphere may be ideal for diffraction characteristics, it is near the worst shape possible for the back wave. It might be OK if fully stuffed with fiberglass and sealed.
This rhomboid thing is a much better idea IMO. It would tend to discourage and break up most internal standing waves I suspect. Also, structurally a polyhedron like this is incredibly strong, equivalent to a thin walled box with a forest of internal reinforcement. Also, the smallish spans and irregular shapes of the plates will tend to reduce plate resonances.
"I actually look forward to making one of these 'hypercubes' because they look cool and would be fun to build. But I bet if you measured a spherical speaker with the same volume it will be very similar."
I love this experimental attitude and am glad you may attempt a build in the future. I am betting against you however with regards to making very similar observations with a spherical enclosure. Remember that my hypercubes were measured without stuffing. I'm guessing that if I were to do the same with a spherical enclosure (which I did not) there would exist a strong resonance at one narrow frequency range vs. what I observed in the hypercube enclosures (low-level wide-band resonance). Good luck and please don't hesitate with further questions regarding your build.
I love this experimental attitude and am glad you may attempt a build in the future. I am betting against you however with regards to making very similar observations with a spherical enclosure. Remember that my hypercubes were measured without stuffing. I'm guessing that if I were to do the same with a spherical enclosure (which I did not) there would exist a strong resonance at one narrow frequency range vs. what I observed in the hypercube enclosures (low-level wide-band resonance). Good luck and please don't hesitate with further questions regarding your build.
While it is true that theories are always subject to revision, I don't see the relevance to Thiele and Small. I wasn't aware that they developed a theory. I thought that they developed some formulas to transform the measured electromechanical parameters of loudspeaker drivers into compound parameters which simplified the process of designing speakers. You appear to be asserting that Thiele-Small does not adequately describe the behaviour of the speaker under discussion. If so, this is quite an extraordinary claim. Which, of course, requires extraordinary proof. I look forward to your explanation.
Gentlemen:
As time goes by I hope to present more and more evidence that you may find interesting. Greg is better situated than I am to take such data, at the moment, and I am indebted to him for his willingness to testify to things that will earn him the scorn of some listeners. Sometimes the data does seem almost too good to be true. But his results agree with my own experiences.
Science grows in two ways. It grows incrementally, such as in the case of biology when a new species is cataloged, or in astronomy, when a celestial object is added to the list
It also grows when phenomena intrude that do not fit into mainstream theory. This has happened often enough for the term "paradigm shift" has been invented for it. When it happens, it sometimes takes a lot of independent replication for the mainstream authorities to accept it. Later, of course, it seems strange that it was resisted.
Now we all know that people make lots of claims, often pretty zany, and thank you Scottmoose for providing us with those excellent examples of that. Faster than light? The only scholarly evidence for that is the current neutrino transit time data from Italy, and they are being hotly debated as we speak. I can only scratch my head that the application you mention was granted. You would think we would have heard about it if someone had succeeded in proving such an effect. NASA would sure like to have FTL communications with the Mars probes...
From the years I have put in as a web developer, I can work up some ASP pages on my website to help anyone who wants to calculate piece sizes for given internal volumes. I will also investigate ways to provide precut pieces for those who may wish to investigate this for themselves. In the meantime, we shall continue taking data and reporting our results.
I can tell you this. I have rarely constructed a rhombic dodecahedron loudspeaker (and I have made many, over the years) which failed to satisfy those who witnessed it. (I say rarely because I failed *once*...and I will be discussing the reasons for it soon enough.) I have executed the design in wood, plastic, and metal, in sizes ranging from 4 inch fullranges up to large cabinets using 12 inch Radian coaxials. I have demonstrated them for experts at defense contractors and concert venues. I am glad Greg is the one presenting waterfall plots etc., because I am well aware that my long association with this design makes my claims attackable on the basis of ego involvement.
As time goes by I hope to present more and more evidence that you may find interesting. Greg is better situated than I am to take such data, at the moment, and I am indebted to him for his willingness to testify to things that will earn him the scorn of some listeners. Sometimes the data does seem almost too good to be true. But his results agree with my own experiences.
Science grows in two ways. It grows incrementally, such as in the case of biology when a new species is cataloged, or in astronomy, when a celestial object is added to the list
It also grows when phenomena intrude that do not fit into mainstream theory. This has happened often enough for the term "paradigm shift" has been invented for it. When it happens, it sometimes takes a lot of independent replication for the mainstream authorities to accept it. Later, of course, it seems strange that it was resisted.
Now we all know that people make lots of claims, often pretty zany, and thank you Scottmoose for providing us with those excellent examples of that. Faster than light? The only scholarly evidence for that is the current neutrino transit time data from Italy, and they are being hotly debated as we speak. I can only scratch my head that the application you mention was granted. You would think we would have heard about it if someone had succeeded in proving such an effect. NASA would sure like to have FTL communications with the Mars probes...
From the years I have put in as a web developer, I can work up some ASP pages on my website to help anyone who wants to calculate piece sizes for given internal volumes. I will also investigate ways to provide precut pieces for those who may wish to investigate this for themselves. In the meantime, we shall continue taking data and reporting our results.
I can tell you this. I have rarely constructed a rhombic dodecahedron loudspeaker (and I have made many, over the years) which failed to satisfy those who witnessed it. (I say rarely because I failed *once*...and I will be discussing the reasons for it soon enough.) I have executed the design in wood, plastic, and metal, in sizes ranging from 4 inch fullranges up to large cabinets using 12 inch Radian coaxials. I have demonstrated them for experts at defense contractors and concert venues. I am glad Greg is the one presenting waterfall plots etc., because I am well aware that my long association with this design makes my claims attackable on the basis of ego involvement.
Don,
If Thiele and Small did not call their work a "theory" than I accept your correction. I called it that because any formulaic attempt to model the physical behavior of a system well enough facilitate practical designs seems to me like a theory. I would not denigrate their work by saying they were just doing curve-fitting.
What I am saying, as you correctly infer, is that Thiele-Small analysis *appears* to fail in the case of the rhombic dodecahedron enclosure. I say this because I have seen lower distortion, improved efficiency, cleaner waterfalls, and shallower bass rolloff in RD systems made by cannibalizing commercial systems designed by competent engineers.
The first such system converted was a JBL L40 back in the late Seventies. Back then the measurement technology was a little cruder than what we have today, of course. I am pretty sure the JBL engineers know what they are doing...but our frequency response curve was better. This was a study in which the JBL speaker was measured, then its woofer, tweeter, and crossover were removed and placed in our sheet aluminum (!) rhombic dodecahedron and the measurement was repeated. We chose unpadded sheet aluminum for that study because of its high acoustic reflectivity -- we were trying to give the JBL box every advantage (it was a box thickly padded with fiberglass and even had a basket of padding attatched to the back of the woofer -- which we of course removed before we put the woofer into our own cabinet). This was the first time we obtained independent testing, and it was not the last. It was over thirty years ago, and I have moved several times over the years, but I will do my best to find a copy of the report and scan it for you to peruse.
If Thiele and Small did not call their work a "theory" than I accept your correction. I called it that because any formulaic attempt to model the physical behavior of a system well enough facilitate practical designs seems to me like a theory. I would not denigrate their work by saying they were just doing curve-fitting.
What I am saying, as you correctly infer, is that Thiele-Small analysis *appears* to fail in the case of the rhombic dodecahedron enclosure. I say this because I have seen lower distortion, improved efficiency, cleaner waterfalls, and shallower bass rolloff in RD systems made by cannibalizing commercial systems designed by competent engineers.
The first such system converted was a JBL L40 back in the late Seventies. Back then the measurement technology was a little cruder than what we have today, of course. I am pretty sure the JBL engineers know what they are doing...but our frequency response curve was better. This was a study in which the JBL speaker was measured, then its woofer, tweeter, and crossover were removed and placed in our sheet aluminum (!) rhombic dodecahedron and the measurement was repeated. We chose unpadded sheet aluminum for that study because of its high acoustic reflectivity -- we were trying to give the JBL box every advantage (it was a box thickly padded with fiberglass and even had a basket of padding attatched to the back of the woofer -- which we of course removed before we put the woofer into our own cabinet). This was the first time we obtained independent testing, and it was not the last. It was over thirty years ago, and I have moved several times over the years, but I will do my best to find a copy of the report and scan it for you to peruse.
Gentlemen:
Why Hypercube Speakers Are Better Futiquity
Here is a post on my blog presenting more data. These plots were done by an engineer in Boston in the mid-Nineties, using MLSSA testing. When he was finished, he called me in Maryland, where I was living at the time, to ask me exactly how I had calculated the volume of enclosure I had used to get such good results. I was embarrassed to tell him I had not "calculated" what box volume to use. I just made the RD big enough to fit the woofer on the square baffle plate. Perhaps I should have made up some scientific-sounding answer for him.
I apologize for the crudeness of the images. He sent paper printouts which I scanned in, and maybe I could have done a better job of that. This was a 2-way system with a 8" Dynaudio woofer and a dome tweeter.
You can click on the graph thumbnails to see larger versions. It's pretty clear from the graphs that:
(1) the "hypercube"-enclosed speaker had significantly lower distortion, particularly in the lower range where the enclosure would be expected to have more of an effect -- since the tweeter had a sealed back.
(2) the "hypercube" waterfall is clearly superior to the standard speaker. It dies away much more within the time window of the measurement, again, especially in the lower portion of the audio spectrum.
Why Hypercube Speakers Are Better Futiquity
Here is a post on my blog presenting more data. These plots were done by an engineer in Boston in the mid-Nineties, using MLSSA testing. When he was finished, he called me in Maryland, where I was living at the time, to ask me exactly how I had calculated the volume of enclosure I had used to get such good results. I was embarrassed to tell him I had not "calculated" what box volume to use. I just made the RD big enough to fit the woofer on the square baffle plate. Perhaps I should have made up some scientific-sounding answer for him.
I apologize for the crudeness of the images. He sent paper printouts which I scanned in, and maybe I could have done a better job of that. This was a 2-way system with a 8" Dynaudio woofer and a dome tweeter.
You can click on the graph thumbnails to see larger versions. It's pretty clear from the graphs that:
(1) the "hypercube"-enclosed speaker had significantly lower distortion, particularly in the lower range where the enclosure would be expected to have more of an effect -- since the tweeter had a sealed back.
(2) the "hypercube" waterfall is clearly superior to the standard speaker. It dies away much more within the time window of the measurement, again, especially in the lower portion of the audio spectrum.
Gentlemen:
Arrgh! Correction: The data in the above post were taken using 8 inch Radian coaxials. I was just looking at the waterfalls and saw the Radian label on one of them. My mistake.
My remarks about the tweeter still apply, however, because the tweeter -- located at the center of the woofer and using the woofer cone as a kind of extension to its horn compression driver -- had no way of venting back into the enclosure.
Arrgh! Correction: The data in the above post were taken using 8 inch Radian coaxials. I was just looking at the waterfalls and saw the Radian label on one of them. My mistake.
My remarks about the tweeter still apply, however, because the tweeter -- located at the center of the woofer and using the woofer cone as a kind of extension to its horn compression driver -- had no way of venting back into the enclosure.
Thanks sharing.
Wonder could there be placed a temporary lets say simple 3x3 feet front baffle that kind of clamped on the two measured boxes to bring the two contenders at same baffle step/diffraction specs and then see if better specs for hypercupe already measured hold on now when front baffle is same, or isn't this going to work.
Wonder could there be placed a temporary lets say simple 3x3 feet front baffle that kind of clamped on the two measured boxes to bring the two contenders at same baffle step/diffraction specs and then see if better specs for hypercupe already measured hold on now when front baffle is same, or isn't this going to work.
You left out all the really interesting effects and observations mentioned in the second link.
I have to admit I'm a bit skeptical though - when I hear about speaker cones that produce sound without moving or ant hills that get up and walk away from the house when the music starts playing, I can't help wondering if those sorts of effects can be reproduced without the use of recreational pharmaceuticals.
IMHO, when most normal people reminisce about experiences like that years later, they tend to say things like: "Man that was some good acid!", not: "Dude, we debunked the whole of physics from Newton to Einstein, why won't anyone take us seriously?".
I am interested too. Please, do that. An include cutting angles in calculations.
Perhaps it would sound like a Nautaloss?
http://www.diyaudio.com/forums/full-range/247598-nautaloss-ref-monitor.html
Repeat the measurements at a greater distance, at least 1 meter (40") with appropriate gating, you will see that there is no increase in efficiency versus simple sealed box.
It is very difficult to repeat exactly the same 2" measuring distance in both cases (1/16" error will make a difference), I believe it is the source of that illusive "3 dB increasing efficiency".
Besides the better stiffness of the dodecahedron sides vs. simple rectangular box, producing clearer waterfall, there is no advantage, and the making of the dodecahedron is way more complex. Simple rectangular box with adequate bracing/damping will produce the same fast decay.
As for the rest on the Mathew Kennedy site, I agree with Sreten – it is a nonsense.
Agreed. But maybe it will let you hear things before someone else hears it but you don't quite know what you heard?
I will be happy to do this as soon as I can (within a few days). I originally planned to do outdoor measurements but did not anticipate how long I would end up waiting for the acryic pieces. 2 of the 3 orders had to be redone to meet spec and it is a bit cold here in NY (Long Island) for me to drag my equipment out to the middle of the yard. If you're interested in these measurements being made in my 11' wide room with cathedral ceiling that peaks at just over 8' then I will happily comply.
As far as a braced sealed box potentially having similar performance, I can offer the following personal experience at this point: The panels of a typical box speaker seem to resonate over a fairly narrow range. Certainly we can raise the pitch that excites the panel by applying stiffening, however the resonances that remain (one predominant freq per panel) seem to disagree to an extent with the driver in terms of
phase/polarity. My little sealed boxes are quite stiff yet sound pretty crappy when compared to the hypercubes while listening straight on and complete garbage when listening from behind the enclosures. The Hypercubes actually sound really good from the rear (just softer and rolled off in the top octave). Some will say this is
the result of improved baffle diffraction but I think that is just part of it. I was thinking I should provide measurements from the rear of each enclosure at some point. What really blows my mind about the hypercubes is that they sound so good (especially in the bass range) without any stuffing. They seem to not need the "band-aid" that is foam or fiberglass stuffing (and stuffing a small enclosure can only absorb so much in the bass range). I'm not as much of an expert on traditional speaker enclosure design as some on this forum so please correct me if I say something wrong in regards to sealed boxes. --Greg
To clarify some of my reasoning from my previous post, if we design a golden ratio closed box, stiffen the hell out of the panels so that they resonate in the khz range and then stuff the box and dampen the panels, don't we still have the issue of rear-wave reflections in the bass range shouting back at the driver? I think what I hear with the hypercubes is a shocking reduction (dare I say elimination?) in the level of this low freq interference. I have objectively observed much less harmonic distortion in the bass range already. I think further testing will need to be done outdoors so I can extend the waterfall plots into the bass range. --Greg
Gentlemen:
Personally I feel one reason that the acoustical performance of this shape has not been thoroughly explored long before I stumbled across it is because the pieces can be tricky to cut. It would be nice if the angles were something simple like 90 or 60 or 30 degrees, but this is not the case. The best results I have had to date involved using a CAD/CAM 3-axis overhead router with two different cutter heads for the 45 and 30 degree bevels. Once I programmed it, it was able to cut out the pieces already beveled, but we had to pay for a custom bit since cabinetry rarely involves 30 degree bevels. The 45 degree bevel bit was standard. Here's a picture of some different size enclosure cut with the router. I'm no cabinet maker, but with this tool we got sharp results:
http://www.matthewrkennedy.com/3encl2.jpg
For those who are interested, the angles can be derived with some basic math. This comes from the fact that the RD can be formed by drawing the diagonals of a 1x1x1 cube -- which cuts the cube into 6 inward-pointing pyramids -- and then turning it inside out as you know, the interior diagonals of a unit cube are length √3 and the diagonals of its faces are √2.
I hesitate to bore you with the details, but if you draw the major and minor diagonals of the rhombus, this slices it into 4 identical right triangles. If you look at one of these triangles, you will see that its side lengths are: √3/2, √2/2, 1/2. Since size does not affect angles, it is simpler to think of these as √3, √2, 1 for the trig solution. We plug these into an inv sin or inv cos function and we get the values of the angles -- or rather, the values for half of them, since we cut each angle in half when we sliced the rhombus into 4 triangles. I'll just use the Windows calc.exe program.
For the smaller angle: alpha = 2 x inv sin(.5/.866) = 70.5311 degrees
For the larger angle, beta = 2 x inv sin(.707/.866) = 109.4515 degrees
The bevel angle for the rhombus is 30 degrees. In other words, cut away 30 degrees leaving 60 degrees of wood so that when two pieces are joined you get the rhombic dodecahedron's 120 degree dihedral angle (angle between planes).
For the square truncation mounting greg used (i.e., a square baffle plate), the four triangles which border the square have side lengths 1, √3/2, √3/2. Their narrower angle is the same as that of the rhombus (70.5311 degrees) but their smaller angles are exactly half of the rhombus wider angle so they are 54.7257 degrees.
For the bevel angles of the long sides of the triangle and the sides of the square they join to, I suggest the following. Leave the triangle's long side unbeveled, and bevel all four of the square sides at 45 degrees. This makes the bevel widths unequal (the square bevel is a little wider), but it also makes the square baffle plate drop neatly into the square hole bordered by the triangles like the keystone of an arch.
Personally I feel one reason that the acoustical performance of this shape has not been thoroughly explored long before I stumbled across it is because the pieces can be tricky to cut. It would be nice if the angles were something simple like 90 or 60 or 30 degrees, but this is not the case. The best results I have had to date involved using a CAD/CAM 3-axis overhead router with two different cutter heads for the 45 and 30 degree bevels. Once I programmed it, it was able to cut out the pieces already beveled, but we had to pay for a custom bit since cabinetry rarely involves 30 degree bevels. The 45 degree bevel bit was standard. Here's a picture of some different size enclosure cut with the router. I'm no cabinet maker, but with this tool we got sharp results:
http://www.matthewrkennedy.com/3encl2.jpg
For those who are interested, the angles can be derived with some basic math. This comes from the fact that the RD can be formed by drawing the diagonals of a 1x1x1 cube -- which cuts the cube into 6 inward-pointing pyramids -- and then turning it inside out as you know, the interior diagonals of a unit cube are length √3 and the diagonals of its faces are √2.
I hesitate to bore you with the details, but if you draw the major and minor diagonals of the rhombus, this slices it into 4 identical right triangles. If you look at one of these triangles, you will see that its side lengths are: √3/2, √2/2, 1/2. Since size does not affect angles, it is simpler to think of these as √3, √2, 1 for the trig solution. We plug these into an inv sin or inv cos function and we get the values of the angles -- or rather, the values for half of them, since we cut each angle in half when we sliced the rhombus into 4 triangles. I'll just use the Windows calc.exe program.
For the smaller angle: alpha = 2 x inv sin(.5/.866) = 70.5311 degrees
For the larger angle, beta = 2 x inv sin(.707/.866) = 109.4515 degrees
The bevel angle for the rhombus is 30 degrees. In other words, cut away 30 degrees leaving 60 degrees of wood so that when two pieces are joined you get the rhombic dodecahedron's 120 degree dihedral angle (angle between planes).
For the square truncation mounting greg used (i.e., a square baffle plate), the four triangles which border the square have side lengths 1, √3/2, √3/2. Their narrower angle is the same as that of the rhombus (70.5311 degrees) but their smaller angles are exactly half of the rhombus wider angle so they are 54.7257 degrees.
For the bevel angles of the long sides of the triangle and the sides of the square they join to, I suggest the following. Leave the triangle's long side unbeveled, and bevel all four of the square sides at 45 degrees. This makes the bevel widths unequal (the square bevel is a little wider), but it also makes the square baffle plate drop neatly into the square hole bordered by the triangles like the keystone of an arch.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.