Bang & Olufsen BeoLab 5.

Manufacturer: Bang & Olufsen America, Inc., 780 West Dundee Road, Arlington Heights, IL 60004; 847/5904900;

www.bang-olufsen.com

Price: $18,000/pr

Source: Manufacturer loan

A Variety of Goals: What kind of dispersion, or horizontal radiation pattern, should a loudspeaker have? Almost everybody now finally realizes that in order to sound superior, a speaker system must have a smooth horizontal radiation pattern, whatever the pattern is. In the great majority of designs this means its output around crossover points must be smooth and uniform--not beaming narrowly as the lower drivers go higher but then suddenly flaring in widely where the upper drivers come in. While hard to do well, this is the new (albeit old) goal. When it happens even partly, images stay steady and don't suddenly change size (mostly their width) with frequency, and spectral balances do not change with head movement.

But given directivity smoothness--uniformity in the transitions from driver to driver--what else? Narrow in the highs, for dry and focused treble sound? Or wide dispersion in the treble, for airier sound? Narrow in the treble and also continuing narrow on down below the treble, into the midrange and lower midrange? (Good luck.) How about a crude figure-8 polar pattern, as from those so-called "dipole" designs? Or what about broadband omni, that is, equi-omni, meaning the output is at the same level in all directions at all frequencies? If not that, then how about sloppy omni?

Finally, what about omni that is not quite equi-, that is, favoring one direction?

Below in Figure 1 is an ideal type of horizontal radiation pattern, graphed the same way as my measured curves for Sensible Sound. This means the curves show on-axis response at the top (perfectly flat output), followed by speaker system response to the sides, that is at -30, -60, and -90 degrees, and then moving around toward the back, that is at -135 and finally -180 degrees, which is directly behind the speaker and reflects into the room from the front wall.

[FIGURE 1 OMITTED]

This first set of curves shows the classic narrowing-dispersion-with-higher-frequency "goal." Goal is in quotes because this rolloff by angle was a design objective only since it made a virtue out of necessity. Beaming is simply how conventional drivers behave: they disperse their sound more narrowly as they go higher. They change from being like bare lightbulbs to being like flashlights to being like penlights. So manufacturers said, Let's just say in our marketing lit that this is how speakers *should* play music and voice. (Almost no one over the decades has asked, Uh, why should the highs be distributed less than everything else?)

In this ideal graph there are no lumps at any crossover stitches. Indeed, you can't tell if this is a multiway system or some single magical driver design (the KEF Uni-Q came close). Regardless, the resultant sound from such an ideal design can be very good, with very good imaging. The more rolloff the drier the sound, as shown in figure 1, and the less rolloff the more spacious the sound.

This goal is still adhered to by the majority of loudspeaker manufacturers, who still defend it as desirable. The folks at the Infinity group (Harman), who have been thoroughly and cleverly quantifying their rediscovery of this decades-old notion that smooth output at all angles (especially including smooth crossover seams) does indeed produce clearly preferable sound, point to figure 1's type of horizontal radiation pattern as optimal. "There were clear correlations between listeners' loudspeaker preferences and a set of acoustic anechoic measurements. The most preferred loudspeakers had the smoothest, flattest, and most extended frequency responses maintained uniformly off axis" is how Infinity's Sean Olive put it recently in the Audio Engineering Society Journal and elsewhere, referring to the actual, similar curves shown in figure 2.

[FIGURE 2 OMITTED]

The Infinity bundling (graphing) of horizontal outputs is different from mine, so the figure 2 two graphs are not fully comparable with the ideal of figure I or with the various actual system compromises I have displayed in T$S over the past several years. But you get the idea.

For this magazine, the better speakers I have heard and measured that come to mind were designs from NHT, API (an Athena sat/sub system, the cheapest and overall probably as good as any in this regard), Eggleston, Buggtussel, and Canton. Figure 3 shows the Athena $1 satellite's horizontal radiation pattern. This smoothly baffled 5"/1" (woofer/tweeter) design demonstrates that achieving something which approximates the figure 1 ideal is not all that difficult.

[FIGURE 3 OMITTED]

Now, what would a uniformly beamy design look like in its ideal form?

Figure 4 shows an impossible dream: narrow dispersion not just in the treble but across the audio band, all the way down to where it does not matter. As you move off axis, the output gets softer across the board, in the treble and also in the midrange down to the bass. This design would produce very dry and precise imaging, tightly focused sound. We would not necessarily notice airlessness because the frequency response of the later, reverberant field would be timbrally similar to the early-arriving sounds, even if much quieter. All of this pretty much remains a dream, however: making midrange and lower frequencies directional, with their increasingly longer wavelengths, is practically impossible. In the late 1980s, Ken Kantor's AR MGC-1 attempted to achieve (along with its many other clever effects) lower-treble directivity by sinking the midrange driver on the cabinet side and also into a deep "horn" of foam.

[FIGURE 4 OMITTED]

Other companies have tried to run two midranges in opposite polarity (out of "phase") and achieve some cancellation, with limited success. It works a bit in the treble, creating a quasi-"null," as I showed with a Polk surround satellite in T$S a few years ago. This probably could work somewhat for the midrange, too.

The Quest for Omni: But the best-sounding developments have been in the opposite direction: omni, even equi-omni, across the audio spectrum. Bose had splashed their (mostly mid) sound around, which was a groundbreaking start, but more precision was wanted. Over the decades, almost all serious manufacturers' designs used smaller and smaller tweeters, and/or baffles (the cabinet face) tended to get narrower (B&W may have started that fine development; Waveform refined it further) or smoother (Genelec is probably the master in that regard). The result has been that the relative amount of treble and upper treble distributed into the room's reverberant field has increased. "Yay!" was the response from listeners used to live acoustic music. A truly revolutionary step after the original AR dome tweeters was the sensational, unpatented, strangely unimitated Allison tweeter, which delivers as much sound sideways as it does forward.

Some manufacturers calculatingly

put a second tweeter on the back or side of the cabinet (Snell, DCM, a few others later), which obviously helps disperse the top octave-plus. Ohm's Walsh-radiator designs strived toward broadband omni output as well. Dbx's Soundfield Imaging designs used extra tweeters, and sometimes extra mids, to achieve a smooth and consistent horizontal radiation pattern--broadband, down to below 200 Hz--that was almost a full B (bel) louder (twice-loudness, that is, approximately 9 dB) toward the other stereo speaker than elsewhere. The goal was to maintain image centrality well off to the sides, off-axis. Its stereo-everywhere imaging aside, several reviewers thought the top dbx Soundfield Imaging models had the most striking, and about the best, loudspeaker sound they had ever heard. By which I mean that what became immediately clear in their reviews was that much of the raving was not, or not only, about the anchored localization of the images as listeners moved about or sat in various off-center spots. Nor was it about the extreme flatness and tonal accuracy of the frequency response. It was about the spacious, floating, cabinet-decoupled, natural quality of playback of all kinds of music. As I was involved in that project, led first by Mark Davis and later by Michael Chamness and John Buzzotta, I hardly disagreed with this praise. One eminent designer came to audition the 1A rev of the design, and before we got to any music he put on a test CD and immediately said, "Boy, listen to that pink noise." It dawned on me that similarly natural-sounding, airy, balanced playback, jaw-dropping for having a big headphonelike quality that was disconnected from the units you saw before you in the room, was possible from any really competent wideband omni or near-omni design, and sometimes even with the big panel designs and a reflective front wall.

Today we have multichannel, so the pleasures of adequate, natural treble in the reverberant field are available and known to more and more listeners. (I'm convinced that if speaker designs had done better with the highs over the years, either from including more conventional tweeters or from using Allison-type tweeters, the excitement over multichannel systems, for music anyway, would have been much less.)

Figure 5 shows what equi-omni sound would look like as typically realized. In other words it's not several flat lines one right atop the other; in the graph the sound behind the speaker (-180 degrees) is a few dB lower in level than the on-axis sound, from below 200 Hz all the way on up. I wanted to draw the performance realistically.

[FIGURE 5 OMITTED]

The special case of the Soundfield Imaging stereo-everywhere horizontal radiation pattern would look like figure 6, which is akin to figure 5 but with the listener axis no longer the loudest axis:

[FIGURE 6 OMITTED]

In the final years of dbx consumer, the late 1980s, the creation of figure 6's horizontal radiation pattern via lensed horns (instead of employing multiple conventional drivers fed by complex, elaborate, software-designed crossovers) was an area of intensive research and mockups for the next, never-realized iteration of the Soundfield Imaging line. Resonances were a chief problem. When I left that dying company and became a speaker reviewer, I took my keen interest in horizontal radiation patterns and treble dispersion issues with me. I kept my eyes and ears open for developments in the area. The dbx Soundfield Imaging line had been enough of a commercial success that there was interest in doing something similar, without violating dbx's patents, from Allison, Bose, SoundWave, Canon, and others. (The Allison IC20, with user-controllable horizontal radiation pattern including a centered-everywhere option, was by far the most successful imitation.)

Wannabe-Great Minds Think Alike, Too; Little did I know that, throughout this same time period, somebody was cleverly figuring out how to make a precise lensed horn that did indeed distribute mids and highs very widely and very smoothly. I knew who David Moulton was--everybody in audio around Boston did, as he was a teacher, consultant, and up-and-coming recording engineer who had written interesting articles on a range of topics and had a number of interesting ideas. He was a Juilliard-trained musician, and as a sometime working classical-music critic I knew how unusual that was in the industry. For my own part, I had begun preaching my dispersion gospel (along with my boundary-augmentation gospel) in a variety of audio publications. I wish I had stumbled across Moulton's earlier, mid-1980s writings where he had claimed:

A novel acoustic lens has been devised which allows the acoustic radiation field of two or more transducers to be coupled in phase such Nat the resultant acoustic radiation pattern is interference-free at all frequencies, and displays wide and nearly frequency independent beam width. The lens is utilized in the high frequency transduction system of a loudspeaker array in order to achieve high frequency beam widths which approach that of the low frequency components while eliminating beaming, upper frequency beam collapse, and combing.

[A year later] ... An acoustic transduction system has been designed which couples the sound field of two transducers in phase and produces a resultant acoustic radiation pattern which has a full 360 degree horizontal by 90+ degree vertical frequency invariant beamwidth. The system incorporates an acoustic lens, rather than a phased array or electronic technique, to couple the sound fields and produce the wide dispersion. The system is used in a loudspeaker to achieve a 360 degree sound field which is devoid of high frequency beam collapse, and combing.

If only Moulton had approached dbx (I like to think so, anyway; one never knows how such overtures turn out). Through the 1980s we were not reading his AES preprints, and he had not familiarized himself with the Soundfield Imaging experiments, designs, and implementations.

Almost a decade later, in the middle of the 1990s, Roy Allison was invited to give a paper at an NYC AES convention, and he chose to write about "Imaging and Loudspeaker Directivity: To Beam or Not To Beam." The paper covered loudspeaker imaging and tonal balance as a function of dispersion and direct/reflected treble output; it also touched on the interesting history of major tweeter designs, starting with Edgar Villchur's domes at AR and including, naturally, Allison's own. But Allison was unable to make it to New York, and asked me to deliver the paper instead. Totally psyched, of course, I did so. I found it striking during the Q&As afterward that even experienced listeners sometimes confused the distribution of treble in the reverberant field with its level. I also was psyched because scheduled next to the Allison paper was one called "The Significance of Early High-Frequency Reflections from Loudspeakers in Listening Rooms," by David Moulton. But Moulton had to bail out, alas, admitting that the paper had not quite panned out, or not yet. By now I'd heard he was investigating treble "lenses," and he had already written these intriguing words:

Through the years we have learned much about which measurements can most accurately predict the way loudspeakers will sound in domestic listening rooms. As a result, we are better able to make loudspeakers that will sound the way we want them to sound. One of the most important design elements is loudspeaker directivity. This paper discusses means for controlling directivity in loudspeakers mean t for home use, and how these design choices affect what we hear.... Work with wide-dispersion high-frequency lenses in loudspeakers has led to the study of the significance of early high-frequency reflections of loudspeaker sounds in reverberant spaces. The effect of such reflections in stereophonic playback of conventional recordings as it relates to loudspeaker design and room design will be discussed.

Audition Time: More years passed, however, and I heard nothing further about Moulton's investigation. Then one day just a few years ago I learned there was a prototype lens design at a local art gallery. Seemingly chronic problems had been solved, as Moulton described it, after he teamed with a former student, Manny LaCarrubba, who made significant advances on his own with ALT (acoustic-lens technology). Eventually Bang&Olufsen got interested. (The whole history is told in fascinating detail on the September 2004 Boston Audio Society Speaker newsletter, v26n3.) Still hopeful about Moulton's design goals and his recognition of the importance of treble throughout the reverberant field, I went to the art-gallery installation and with permission took several quick individual and system measurements with my dbx RTA-1 and instrumentation mikes. I looked at full-band frequency response both by angle and in toto. My instant conclusion from listening and measuring was that this was the real deal, at least as implemented in the prototype. The playback was unusually flat all the way round the cabinet (not quite totally "equi" in level, but close), smooth in totaled output in-room, and it sounded exactly as you'd expect: utterly natural, neutral, balanced, and fully spacious but not sloppily so.

In a matter of time the BeoLab 5 appeared on the market. The cabinet comprises a 15" downward-facing woofer, a 6.5" midrange (called the upper woofer) covering the upper bass and lower midrange, and the acoustic lens, which loads an upper mid and tweeter in two-decker-style. At an srp of around $16,000, including beefy built-in amps and steep crossovers, EQ both static and dynamic, onboard hi-rez digital converters (inputs are both analog and digital), and space-age design in both the good and the bad senses of the word, the BeoLab 5 made news almost as much for its features, appearance, and price as for anything else. The static EQ evidently takes care of any lens resonances that exist, along with any other anomalies; the dynamic EQ modifies the bass output with room position, correcting for boundary augmentation and other variables, using a built-in moving mike (the delta is small, a few inches down near the floor) with analyzing computer. Claims are made for correcting room resonances (not the same thing as boundary-augmentation ripple), but I am skeptical of how completely that matter is or can be addressed with their approach.

This all sounded like an almost overwhelming amount of ambition for a speaker design. I got in touch with Moulton and told him I wanted to listen to and measure his and his colleagues' work in its final incarnation of the BeoLab 5. He invited me to come out to his studio.

Moulton's listening and mixing room is a very large domestic space with a high vaulted ceiling. Five BeoLab 5s are deployed in the roolTl (front left / center / right, and rear left / right), as he does more and more multichannel work. I auditioned only the front stereo pair, and my room-response measurements were of those two units plus the center channel. I brought a range of extremely familiar material, most of old, some of it even pretty mediocre as to recorded sound (congested, thickly mixed, and so on). Most of it I have been using to audition loudspeakers for over 20 years, however.

To get right to the point, in every instance I heard things I had never noticed before in quite that way: inner details, lines and strands that were newly followable, everything revealed under fresh, clear lighting. I worked my way through Mozart string quartet passages, then '60s and '70s rock (and not in gussied-up new mixes, either: old Fleetwood Mac, old live Cream). From CD to CD, tonal balance seemed oddly right--neutral, accurate were the words that kept coming to mind--and since some of the rock is on the thin side timbrally, this was impressive. Occasionally I wanted to broadly turn down the treble --it was obviously creamy-smooth, and not piercing, but overall it sometimes seemed bright for my taste. I heard no lumps or missing spots in the lower midrange and bass, which is extremely rare. All was full and solid. On pink noise it sounded plenty full, maybe a bit much, so I wanted to turn it down a tad in that range, too, using a broad bass control. The fullness of the lower midrange and bass evidently balanced out that flat treble.

These are minor reservations, and easily fixed. Generally this was sound such as we seldom hear from loudspeakers, and I must repeat earlier adjectives: floating, very detailed ("complete" is my wife's apt phrase), gratifyingly spacious without any readily apparent source, and uncannily smooth. I keenly sensed that I had probably never heard playback with this quality of airiness that was also this flat and neutral. I believe the only other time I had the feeling of such a high degree of rock-solid, locked-in, dead-on accuracy was the first time I put Etymotics ER-4S headphones into my ears and played familiar CDs for hours. You also can characterize it in absences: no boxiness, no harshness; the system disappears sonically. The experience is of great delicacy.

Like so many recording engineers, Moulton listens at loud levels. Not deafening, not damaging, but somewhat louder than life (except with the rock, anyway). This is easy to do when the system is so powerful and so unstrained; the high levels do not bother your ears. He readily turned it down when I asked him to, as with chamber music. Which sounded just astonishing, altogether like "you are there," and in a nice big room like his, also like "they are right here." Wow.

As mentioned, I listened to pink noise, too, which roared as smoothly as Niagara Falls and was unchanging as I walked about. It was every so slightly shadowed behind the speakers, but note that the design goals are claimed only for the forward 180 degrees. Yet the timbral changes were not large when you went in back of the units, and on music they were unnoticeable. So if the BeoLab 5 is not precisely omni, nor claimed to be, it is closer than anything else.

I did not angle or toe the stereo units in toward the center listening positions, nor face them toward each other, but I am sure that doing either would produce balanced mono and stereo everywhere across the front. Figure 8 (about which more later) shows that there would not be much point in doing this anyway, and as it is, two BeoLab 5s do a good job maintaining mono and stereo balance with changing listening position.

[FIGURE 8 OMITTED]

Measurement Time: My measurements speak louder than any of my words. The averaged in-room response (Figure 7) speaks for itself. (5 dB / division vertically.) The 5-6-dB elevation below middle C may look excessive, but not only is it not unpleasant on music, it does serve to balance the flat and extended mid and treble. Note that 20-33 Hz is at the same level as everything above 250 Hz, an unusual sight.

[FIGURE 7 OMITTED]

The speakers are too heavy (over 130 lbs each unit) to lug outdoors readily for my customary 2-pi, anechoic-with-a-floor measurements of horizontal radiation pattern. So to produce figure 8 I improvised while staying in Moulton's listening studio/lab. Standing a meter or less away and with the dbx RTA-1 set to its forever (continuous) averaging, I moved my Earthworks mike broadly up and down and to and fro from upper lens to 6.5" driver, taking a reading of pink noise output at 0 degrees (on axis) +/-5 degrees (in other words also moving it slightly back and forth as well for each "slice). Then the same at -30, -60, -90, -135, and behind (-180 degrees). This is the procedure I have done for over 15 years outdoors in my quiet back yard, only there I am farther away, at typical listening distances (7-9') and seated ear height, which perforce blends what I was trying to capture indoors with the moving mike. I am persuaded that these results are a reasonable facsimile of my usual protocols.

[FIGURE 8 OMITTED]

Whether it is like what I have done before or not, the normalized horizontal radiation pattern is astounding. Normalized means the axis response has been made perfectly flat, and thus the graphs show the difference in output as you move off-axis. These outputs contribute to the total sound (of course) and also form the imaging. Look at the overlap. Note the similarity to the ideal of figure 5. Enough said. Oh, the few dB of ripple from 160 to 400 Hz in the rearward output (behind the cabinet) might be worth investigating and smoothing if possible, but it is probably not possible to fix it, and hardly worth fretting.

Nits: What can I critique? Only a few things:

Cost: It's awfully expensive, and the old speaker designer's quip is that for kilobucks you better be able to do a good job; it's making a great inexpensive speaker that's the real feat. $18,000 for stereo is hardly Sensible sound. But for this quality, it's not easy to say what's sensible.

Voicing: For this much money, the balancing warmth of the upper bass and lower midrange is slightly excessive and should be reduced a little. B&O always had a deserved reputation for lean bass at best, and I bet this is a reaction to that history.

EQ: It should actually be even better than it is, and it would be really cool if it and other functions could be set remotely.

Appearance: B&O says it is a hit aesthetically, too, but I have not met any non-audio person (especially female) who could happily live with it. The wife points out that it looks like Mad magazine's Spy vs Spy; forget the Space Age. Perhaps an elegant cloth / frame covering would be an option to consider, or a different lower cabinet.

That's it. In my view the BeoLab 5, with its exceptional horizontal radiation, is demonstrably the finest loudspeaker system designed and manufactured thus far.

There is already a BeoLab 3 satellite, with an srp of around $3000 / pair and its own "statement" look, which sounded superb in a brief audition at Moulton's.

A BeoLab 4 is on the way, rumored to be along the lines of a less-expensive passive version of the 5. I can't wait.

Postscript: The achievement of Moulton and LaCarrubba's lens is an extremely exciting audio development. As Moulton explains it:

An ellipse is a geometric shape that has two focal points, and all reflected paths from one focal point to the other are of equal length. The implication is that essentially all energy passing through one focal point will subsequently pass through the other, with no change in relative phase or power.... The Acoustic Lens is a complex shape consisting of ellipses in rotation, oriented so that one focal point is common to all of the ellipses in rotation while all the other focal points constitute a focal ring (a circular arc in a plane that includes all the focal points). There is a concave section of the lens as well as a convex section. The acoustical center of the driver is placed at the common focal point, which is located within the concave portion of the lens. All energy emitted from the driver that strikes the lens will re-emit in phase, through the focal ring, in a 180[degrees]arc (because the acoustic lens is designed for 180[degrees] dispersion).

[ILLUSTRATION OMITTED]