Good home stereo systems
sound great if you're sitting directly between them in the direction that
the speakers are facing. The sounds paths are nearly the same in each ear
for each speakers and all of the sound is reaching you in phase. You can
almost see the musical instruments in the orchestra and where they belong.
But what if you are not sitting in the "sweet-spot" where all the phases
line up and the frequency patterns are optimal? What if two or three
people want are trying to listen simultaneously? Only one person can be
in the optimal place at a time. What if it's a wedding reception and
people are walking around?
The whole key is the
"stereo image." When all of the sounds reach you ears in
phase at the correct amplitudes on a good recording, you can almost see
every instrument as if it were in real 3-dimensional space. But when you
are moving around, the distance from you ears to the speaker is changing.
This would be fine if the frequency content didn't change with your
position, but this dispersion pattern can change drastically with the
angle you ear makes to the axis of a regular tweeter (figure at left).
Putting two or more tweeters close together only makes the problem worse:
Now the sound at any point is impossible to predict. A solution can come
in several ways: A Line-source speaker, a omni-polar source speaker, and a
point-source speaker are a few to consider. Each one has distinct
benefits and drawbacks.
Line-Source
speakers[1]:
Usually a ribbon or magnetoplanar-type speaker that uses a long (up to 7
foot), very thin element which vibrates by magnetic attraction and
repulsion in a magnetic field. The magnets that create this field must be
strong and are quite expensive. The element must be very precisely
machined and positioned and can be very easily damaged. The since the
element is a somwhat resonant metal, it can add unwanted colorations to
the sound. Another problem is that the magnetic housing causes near-field
sound reflections, which can cause unpredictable constructive and
destructive interference at certain frequencies. These speakers
traditionally rediate sound as a directional vertical bar, which helps the
problem of varying your ear level vertically, but does not significantly
increase the optimal listening area. These speakers cannot be used
horizontally because the distinct left and right channels would not
originate from one direction, but rather a range of directions. The
frequency response for these type of speakers is generally midrange and
higher.
Omni-Polar Speakers: An
Omni-polar speaker consists of a siamese tractrix (or two
"pseudo-spheres") with a horn speaker at the top and bottom. The
horn-speakers emit sound toward each other. As the sound approaches the
center of the siamese tractrix pair, it is gradually bent away from the
axis until it is perpendicular to the axis. This redirected sound creates
a disc dispersion pattern that radiates 360 degrees radially. This type
of pattern is great unless your ears are changing in distance from the
ground level, i.e., if you are sitting and standing. The size of such
siamese tractrices are around 6 feet tall and would be ideally quite
heavy. One advantage is that midrange-tweeter horns can be used if the
tractrix is large enough. This type of speaker solves only the horizontal
movement problem and does not address the vertical problem.
Hemisphereical Point-sources
can be created by beaming ultrasonic frequencies at a hard, flat surface.
This type of speaker, termed the HyperSonic[2] sound source makes use of high frequency beat
effects. To create a 10kHz tone, it uses a 200kHz tone and a 210kHz tone
together. One disadvantage is that this type of speaker still must move a
large amount of air to create low frequencies, which cannot happen
using small ultrasonic transducers. They also require special driving
circuitry to create the beat frequencies.
True Point-Source speakers
have been the goal of audio engineers for decades, but a feasible,
practical solution is very difficult to even dream up. A true
point-source radiates sound 360 degrees spherically.
this means that, no matter where you sit or stand, the frequency content
and the phase is the same on each sphere-slice. With two or more
point-sources, the stereophonic effect is even more amazing: No matter
where you stand, you get a very clear stereophonic image, undistorted in
phase or amplitude. Even the sound you don't directly hear, that is, the
sound directed away from you, can get reflected, depending on the room's
acoustics, and add to the stereophonic effect by providing depth and
width, like an orchestra in a large hall. Dr. Amar Bose used this concept
in his highly-accaimed Bose 901's, which use eight rear-firing speakers
and one foward-firing speaker, effectively mimicing the 88% reflected
sound and 12% direct sound of a typical orchestra hall.
The only feasible true point
source is the ion speaker, which uses a high-voltage ion flame to produce
the sound. The sound is limited in frequency output to the size of the
flame. For a 1cm flame, the frequency response is approximately 4kHz to
100kHz[3]. The problem is lowering
the frequency response in order to get a more full-range sound. Before
the flame can be significantly enlarged, however, the existing plasma
flame creation methodologies must be made more efficient. Existing ones
dissipate too much heat to be widely practical and require too much
power.
-
[1]http://audiocircuit.com. This site contains a very
large source of infomation on all kind of speakers from DIYer's to
production models. It contains many fine examples of electrostatic
speakers, horn speakers, magnetoplanar.
-
[2]http://www.atcsd.com/HTML/newwhitepaper.html. This
site explains a HyperSonic sound transducer that creates a hemispherical
point source by beaming an ultrasonic wave at a hard flat surface. The
ultrasonic waves produce lower frequencies by making use of beat
frequencies.
-
[3]http://signup.sundial.net/~rogerr/ionovac.htm:
This site by Roger Russel has some testing plots and data for an Ion
Tweeter. http://www.cco.caltech.edu/~boyk/spectra/spectra.htm:
This site contains an argument for having frequency content above 20kHz
and all the way up to 102kHz, since most musical instruments and sounds
have signifcant content above 20kHz.