Also, as related to the Sweet 16 project, would it be fun to try and wire those drivers with all of them getting the lows but only some getting the mids and highs to shape the waveform; sorta ala Tekton? Maybe time to revisit with DSP crossovers that weren't available back then...
Multiple sources mounted as an equidistant array - and i guess that is what you had - will not produce
(any of) the desired qualities mentioned.
You will not even get a "smooth" sound power curve. One key element is to use a configuration that has
low autocorrelation.
These classes of arrays e.g. may be used as a "starting point" in design:
1 dimension
https://en.wikipedia.org/wiki/Golomb_ruler
2 dimensions
https://en.wikipedia.org/wiki/Costas_array
Last edited:
@sjhorton1184 (you wrote)
"Had a thought about the whole stereo image above 3khz thing; ..."
(>3Khz central phantom image disappearing)
As far as i understood the Bennet et al. Paper linked above (Figure 5), it is really about the "physical
ability" to (closely) produce a "plane wave" at zero angle in the region of the head.
The Authors work with "residual components" that come from the sides and are detrimental to that
plane wave component. That detrimental components rise in relation to the plane wave component
with wavelengths getting shorter.
So i took that more as a physical limitation and upper frequency limit, that is kind of "inherent" to
stereo, and there is not so much psychocaoustics involved here?
"Had a thought about the whole stereo image above 3khz thing; ..."
(>3Khz central phantom image disappearing)
As far as i understood the Bennet et al. Paper linked above (Figure 5), it is really about the "physical
ability" to (closely) produce a "plane wave" at zero angle in the region of the head.
The Authors work with "residual components" that come from the sides and are detrimental to that
plane wave component. That detrimental components rise in relation to the plane wave component
with wavelengths getting shorter.
So i took that more as a physical limitation and upper frequency limit, that is kind of "inherent" to
stereo, and there is not so much psychocaoustics involved here?
Doesn't cover frequency issues, but from The Master Handbook of Acoustics, by Everest:
"the ear and brain have the remarkable ability to gather all reflections arriving within about 50 msec after the direct sound and combine (integrate) them to give the impression that all this sound is from the direction of the original source, even though reflections from other directions are involved"
. . .
"A number of researchers had previously found that very short delays (less than 1 msec) were involved in our discerning the direction of a source by slightly different times of arrival at our two ears. Delays great than this do not affect our directional sense."
. . .
"Haas found that in the 5 to 35 msec delay range the sound from the delayed loudspeaker has to be increased more than 10 dB over the direct before it sounds like an echo. . . . these early reflections increase the loudness of the sound, and as Haas has said, result in '. . . a pleasant modification of the sound impression in the sense of broadening of the primary sound source'"
. . .
"In the reproduction of stereo sound, both intensity (level) and phase (time) are invovled. . . . If the level of one loudspeaker is increased over that of the other, the perceived location tends to migrate toward the stronger. If the two loudspeakers radiate identical signals of the same level, the signal of one is delayed slightly, the perceived location migrates toward the earlier sound. This is the law of the first wavefront"
-----------------
And a few additional comments from me: some earlier posts in this thread give blanket references to other sources without direct quotes or images that would be helpful to quickly make a point. I'm way too lazy to follow a bunch of links to other sites and read full papers to try to see if a point is even a correct interpretation of what is presented. Link rot also makes a lot of this stuff unreachable down the road.
@LineArray I would have enjoyed to discuss further, but I hate how you're citing my comments in the wrong context. I am out of here. Bye, bye!
@mattstat (you wrote)
"And a few additional comments from me: some earlier posts in this thread give blanket references to other sources without direct quotes or images that would be helpful to quickly make a point. I'm way too lazy to follow a bunch of links to other sites and read full papers to try to see if a point is even a correct interpretation of what is presented. Link rot also makes a lot of this stuff unreachable down the road."
@mattstat
What you wrote is understandable: So i will upload a few pictures here directly.
These pictures show an example baffle and driver configuration and a "quick and dirty" simulation
of frequency responses "on axis" and under angles (summed over 4 points for each range of angles).
The quite minimalistic Tool used to estimate baffle diffraction is "The Edge":
https://www.tolvan.com/index.php?page=/edge/edge.php
Diagrams and summations were done in Excel. There are some tools out there, which can do "all
the job" in a more convenient way, i am aware of that.
Example is for a small test baffle with 3 partly "decorrelating" drivers fed with same signal.
Schematic of example baffle and driver arrangement:
Frequency response on axis (far field, distance 2m) and under angles
X axis shows just Excel line numbers: "60" is about 600Hz, "100" is about 6000Hz (for orientation)
Y axis is in decibel
Remarks:
Sum over all angles and inroom response estimate is IMO "quite smooth" (although not "flat"),
despite the fact that very few points in space were taken into account. Curves tend to be much
smoother, when spatial averaging uses more points in space.
I have constructed comparable/similar arrangements "in real" since about 10 years now and
measured them (direct sound, inroom response, sound power estimate in a room having quite
constant reverberation time over frequency) and there is quite good agreement. There are no
surprises happening in the "overall behaviour" here, especially when using small drivers, that
act "pistonic" in the relevant frequency range and using a cabinet that is non resonant.
"And a few additional comments from me: some earlier posts in this thread give blanket references to other sources without direct quotes or images that would be helpful to quickly make a point. I'm way too lazy to follow a bunch of links to other sites and read full papers to try to see if a point is even a correct interpretation of what is presented. Link rot also makes a lot of this stuff unreachable down the road."
@mattstat
What you wrote is understandable: So i will upload a few pictures here directly.
These pictures show an example baffle and driver configuration and a "quick and dirty" simulation
of frequency responses "on axis" and under angles (summed over 4 points for each range of angles).
The quite minimalistic Tool used to estimate baffle diffraction is "The Edge":
https://www.tolvan.com/index.php?page=/edge/edge.php
Diagrams and summations were done in Excel. There are some tools out there, which can do "all
the job" in a more convenient way, i am aware of that.
Example is for a small test baffle with 3 partly "decorrelating" drivers fed with same signal.
Schematic of example baffle and driver arrangement:
Frequency response on axis (far field, distance 2m) and under angles
- in range 10 degrees (sum of 4 points "up, right, down, left")
- in range 20 degrees (sum of 4 points "up, right, down, left")
- range 45 degrees (sum of 4 points "up, right, down, left") and
- sum over all 12 points (gives also a hint how DI behaves over Frequency)
- estimate of an "inroom response" by summing on axis and off axis in weighted manner
X axis shows just Excel line numbers: "60" is about 600Hz, "100" is about 6000Hz (for orientation)
Y axis is in decibel
Remarks:
Sum over all angles and inroom response estimate is IMO "quite smooth" (although not "flat"),
despite the fact that very few points in space were taken into account. Curves tend to be much
smoother, when spatial averaging uses more points in space.
I have constructed comparable/similar arrangements "in real" since about 10 years now and
measured them (direct sound, inroom response, sound power estimate in a room having quite
constant reverberation time over frequency) and there is quite good agreement. There are no
surprises happening in the "overall behaviour" here, especially when using small drivers, that
act "pistonic" in the relevant frequency range and using a cabinet that is non resonant.
I don't see the logic at all of what you try to do. I think you should read the works of Dr. Earl Geddes and Dr. Floyd Toole, then you would see that what you try to do does not work at all. They did long studies about this subject, and their views lead to how we now make speakers.
Next is a slightly scaled up version, which could serve as a schematic for a 2-Way floorstanding
loudspeaker. The rectangle in the bottom right of the baffle symbolizes the woofer's preferred
position. The baffle shown would be the left loudspeaker, so the woofers are close to the floor
and shiftet "inwards".
The crossover frequency is assumed to be lower than 500Hz.
Low Frequencies (Woofer):
Next picture show a 2D visualisation (of course a "simulation" would be 3D) of how that speaker
can be placed in a room. There are 2 wavelengths shown, first with a point source radiating without
a cabinet and then a point source mounted on the symbolized cabinet.
Remarks:
The cabinet with the woofer placed as shown is meant to act like a "scattering body" (in conjunction
with the room's adjacent walls) to reduce combing in the upper bass to lower midrange due to early
reflections from the front wall and the side wall of the room's closest corner.
Due to woofer placement, in the freefield the baffle step of the cabinet itself (alone) tends to be
smooth without exibiting a large "baffle hump".
The "bottom woofer" placement (classical) avoids detrimental floor reflection and associated
"notches" or "suckouts" in the low frequency response, when the cabinet is placed on the floor
and in a room.
loudspeaker. The rectangle in the bottom right of the baffle symbolizes the woofer's preferred
position. The baffle shown would be the left loudspeaker, so the woofers are close to the floor
and shiftet "inwards".
The crossover frequency is assumed to be lower than 500Hz.
Low Frequencies (Woofer):
Next picture show a 2D visualisation (of course a "simulation" would be 3D) of how that speaker
can be placed in a room. There are 2 wavelengths shown, first with a point source radiating without
a cabinet and then a point source mounted on the symbolized cabinet.
Remarks:
The cabinet with the woofer placed as shown is meant to act like a "scattering body" (in conjunction
with the room's adjacent walls) to reduce combing in the upper bass to lower midrange due to early
reflections from the front wall and the side wall of the room's closest corner.
Due to woofer placement, in the freefield the baffle step of the cabinet itself (alone) tends to be
smooth without exibiting a large "baffle hump".
The "bottom woofer" placement (classical) avoids detrimental floor reflection and associated
"notches" or "suckouts" in the low frequency response, when the cabinet is placed on the floor
and in a room.
@waxx (you wrote)
"I don't see the logic at all of what you try to do."
I truly believe that and i am fine with that. But let me state this in a friendly and humble manner:
I am not just trying to do something, i am truly doing something (since many years now).
Even in case i am the only person on the planet - which i am not - realizing what the background
is and what the goals may be, i am also fine with that.
Remark:
I regard any further comments just insisting on how wrong, superflouus, technically incorrect, ... etc.
the content provided in this thread is as being superflouus. This is simply because such one-sided
and highly predictable comments are boring me (to death).
This is also because i know very well (e.g.)
Members willing to just "put down" things presented here, are invited to ignore this thread. There
is no further comment necessary when doing so.
"I don't see the logic at all of what you try to do."
I truly believe that and i am fine with that. But let me state this in a friendly and humble manner:
I am not just trying to do something, i am truly doing something (since many years now).
Even in case i am the only person on the planet - which i am not - realizing what the background
is and what the goals may be, i am also fine with that.
Remark:
I regard any further comments just insisting on how wrong, superflouus, technically incorrect, ... etc.
the content provided in this thread is as being superflouus. This is simply because such one-sided
and highly predictable comments are boring me (to death).
This is also because i know very well (e.g.)
- what a waveguide is and what it is commonly used for (or what it can do and what it cannot do)
- how (multiway) loudspeakers are designed usually and what common arguments are due to that
- ...
Members willing to just "put down" things presented here, are invited to ignore this thread. There
is no further comment necessary when doing so.
Last edited:
I don't wholly understand what is being plotted and how it is related to the frequency response at a set of points in space around the listening position. A typical plot might be something like 5 x 5 set of frequency responses and an average. It needs a degree of interpretion because our 2 ears and a brain don't perceive cancellation dips in the way a microphone measures them. On the other hand, hiding information by showing only averages will put off people trying to assess the pros and cons of what is being proposed.
Talking around simulations is good although I am expecting what is being proposed to work less well than configurations that make use of the directivity of the drivers. This should of course be visisble in the simulations/measurements so long as they simulate/measure what is relevant. I presume your current simulation is only including the direct sound? Do you have access to software that includes the reflections becasue they are a signficant part of what we hear?
@andy19191 (you wrote)
"I don't wholly understand what is being plotted and how it is related to the frequency response at a set of points in space around the listening position. ..."
Let me try to make things a bit clearer.
The plots referring to sets of points in space are not "around the listening position".
- first there is a 0 degree ("0 Grad") on axis plot at 2m distance. That is what you get when measuring in the free field on axis (direct sound).
- then there is a 10 degree ("10 Grad") off-axis plot, averaging over 4 points in space, that are 10 degrees off-axis. Plotted is the relative level
when compared to the on axis response. As you can see, there is only little directivity far below 600Hz.
- same is done for the 20 degree ("20 Grad") off-axis plot and the 45 degree ("45 Grad") off-axis plot.
- the whole set of off-axis points and associated average responses from 10 to 45 degrees are used just as an estimate ("mean of the
means" if you like) for the energy radiated into the front hemisphere under a range off-axis angles.
Placing the loudspeaker in a room, the sound radiated into this "donut" of off-axis angles would mainly be responsible for the
early reflections.
So the off-axis curves serve as an estimate for what can expected in a reflective listening room (having close to constant
reverberation time over frequency, which untreated european living rooms usually have above the "modal" range).
Toole e.g. uses a similar "donut" of angles, to make up his "ERDI" (Early Reflection Directivity Index) and an associated sound
power curve for the early reflections: Such sound power and DI curves are generally obtained from spatially averaged data.
What i have done here is "a lot more sloppy" (only few points in space, only selected angles) but followed the same concepts
roughly.
"I don't wholly understand what is being plotted and how it is related to the frequency response at a set of points in space around the listening position. ..."
Let me try to make things a bit clearer.
The plots referring to sets of points in space are not "around the listening position".
- first there is a 0 degree ("0 Grad") on axis plot at 2m distance. That is what you get when measuring in the free field on axis (direct sound).
- then there is a 10 degree ("10 Grad") off-axis plot, averaging over 4 points in space, that are 10 degrees off-axis. Plotted is the relative level
when compared to the on axis response. As you can see, there is only little directivity far below 600Hz.
- same is done for the 20 degree ("20 Grad") off-axis plot and the 45 degree ("45 Grad") off-axis plot.
- the whole set of off-axis points and associated average responses from 10 to 45 degrees are used just as an estimate ("mean of the
means" if you like) for the energy radiated into the front hemisphere under a range off-axis angles.
Placing the loudspeaker in a room, the sound radiated into this "donut" of off-axis angles would mainly be responsible for the
early reflections.
So the off-axis curves serve as an estimate for what can expected in a reflective listening room (having close to constant
reverberation time over frequency, which untreated european living rooms usually have above the "modal" range).
Toole e.g. uses a similar "donut" of angles, to make up his "ERDI" (Early Reflection Directivity Index) and an associated sound
power curve for the early reflections: Such sound power and DI curves are generally obtained from spatially averaged data.
What i have done here is "a lot more sloppy" (only few points in space, only selected angles) but followed the same concepts
roughly.
Last edited:
An observation. Looking at five and ten year old posts I often find that little to none of the links to external materials referenced work. I see that the OP uses a copious amount of links in their posts. It would be a shame if this all went to waste in a few years when the owners of these other sites decide to redesign their website and change the address by a single character breaking all of it. It is pretty easy to upload files to this site or copy and paste the relevant text directly into a post. Just looking to ensure future access to great content like this.
Thanks. What you have chosen to respond to and what you have chosen to ignore seems to indicate that we are interested in different things. No problems.
I just wanted to explain, what i have posted so far (first things first ...), since that did not seem
to be clear enough up to then.
You'd like to see some "dirty data" (non averaged), right?
I will see, how i can "present" some ... (coming soon)
@olsond3 Thank you for the kind words, i will keep uploading pictures directly (since about post #47)
@Andersonix
If you are interested in the technical background or even historical loudspeaker concepts that may show similar
aspects of acoustic behaviour then I will be happy to go into it later. The "Kabuki" speaker - i can assure you - will
not be included in a historical outline.
Otherwise: "Please stop the noise and refer to Post #50". Thank you in advance.
If you are interested in the technical background or even historical loudspeaker concepts that may show similar
aspects of acoustic behaviour then I will be happy to go into it later. The "Kabuki" speaker - i can assure you - will
not be included in a historical outline.
Otherwise: "Please stop the noise and refer to Post #50". Thank you in advance.