Again, point-source or section-of-sphere behavior is a nice dream, but the real world tends to get in the way.
We don't have actual polar measurements of actual 15As, but we do have some pretty good simulated polars thanks to John Sheerin. These are best-case-scenarios with half-space baffles to take the edge off of mouth diffraction, but they still show a lot of lobing and cancellation, which tells us that the 15a output at these frequencies is quite far from being a point source.
Now, you can say that the sound does all emanate from within a rectangle defined by the horn mouths, but calling it point-source is IMO a gross misnomer.
I'll insert my usual caveat here: even though the 15A horns are likely rife with internal cancellations, multi-path, and mouth reflections/diffraction which all contribute to a multi-lobed response at a range of frequencies, STILL it doesn't mean they don't sound great.
We don't have actual polar measurements of actual 15As, but we do have some pretty good simulated polars thanks to John Sheerin. These are best-case-scenarios with half-space baffles to take the edge off of mouth diffraction, but they still show a lot of lobing and cancellation, which tells us that the 15a output at these frequencies is quite far from being a point source.
Now, you can say that the sound does all emanate from within a rectangle defined by the horn mouths, but calling it point-source is IMO a gross misnomer.
I'll insert my usual caveat here: even though the 15A horns are likely rife with internal cancellations, multi-path, and mouth reflections/diffraction which all contribute to a multi-lobed response at a range of frequencies, STILL it doesn't mean they don't sound great.
Last edited:
I think people are mixing the notion of "a single point source" meaning all the sound from one element, vs. an "ideal point source", meaning omnidirectional radiation for all frequencies (assuming that is actually ideal).
A full range W.E. horn would be a good example of a single point source. Moving around it would not reveal multiple source locations or interference regions.
David S.
Did you mean to say WOULD reveal multiple source locations or interference regions? Because the simulated polars clearly show interference (lobing), which points to multiple virtual "sources."
Or did I completely misunderstand you?
I was thinking more of a typical multiway, where, even with very good response on the design axis, you can usually see that it is a multiway when moving to various off axis points. Good phase allignment at one point can not be maintained at all points unless the units are coaxial or coincident.
You are referring to the messy polar response and whether that appears as if it was interference between multiple sources, even though it is a one way speaker. I'm not sure if that is the best way to characterize it, but I can see that it might model out that way.
David S.
Why is that good? Why would that be better than, for example, a line source? Or a cardioid? Or a dipole?
Hi Guys
Speaker Dave is right or sees it like I do anyway in his descriptor of it being a single point source.
Also we have two ways of seeing it, the theoretical way and the practical way.
By that I mean consider the square wave as a diagnostic signal, in theory, it requires bandwidth from DC to light, in the practical world, it takes a bandwidth about 1/10 to 10X the Square F to look perfect on an oscilloscope.
Also, the strength of a computer modeling program is only as much as it’s ability to predict what you measure form the real thing.
Until one has compared measurements to a model, one is still speculating.
For example, I use AKABAK to design tapped horns at work which is a pretty accurate program BUT one must make some sawdust and discarded prototypes before you figure out what isn’t included in the first model , eventually your adjusted models predict what you measure pretty closely.
Predicting what happens inside a horn is one of those things that need to have real measurements to compare to before putting too much stock in it.
What would be cool is to have measured polar’s of such a horn.
I would not be surprised to see the hf depart far from the simple case but in that day, that hf response was already very good and our acuity seems to be related to our ears response.
Sy asks about the ideal being a point source, there is nothing magic about it, it just happens to have a desirable property compared to some other options.
Set up a speaker indoors or better outside where you have NO room effects.
Play pink noise through one speaker at a comfortable level and move around the speaker and listen.
If you hear comb filtering, swishy swishy as you move around, you are hearing an interference pattern created by having more than one path length between the source and your ears or having more than one source (which then has more than one path to your ears).
If one took a polar measurement of that, one sees a pattern of lobes and nulls which reflect the addition and cancellation at various points in space and frequencies. If these are dense enough, the idea is you can’t hear them.
What I was trying to explain earlier was that we hear in 3d, that interference pattern is (I think) partly what lets your ears guestimate how far away the speaker is while playing a soft voice with your eyes closed.
A speaker with the same spectral response etc but not radiating that interference pattern, can have a sonic character much less associated with the depth location, instead the voice my sound like it’s behind or in front of the speaker.
This wouldn’t mattered in the WE days but in stereo it is a very important thing (I think).
When a speaker radiates “where it is” in depth, that remains when producing a mono phantom image. Instead of hearing only the mono phantom, you also have a right and left source detracting from the image faithfulness.
Bill F, a point source radiating a section of a sphere also falls into the theory vs practical world issue, one cannot have an infinitely steep side lobes but the concern about mouth diffraction is not a big inherent problem.
While not predictions but third party full spherical measurements in free space, you can see the lack of this if you open up the CLF data files for an SM-60 or SH-50 etc and examine the polar plots at various frequencies.
Above the polar window is the 3D balloon view which you can move around and examine at different angles. Note there are no lobes and nulls, you can’t tell where the crossovers are, they act like a single driver on a large point source CD horn.
CLF files and viewer here;
Danley | Technical Downloads
These are practical full range point sources that have a limited radiation angle and a point of origin only a couple inches from the rear of the cabinet.
Unlike probably 99% of “array able” loudspeakers in commercial sound, you can place two of these side by side (or one against a wall or ceiling) and not be able to hear a seam between them.
In a narrow room, that is a cool way to eliminate the side wall reflections, place them hard against the side wall, with the wall acting like the mirror image the other speaker would have produced.
Best,
Tom
Speaker Dave is right or sees it like I do anyway in his descriptor of it being a single point source.
Also we have two ways of seeing it, the theoretical way and the practical way.
By that I mean consider the square wave as a diagnostic signal, in theory, it requires bandwidth from DC to light, in the practical world, it takes a bandwidth about 1/10 to 10X the Square F to look perfect on an oscilloscope.
Also, the strength of a computer modeling program is only as much as it’s ability to predict what you measure form the real thing.
Until one has compared measurements to a model, one is still speculating.
For example, I use AKABAK to design tapped horns at work which is a pretty accurate program BUT one must make some sawdust and discarded prototypes before you figure out what isn’t included in the first model , eventually your adjusted models predict what you measure pretty closely.
Predicting what happens inside a horn is one of those things that need to have real measurements to compare to before putting too much stock in it.
What would be cool is to have measured polar’s of such a horn.
I would not be surprised to see the hf depart far from the simple case but in that day, that hf response was already very good and our acuity seems to be related to our ears response.
Sy asks about the ideal being a point source, there is nothing magic about it, it just happens to have a desirable property compared to some other options.
Set up a speaker indoors or better outside where you have NO room effects.
Play pink noise through one speaker at a comfortable level and move around the speaker and listen.
If you hear comb filtering, swishy swishy as you move around, you are hearing an interference pattern created by having more than one path length between the source and your ears or having more than one source (which then has more than one path to your ears).
If one took a polar measurement of that, one sees a pattern of lobes and nulls which reflect the addition and cancellation at various points in space and frequencies. If these are dense enough, the idea is you can’t hear them.
What I was trying to explain earlier was that we hear in 3d, that interference pattern is (I think) partly what lets your ears guestimate how far away the speaker is while playing a soft voice with your eyes closed.
A speaker with the same spectral response etc but not radiating that interference pattern, can have a sonic character much less associated with the depth location, instead the voice my sound like it’s behind or in front of the speaker.
This wouldn’t mattered in the WE days but in stereo it is a very important thing (I think).
When a speaker radiates “where it is” in depth, that remains when producing a mono phantom image. Instead of hearing only the mono phantom, you also have a right and left source detracting from the image faithfulness.
Bill F, a point source radiating a section of a sphere also falls into the theory vs practical world issue, one cannot have an infinitely steep side lobes but the concern about mouth diffraction is not a big inherent problem.
While not predictions but third party full spherical measurements in free space, you can see the lack of this if you open up the CLF data files for an SM-60 or SH-50 etc and examine the polar plots at various frequencies.
Above the polar window is the 3D balloon view which you can move around and examine at different angles. Note there are no lobes and nulls, you can’t tell where the crossovers are, they act like a single driver on a large point source CD horn.
CLF files and viewer here;
Danley | Technical Downloads
These are practical full range point sources that have a limited radiation angle and a point of origin only a couple inches from the rear of the cabinet.
Unlike probably 99% of “array able” loudspeakers in commercial sound, you can place two of these side by side (or one against a wall or ceiling) and not be able to hear a seam between them.
In a narrow room, that is a cool way to eliminate the side wall reflections, place them hard against the side wall, with the wall acting like the mirror image the other speaker would have produced.
Best,
Tom
But that's certainly not exclusive to point sources (or equivalent omnidirectional radiators). A cardioid or line source would have no swish-swish, and a dipole would only have a null-which could be considered and advantage in a room.
Yes, I've got a copy of it and have read it carefully, a terrific book. Unfortunately, it does not answer my question.
SY,
IMO, the discussion of which radiation characteristic is ideal is far more academic than practical, if only because so few speakers that claim to function in one of these modes actually does so over the majority of its bandwidth.
Earlier, in my lost-and-found post, I made the point that any speaker that claims to be an ideal point source up to 20kHz had better not be any bigger than a lentil, or we can know the claim is a lie.
Line sources (unless they are floor-to-ceiling) only behave as line sources down to about the frequency where their height is > 1/2 lambda.
Most open-baffle "dipoles" begin displaying hypercardioid behavior above the point where their diaphragm and/or magnet diameters become a significant fraction of lambda.
"Cardioids" become dipoles, become hypercardioids at various frequencies.
I'm not saying all speakers migrate through different modes like this, but the vast majority do.
So what we are left with is speakers that tend to morph from pattern to pattern over their bandwidth, not adhering strictly to any ideal . And yet we still sometimes hit upon designs that sound terrific. What a blessing!
IMO, the discussion of which radiation characteristic is ideal is far more academic than practical, if only because so few speakers that claim to function in one of these modes actually does so over the majority of its bandwidth.
Earlier, in my lost-and-found post, I made the point that any speaker that claims to be an ideal point source up to 20kHz had better not be any bigger than a lentil, or we can know the claim is a lie.
Line sources (unless they are floor-to-ceiling) only behave as line sources down to about the frequency where their height is > 1/2 lambda.
Most open-baffle "dipoles" begin displaying hypercardioid behavior above the point where their diaphragm and/or magnet diameters become a significant fraction of lambda.
"Cardioids" become dipoles, become hypercardioids at various frequencies.
I'm not saying all speakers migrate through different modes like this, but the vast majority do.
So what we are left with is speakers that tend to morph from pattern to pattern over their bandwidth, not adhering strictly to any ideal . And yet we still sometimes hit upon designs that sound terrific. What a blessing!
Last edited:
But as far as I have been able to figure out,
a conventional box loudspeaker is just about the worst thing you can put in your living room for sound. Unless you flush mount it into the front wall.
a conventional box loudspeaker is just about the worst thing you can put in your living room for sound. Unless you flush mount it into the front wall.
Why should that be? Certainly it's true if the speaker is a piston, but a pulsating sphere (for example) can be arbitrarily large and still act as a point source.
Granted. But I have yet to encounter any tweeter--MBL included--that did anything more than a very rough approximation of a pulsating sphere. It's quite the daunting engineering challenge.
As I mentioned earlier, the one exception I can think of is corona discharge on the spike of a plasma tweeter.
As I mentioned earlier, the one exception I can think of is corona discharge on the spike of a plasma tweeter.
Well Sy, it seems like a simple question but it actually begs a lot of different questions.
How directional? The system directivity index relates the power response to the level of the axial response. High directivity index means a lower reverberent field for a given direct field. Do you want a Bose 901 (slightly negative d.i.) or do you want ESL 63 (high positive d.i.).
Is d.i. flat? This asks whether the power response and the axial response have the same shape. Is directivity constant? Realize that using a CD horn does not mean directivity is constant, since the horn is only part of the speaker's range. Most systems with CD horns start with virtually no directivity and build their way up. Also, many horns today are termed "constant directivty" when they are only "smooth directivity". Their directivity rises across the frequency band.
Since d.i. probably isn't truly flat, then you have to decide whether you want the axial response flat or the power response flat? (or the room response flat, or the early sound flat?)
D.i. is only the generic look at directivity. It doesn't define the polar curves or the even the horizontal and vertical beamwidths. Many speakers can have the same directivity but they may send their energy around in very different patterns. This is important since horizontal reflections and vertical reflections are perceived differently.
What we know from various studies is this: Flat power response is undesirable and always sounds too bright. Rising axial response to achieve flat power response will not sound good. Many speakers have falling power response and that seems to be the best approach. That would tend to disqualify the omnidirectional point source. Power response shape isn't totally unimportant, but it seems to be much less important than axial response. Dips in the power response are generally inaudible.
Very low directivity (an omni radiating point source) will lead to a diffuse sound and less of a sense of the speaker's presence. The reverberent to direct field ratio will be higher and your sense of being surrounded with sound in your room will increase. High directivity (d.i. of 10 or more) will lead to a greater sense of the speaker's presence as the source of sound. It will lead to greater "clarity" but less a sense of envelopment in a real listening space.
My personal opinion is that an ideal omni point source goes too far in the direction of wide dispersion. I also believe that people who gravitate towards horns tend to like the presence and clarity that comes with higher directivity.
David
How directional? The system directivity index relates the power response to the level of the axial response. High directivity index means a lower reverberent field for a given direct field. Do you want a Bose 901 (slightly negative d.i.) or do you want ESL 63 (high positive d.i.).
Is d.i. flat? This asks whether the power response and the axial response have the same shape. Is directivity constant? Realize that using a CD horn does not mean directivity is constant, since the horn is only part of the speaker's range. Most systems with CD horns start with virtually no directivity and build their way up. Also, many horns today are termed "constant directivty" when they are only "smooth directivity". Their directivity rises across the frequency band.
Since d.i. probably isn't truly flat, then you have to decide whether you want the axial response flat or the power response flat? (or the room response flat, or the early sound flat?)
D.i. is only the generic look at directivity. It doesn't define the polar curves or the even the horizontal and vertical beamwidths. Many speakers can have the same directivity but they may send their energy around in very different patterns. This is important since horizontal reflections and vertical reflections are perceived differently.
What we know from various studies is this: Flat power response is undesirable and always sounds too bright. Rising axial response to achieve flat power response will not sound good. Many speakers have falling power response and that seems to be the best approach. That would tend to disqualify the omnidirectional point source. Power response shape isn't totally unimportant, but it seems to be much less important than axial response. Dips in the power response are generally inaudible.
Very low directivity (an omni radiating point source) will lead to a diffuse sound and less of a sense of the speaker's presence. The reverberent to direct field ratio will be higher and your sense of being surrounded with sound in your room will increase. High directivity (d.i. of 10 or more) will lead to a greater sense of the speaker's presence as the source of sound. It will lead to greater "clarity" but less a sense of envelopment in a real listening space.
My personal opinion is that an ideal omni point source goes too far in the direction of wide dispersion. I also believe that people who gravitate towards horns tend to like the presence and clarity that comes with higher directivity.
David
Last edited:
Don't worry, I have no illusion that it's an easy question or that there's just one answer. My current speakers have asymmetric dispersion and controlled lobe angles (can you say "Ken Kantor"?). The ones before that were line source dipoles. The ones before that were designed to Toole's criteria. They're all different, but no way I can say which one's "better." They're all wrong and all different. The point (no joking intended) is that the dispersion target is axiomatic- the engineering is judged by how closely the speaker comes to that target, not whether the target is "right" or not.
Nice summary of the issues, Dave.
Of course, the room will have a huge impact on what ratio of direct to reverberant will sound best, and how important it will be to have a flat power response.
Room theory is a whole 'nuther can o' worms -- RFZ vs. Moulton vs....
Of course, the room will have a huge impact on what ratio of direct to reverberant will sound best, and how important it will be to have a flat power response.
Room theory is a whole 'nuther can o' worms -- RFZ vs. Moulton vs....