

The experimental speaker using a 4-inch Dayton Audio full-range driver (left) and the Wharfedale DX-1 Satellite two-way speaker (right).

Dispersion of the Wharfedale DX-1 Satellite at 1 kHz (red trace), 2 kHz (green trace), 4 kHz (yellow trace), 6 kHz (orange trace), and 8 kHz (blue trace). Dispersion is similarly broad at all frequencies.

Dispersion of the 4-inch full-range speaker at 1 kHz (red trace), 2 kHz (green trace), 4 kHz (yellow trace), 6 kHz (orange trace), and 8 kHz (blue trace). At 8 kHz, the speaker produces a focused beam of sound.

Frequency response of the Klipsch G-28 Gallery speaker on-axis (blue trace); at 45 degrees off-axis with the speaker placed vertically (green trace); and at 45 degrees off-axis with the speaker placed horizontally (red trace).
The desire for broad dispersion is the main reason most home speakers are two-way designs (with separate woofer and tweeter) or three-way designs (with separate woofer, midrange, and tweeter). By switching to a smaller driver above a certain frequency, a speaker can maintain broader dispersion at high frequencies and more consistent dispersion from the midrange through the treble.
To demonstrate what a two-way design does for dispersion, I measured the frequency response of two speakers at horizontal positions from -90° to +90° in 15° increments. The first speaker is the Wharfedale DX-1 Satellite, a compact model with a 3-inch woofer and a 0.75-inch tweeter. The second speaker is an experimental model I made using a 4-inch, more-or-less full-range Dayton Audio driver. (You can see both speakers pictured in Figure 1.)
Figure 2 is a polar plot that shows the dispersion of the DX-1 satellite at five frequencies, from 1 to 8 kHz. The dispersion is broad — in fact, almost the same — at all frequencies.
Figure 3 shows the dispersion of the 4-inch full-range speaker at the same five frequencies. Even at 2 kHz, it starts to narrow a little. By 6 kHz, it’s quite narrow, and at 8 kHz it’s like a sonic flashlight, sending out a focused beam of sound.
A speaker’s crossover can have a huge effect on dispersion. Consider a two-way speaker with a 6-inch woofer and a 1-inch tweeter. Using the formula from above, we can calculate that the woofer will get beamy above about 2.3 kHz. If the crossover is at 3 kHz, the dispersion will narrow between 2.3 and 3 kHz. Above 3 kHz, the tweeter takes over and the dispersion broadens again.
The solution here might seem simple: Move the crossover point down to about 2 kHz. Yet by moving the crossover point down, the engineer puts more low-frequency energy into the tweeter and runs the risk of higher distortion at best and blowing out the tweeter at worst. The problem is even more acute in speakers that use first-order (6 dB/octave) crossovers, because those crossovers allow more high-frequency energy to pass to the woofer (thus reducing dispersion) and more low-frequency energy to pass to the tweeter (thus increasing distortion). The late Jim Thiel proved it’s possible to design a speaker with a first-order crossover that has broad dispersion and low distortion, but it took him years of hard work and some pretty radical driver designs to achieve it.
The greater the disparity in size between the woofer and tweeter, the more the speaker engineer is forced to sacrifice dispersion or power handling. A speaker that has an 8-inch woofer and a 1-inch tweeter has to have a crossover point around 1.7 kHz or lower in order not to have a dispersion problem, and few tweeters can handle a crossover point that low without distorting. This is why many speakers with 8-inch woofers (and almost all speakers with 10-inch or larger woofers) add a midrange driver.
Dispersion problems aren’t caused only by putting too high a frequency into too large a driver. They can also be caused by interference between two drivers. If your ear is closer to one driver than to the other, the two drivers’ waves will be at least somewhat out of phase. As a result, they’ll reinforce each other at some frequencies and cancel each other at some frequencies. The further apart the drivers are, the more extreme the interference effects will be.
If the drivers are positioned one above the other, the effects aren’t so bad because your ears will probably be roughly at the level of the tweeter and about the same distance from each woofer. But if the drivers are positioned side-to-side, as in a typical center speaker, the effects can be troublesome for the same reason presented a few paragraphs back: The listener could be sitting at various places in the horizontal plane relative to the speaker, and could thus experience different frequency response depending on where she’s sitting.
To demonstrate this effect, I measured the Klipsch Gallery G-28 LCR (left/center/right) speaker on-axis, then at 45° off-axis with the speaker positioned horizontally and vertically. You can see the results in Figure 4. With the speaker standing vertically (green trace), the response at 45° off-axis is pretty smooth up to about 13 kHz. But with the speaker placed horizontally (red trace), a deep dip appears between 1 and 2 kHz at 45° off-axis. Careful crossover design and close positioning of the woofers helps minimize this effect. But the best solution, if space and cost allow, is to use a midrange and tweeter positioned vertically between two horizontally placed woofers, as in the Paradigm Signature C1 and many other high-end center speakers.
The same interference problem can occur between a woofer and a tweeter, because frequencies in the crossover range are reproduced by both drivers. In this case, the interference effects can be minimized by placing the drivers closer together and/or by using steeper crossover slopes such as 3rd-order (18 dB/octave) or 4th-order (24 dB/octave).
All speaker designers are aware of the phenomena I’ve described above, but surprisingly, some don’t make broad dispersion a priority. I still occasionally encounter a speaker with, say, a 7-inch woofer crossed over to a 1-inch tweeter at 3.5 kHz. Not long ago, I heard a prototype soundbar with 3.5-inch woofers crossed over to the tweeter at 6.5 kHz. When I played James Taylor’s Live at the Beacon Theatre DVD through it, it sounded like James was singing through a cardboard toilet-paper tube! But there’s a happy ending: The engineers later moved the crossover point down to 3.5 kHz and now it’s one of the best-sounding ’bars I’ve ever heard.
Brent Butterworth and Geoff Morrison combine their years of gear testing and knowledge in one überblog of irreverence and techiness.










Copyright © 2013 Bonnier Corp. All rights reserved. Reproduction in whole or in part without permission is prohibited.


Interesting article, but then why do my magnepans (or martin logan's I've heard) sound so good?
@funboy: Panel speakers merit a whole article on their own. Their dispersion patterns are completely different from conventional speakers. A typical panel speaker radiates sound evenly front and back, but has a null (i.e., no sound radiation at all) at +90 and -90 degrees off-axis. Because the panels are so tall, they're beamy vertically - but it's a very big beam so that's not a problem, and it's actually something of an advantage because there's less reflection from the floor and ceiling.
There are variations - for example, MartinLogan uses curved panels, and on-wall panel speakers tend to have baffles that redirect the sound radiation off the back of the panel.
They sound good in large part because they're radiating sound against the wall behind them as well as directly toward you, thus producing a higher ratio of reflected to direct sound. This tends to give you a bigger soundstage and a more enveloping sound, but typically with less-focused stereo imaging.
Comment & question:
EPI 100: I have a set of these. 8" woofer crossed over by a capacitor (alone) at about 1.8 kHz to a 1" inverted-dome tweeter in a sealed box. Pretty much verifies your dispersion formula - no serious change in sound quality out to 45 deg or so off-axis, and after that it's background music reflecting off the walls anyway so changes don't matter. Sound is awesome and has very little distortion; bass is solid down to 50 hz or below and highs are clean well beyond my (older) hearing. Power handling despite the crude crossover is good - rated 60w - but any amp over 30w is ample. Wish EPI was still around - I want another pair when I go surround, with a sub crossing over around 60-80.
Whizzer cone/other tweaks in a single radiator: I used to have a pair of tiny Goodmans speakers -- 6" with an aluminum dome in the center. They sounded amazing for something that small, and needed very little power to drive (10wpc amp was ample). Yes, the very high treble was rolled off and beamed straight forward, but below 10-12K it didn't seem bad and in a dorm room the beaming didn't matter - bounced off the other wall and splattered rather than dispersing. My mom has them now and loves them. So how do those little tweaks (center dome, whizzer cone) help, or do they just make you *think* you're getting some treble without a real tweeter?
@mook: I can't say with great certainty how much those tweaks help, but I can say that every full-range driver I've measured had really lousy dispersion above 6 to 8 kHz, and extremely uneven frequency response above that, too. That said, they sometimes sound pretty good. I have yet to be swayed by a full-range driver larger than about 4 inches, but I'm keeping an open mind about it.
I've heard some good EPIs in the past, too.
Thanks for this "fundamentals" article. I have an appreciation of great sounding gear without an understanding of how and why it works. This article is a starting point for my education.
My understanding is that the reflections from wide dispersion speakers have a detrimental effect on imaging due to the sound from the reflections arriving at your ear later than the direct on axis sound.
Narrower dispersion speakers don't have the room boundry/reflection issue hence the imaging/sound quality is better.
It seems the best speaker design would be to have a narrow dispersion and good off axis frequency response
@mikeba316: The room reflections created by broader dispersion speakers improve the perceived sound quality. That's why there are few, if any, speakers available on the consumer market that show a conscious effort to narrow the dispersion. This principle has been established through decades of research at Canada's NRC, continued at Harman International, and outlined in depth in Dr. Floyd Toole's book "Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers in Rooms." Strongly recommended reading.
this kind of article is dangerous. it is at least misleading and claim to offer 'fundamental' insights when it is in fact based on compromised theories.
the whole 'stereo everywhere' school of thought will get you nothing but mediocre sound everywhere, and with wide dispersion, you are just listening more to your room than what is on the recordings.
for a counter perspective, take a look at:
http://www.sanderssoundsystems.com/technical-white-papers/dispersion-wp
http://www.sanderssoundsystems.com/technical-white-papers/room-acoustics...
@pureo: "dangerous"? That's a strong statement. My article is based heavily on research done at Canada NRC and Harman (widely acknowledged as two of the best and most important audio research facilities in the world), on information taken from some of the best-known speaker design books, and on the general practices of the most recognized and accomplished speaker designers on the planet. Your articles seem to be based primarily on your own anecdotal experiences and pet theories.