POST #21
some practical considerations
How do we "realize" an infinite vertical line source? I'm sure many are aware of the "image theory" of reflections : to first-order, a reflecting surface can be replaced by a "virtual source" behind (or above, or below) the surface, and the same sound field will result. So, one very practical way to create an essentially "infinite" line is to build a floor-to-ceiling line, and allow a reflecting floor and reflecting ceiling to "extend" the line.
How do we build a floor-to-ceiling line source? Most commonly, of course, from a vertical array of small-ish drivers. The CTC (center-to-center) spacing rules have been well-developed elsewhere ... anyone may please feel free to add some info to this thread!
What about wall reflections? Another big topic, to be sure ... but, following the same "image theory" of reflections, it's pretty tempting to consider putting floor-to-ceiling line arrays right in the forward corners of the room, to virtually eliminate the 'earliest' wall reflections. Certainly, some wall treatment would be in order for the first reflection of the left speaker from the right wall, and vice versa. But please remember that this whole plan is certainly not screaming for floor or ceiling treatments ...
And this pretty much describes something that i would personally consider to be a very interesting "experiment" :
- floor-to-ceiling line sources, made of an array of smallish drivers
- uniformly driven (or excited), but we'll need some good EQ (probably FIR) for that ~3dB per octave issue
- place a stereo pair right in the forward corners of a rectangular-ish room
- the only room treatments to consider, at first, may be wall treatment on the left wall for first reflection from right speaker (and vice versa)
- I might just be bold enough to suggest that we can get away with no center channel, for movies ... the "virtual center" approach may be good enough, because of the wider "sweet spot" from only -3dB per doubling-distance
I haven't built what i've described ... but i bet others have (or something close). So ... time for everyone to share thoughts, opinions, and experiences!
My work here is done
some practical considerations
How do we "realize" an infinite vertical line source? I'm sure many are aware of the "image theory" of reflections : to first-order, a reflecting surface can be replaced by a "virtual source" behind (or above, or below) the surface, and the same sound field will result. So, one very practical way to create an essentially "infinite" line is to build a floor-to-ceiling line, and allow a reflecting floor and reflecting ceiling to "extend" the line.
How do we build a floor-to-ceiling line source? Most commonly, of course, from a vertical array of small-ish drivers. The CTC (center-to-center) spacing rules have been well-developed elsewhere ... anyone may please feel free to add some info to this thread!
What about wall reflections? Another big topic, to be sure ... but, following the same "image theory" of reflections, it's pretty tempting to consider putting floor-to-ceiling line arrays right in the forward corners of the room, to virtually eliminate the 'earliest' wall reflections. Certainly, some wall treatment would be in order for the first reflection of the left speaker from the right wall, and vice versa. But please remember that this whole plan is certainly not screaming for floor or ceiling treatments ...
And this pretty much describes something that i would personally consider to be a very interesting "experiment" :
- floor-to-ceiling line sources, made of an array of smallish drivers
- uniformly driven (or excited), but we'll need some good EQ (probably FIR) for that ~3dB per octave issue
- place a stereo pair right in the forward corners of a rectangular-ish room
- the only room treatments to consider, at first, may be wall treatment on the left wall for first reflection from right speaker (and vice versa)
- I might just be bold enough to suggest that we can get away with no center channel, for movies ... the "virtual center" approach may be good enough, because of the wider "sweet spot" from only -3dB per doubling-distance
I haven't built what i've described ... but i bet others have (or something close). So ... time for everyone to share thoughts, opinions, and experiences!
My work here is done
Last edited:
Thank you, one and all, for indulging me and participating !!!! i hope i've added something of value, to this wonderful community
Finally, I'd like to acknowledge and thank James Griffin, for his paper "Design Guidelines for Practical Near Field Line Arrays" (2003). His paper was my inspiration for beginning a study, in earnest, of line arrays.
Finally, I'd like to acknowledge and thank James Griffin, for his paper "Design Guidelines for Practical Near Field Line Arrays" (2003). His paper was my inspiration for beginning a study, in earnest, of line arrays.
Last edited:
Jim Griffins paper is the most commonly cited: http://www.audioroundtable.com/misc/nflawp.pdf
But a much earlier (1964) paper by Paul Taylor also has relevance (and some very telling diagrams):
http://p10hifi.net/TLS/downloads/taylor-line-array.pdf
dave
......
The comment about having to use EQ I find interesting.
I'm enjoying my 12 Pluvia 7's per side crossed at 150Hz to a currently experimental subwoofer set. No EQ. But the array is slightly curved at towards the top...
Thank you for this thread! It has opened my ears to the invisible/reflective drivers. 1100ft/sec is almost liquid.
The comment about having to use EQ I find interesting.
I'm enjoying my 12 Pluvia 7's per side crossed at 150Hz to a currently experimental subwoofer set. No EQ. But the array is slightly curved at towards the top...
Thank you for this thread! It has opened my ears to the invisible/reflective drivers. 1100ft/sec is almost liquid.
First of all, thanks for this thread. It has been and will continue to be lots of fun. I hope my comments aren't taken as discouraging or argumentative. It's just that I don't have much to say about all the stuff I agree with.
In a room with a reflective floor and ceiling, a point source could be considered an array that's just as tall as the essentially "infinite" line approximated by the floor-to-ceiling array. The only difference is that the array consisting of one point source and its virtual sources below the floor and above the ceiling is sparsely populated. It would be interesting to consider the implications of this on the losses for distance.
In a practical sense, the losses across the first several feet don't matter. It's the losses beyond the closest listening position that matter. My intuition tells me that the difference between the sparsely populated array (from the point source and its virtual sources) and the abundantly populated array (from the floor-to-ceiling array and its virtual sources) aren't all that different. Hopefully someone has already studied this, and the answer can be shared.
This is interesting. My belief is that recreating an even more believable sound scape, manipulating a whole vertical cross-section of a room might be needed. This would ultimately mean recording a cross-section of the venue space in say 10x10 places (evenly spred over the area) and replay them in 10x10 places - if not with the same inter-distance, still with the same relations. I would think that a line-source comes closer to this than "point-source" does and as such, it may ultimately stand a greater chance of better reproduction. But my bet is that it is not enough with 2 point ones - that will be proven one day I'm sure. Line array mic recording replayed over line array speakers... intriguing - but maybe no signal-processing - just multi channels...
Sorry for somewhat OT but surely affiliated...
//
"What about wall reflections? Another big topic, to be sure ... but, following the same "image theory" of reflections, it's pretty tempting to consider putting floor-to-ceiling line arrays right in the forward corners of the room, to virtually eliminate the 'earliest' wall reflections. Certainly, some wall treatment would be in order for the first reflection of the left speaker from the right wall, and vice versa. But please remember that this whole plan is certainly not screaming for floor or ceiling treatments ..."
We have this by John L. Murphy
The Murphy Corner-Line-Array Home Page
Here the measured room response after dsp-eq
We have this by John L. Murphy
The Murphy Corner-Line-Array Home Page
Here the measured room response after dsp-eq
Last edited:
My work here is done
Thanks for the detailed exposition.
I probably need to reread it because I can't reconcile it with my mental picture of the physics.
Of course, that doesn't mean much, visualizations and physical intuition are notoriously unreliable.
But why doesn't the infinite line source with a Dirac pulse simply produce a cylindrical pulse wave-front?
Since a point source with a Dirac pulse produces a spherical pulse wave-front, I would expect a similar behaviour.
Similarly for an infinite plane source.
0, 1 and 2 dimensional sources and three natural co-ordinate systems, spherical, cylindrical and Cartesian.
What is your analysis result for an infinite plane source?
Best wishes
David
Maybe because of multiple source interference and distance? If so, perhaps as frequency goes up, waves from nearby point sources are more likely to arrive at a listening point out of phase. Also, source points farther away contribute less to SPL at a listening point due to expansion of their individual wavefronts. Between the two effects, it adds up that HF at a listening point is rolled off.
The problem is I can't reconcile this with the Huygens construction of the wave-front from wavelets.
Best wishes
David
Is the difference the assumption of an infinite monopole source?
Last edited:
I think you have to account for some cancellation associated with phase shifts due to distance and propagation time, at the same time as accounting 1/r^2 point source attenuation with distance related to wavefront area expansion.
Taken separately, an infinite number of points making up a line source would expand cylindrically if the points only radiated 2-dimensionally normal to the line source axis. But, they radiate spherically, so therefore may arrive at a listening point at various phases, while being attenuated with distance. Because there are an infinite number of points, instead of producing ripple in the frequency response, it produces a smooth roll off with increasing frequency.
Why do I say this? Basically, there are only a few physical things going on. A 3 dB/octave roll off has to occur for fairly simple and straightforward reasons, nothing mysterious. We know a few basic things about waves and frequencies that should be able to help develop some intuitive picture. We just have to figure out what it is.
Last edited:
It is well known in the theory of waves that the impulse response has a tail in all odd dimensions and none in the even ones. Hence there is no tail for a plane wave and it has no drop in level with distance. The cylindrical one has a tail and drops at 3 dB. The spherical one has no tail and drops at 6 dB. Gravity waves will have a tail and should drop at 9 dB.
Yes, I am. The tooling completely wore out and I was either going to have to make a huge investment or stop altogether. Being 66, I just stopped, although from a health perspective I could easily have continued. Because I had made myself a set of three NS-15s, (the last speakers that I made) I do have tooling for those, but nothing else. The NS-15's are not a popular speaker because of size and cost. Although they are clearly the best speaker that I have ever made.
There is a guy who claims that he will invest in the tooling and start to make the speakers, but as with all bright-eyed ambitious young men, things are seldom as easy as they appear at first glance. So, the outlook for this is unclear.
Based on the images and a quick read of that link, that page/product by John Murphy is exactly what i was imagining !!!!
Thank you !!
Dave, the difference is this: the continuous line source contains many, many MANY little "dz" elements or monopoles, all radiating an impulse at the same time ... BUT any single listening or measuring position will receive those impulses at different times, creating an impulse response that's "smeared" or "dispersed". With a single radiating impulse, there's only one radiating monopole for any listening position to hear.
So the point source radiates an "impulsive sphere", while the line source radiates a "dispersive cylinder".
And please keep in mind, in each case that impulse response "form" directly dictates what the frequency response will look like : "flat" for the point source, and dropping at 3dB per octave for the infinite line source.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.