An Alternative Approach to Stereo Imaging.

I have long wanted to give a pair of Bose 901 speakers a really serious audition in my 3400-cubic-foot listening/viewing/testing room. The 901s have been demonstrated to me in assorted and often very substandard dealer demo rooms over the years (this goes all the way back to the original, in the late 1960s), and in each case I felt that the locale had the potential to color the sound more than the speakers deserved.

I once tried to get a pair to audition from Bose, but the company wanted me to jump through more corporate-arranged hoops than I cared to in order to get a pair shipped down to my place. The Bose people obviously are very sensitive about the 901s, particularly when it comes to letting any product reviewer loose with a pair. I got the impression that they first wanted to make sure that I was not a tweak-audio crank with a Bose-bashing agenda.

I am not a Bose-basher, but I can certainly understand their concern, given the way the company has been treated by some members of the high-end press. However, I was not in the mood to be treated all that much differently from typical consumers who obtain a pair and set them up in their living rooms. When I get a product to review, I prefer to fiddle with it unassisted, so that I can see just what customers will experience during the set-up and auditioning stages.

Consequently, I had just about given up on ever hearing a pair of 901s in my room, or in any other good room, until a real fan of the Bose concept, Gary Eickmeier, invited me down to his home to see what his particular setup, which includes the latest Series VI incarnations of the 901, had to offer.

Like me, Gary is a regular participant in several audio-related internet newsgroups. Unlike me, Gary has caught a lot of flack from assorted newsgroup Bose-haters, who have often vilified him for even having a set of 901s in his home, let alone praising their sound. It seems that the speaker is not particularly popular with those who frequent the audio chat groups. This appears to be the case even if some of them no doubt dislike the speaker mainly because of the monumental commercial success of the Bose Corporation (some audio buffs seem to be paranoid about big, successful companies), rather than because of their sound. Indeed, I get the impression that many of the more emphatic Bose 901 haters have either never auditioned the systems at all, or at best have only heard them in the same kinds of less-than-perfect dealer environments I have experienced.

Anyway, one weekend I packed my car with my AudioControl SA-3051 RTA, some test and music discs, and an open mind, and headed for Gary's place, 270 miles away. When I got there, Gary almost immediately showed me the speakers and the room. And what a room that room is!

Over the past three decades, I have seen a lot of high-end audio gear set up in rooms that were not much more spacious than oversized closets. (Years ago, I heard the original Infinity Servo-Static, sub/ satellite system in a room that probably was about 15 feet square.) However, Gary has his system set up in a room that a lot of high-end enthusiasts would be willing to kill for.

How about these dimensions: 21 feet wide and 31 feet deep, with a ceiling that is 8 feet high at the sides, with a 10-foot peak?! Total enclosed space: about 5800 cubic feet.

Not only that, but this is no large "family" room with a bunch of large glass areas, furnishings, and doorways flanking the speakers, and maybe with a kitchen or dining room off to the side. Rather, it is a dedicated home theater room, with no window openings whatsoever in the front and side walls, with minimal furnishings at the speaker end, and with a plush carpet covering the floor area.

It has a high-quality, ceiling-mounted, Sony CRT projector aiming its beams at a 9-foot screen on the 21-foot wide front wall. The equipment array is in an adjacent room (controlled mostly by an RF remote control), which not only gets a typically unsightly stack of equipment out of sight, but also eliminates any chance whatsoever for microphonic feedback to add colorations, even at high listening levels. (Note that neither Gary nor I believe this is a potential problem, but the main room certainly does look cleaner for want of a bulging equipment rack.)

Gary has a fair amount of hardware in that room. I will not go over the full package, but the significant things are a pair of 350 wpc Carver M1.5t amplifiers to drive the four (yes, four) 901s in the main room, two Bose 901 equalizers, a Fosgate surround processor, a Carver C-1 preamp, a Richter Scale crossover for his subwoofer system (more on that below), a Carver C-500 amp to drive those subs, and a Sony CD player to provide most of the musical signals. There are other items (LD player, VCR, tuner, etc.) but those I have listed are the items that will matter most to most audio enthusiasts. The combined power of the three Carver amps allows this system to easily hit 105 dB peaks at the listening position.

OK, now, let's face it. What matters most in this case are the speakers. For those of you who are not familiar with how the 901 works, let me briefly go over the concept.

The 901 is actually a "one-way" speaker system, in that it has nine full-range drivers in the speaker box, with no crossover. Each driver is about 4.5" in diameter. Eight of them are mounted on the backside of the cabinet: four set up in a square configuration in each of two panels that are angled slightly to either side of a line that could be drawn straight back toward the front wall. The remaining driver is aimed straight forward into the listening area. The speaker box itself is rather small, being about the same size as a smallish bookshelf speaker system.

[ILLUSTRATION OMITTED]

Normally, a 4.5" driver will not exhibit flat frequency response over the entire audible bandwidth, with driver-efficiency rolloffs being axiomatic in both the bass and treble ranges. However, a good 4.5" driver's midrange performance has the potential to be exemplary.

These characteristics (in the bass, midrange, and treble) are also the case when nine such drivers are ganged together in the same enclosure, although the coherent reinforcement of the ganged drivers on the rear panels should help to bring up the bass range somewhat. Bose overcomes the remaining bass deficiency and significant treble deficiency by using moderately controllable, "active equalization" to boost the bass and treble ranges. This allows the system to deliver a reasonably flat input to the listening room. The Bose active equalizer may be located in a tape loop, or between preamp and power-amp stages in a system that has separates.

The problem of active equalization with multiple drivers of this kind is something that Bose enthusiasts have had to face for some time. Although the drivers in each cabinet have a total moving-surface area roughly equal to that of a 12-inch woofer, their excursion limits are certainly not as great as what a decent woofer of that size would have. Bose overcomes some of this limitation by using a very special, ported enclosure that increases the efficiency of the drivers down really low.

Of course, many systems that employ long-excursion 12-inch drivers (including subwoofers, such as the Velodyne CT-120) also have ported enclosures, and so they share the same advantages as the 901, in addition to having that longer excursion ability. So, while active equalization can cause the nine drivers to be quite flat down to where the port no longer assists, the maximum output of those drivers down really low may be limited.

In addition, even though the drivers can be equalized for flat output into the treble range, we have the problem of driver beaming above a certain frequency. Yes, the rear-facing arrays are bouncing the majority of sound off the front wall, but each of those two, angled, four-driver arrays actually behave as a single driver in some ways. Consequently, they beam even more emphatically than a single 4" driver would. How much the beaming will affect that rear-aimed sound will depend upon the makeup of the wall itself, as well as the distance from the speakers to the wall (more on this, below), and the distance from the speakers and wall to the listener.

Of course, the forward-aimed driver is going to be more conventional in its directional behavior than the rear-facing panels. From about 2 kHz on up a 4.5 incher will begin to exhibit some directionality that will get progressively worse as the frequency continues to climb upward.

Now this means that the power response in the treble range will fall off (it also falls off with the rear-facing drivers), and the way Bose corrects for this is, as I have noted, to have the active equalization supply more power at higher frequencies than to the middle frequencies. This can indeed flatten out the power response, but of course it can do nothing at all for the reduced radiation efficiency of the drivers themselves as the frequency climbs. The forward-facing one is still going to beam as the frequency climbs, and the upshot of the compensatory increased input to the system in the treble range will be that the on-axis response of all the drivers will be rising.

It is hard to judge just what this means with the rear-aimed drivers, but what Bose probably is counting on with the forward-facing one, which is the driver that supplies the listener with first-arrival imaging information, is that with the speakers aimed according to factory recommendations, the on-axis beaming, which we need to remember is only going to be significant above about 2 kHz, will not usually be aimed at the listener. This means that at a typical off-axis listening position (off axis from each forward-aimed system), the first-arrival response from the forward-facing drivers will parallel the system power response. There is no reason why this should not work reasonably well. The net result of this active-equalizing and proper aiming is that the system is able to generate fairly flat power response into the room, and what the listener will predominantly hear will be that power response.

However, the three driver arrays (double sets of four to the rear and one driver facing forward) will still beam. The system therefore cannot exhibit a uniform radiation pattern at all frequencies, and so the soundstage may vary slightly at different lateral listening positions, provided the listener is sitting close enough to the speakers for the first-arrival signal to be reasonably significant. (This is not uncommon in speakers other than the 901, by the way.) The strong bounce from the front wall inserts a wild card into the situation that makes it difficult to compare the system to systems that are forward-facing designs.

The 901s come with some pretty emphatic placement recommendations. The company suggests setting the 901s up so that their rear-aimed panels are about a foot (give or take a few inches) from the wall behind them. The short delay supposedly allows the speakers to mimic the bounced-around sound of a live-music environment. To be truthful, I really cannot see how a delay that brief (probably in the range of three milliseconds) can possibly mimic the acoustics of a concert hall, church, or nightclub.

However, from what has been written about the 901, the bounce created by the signals from the angled rear panels does indeed allow the speakers to sound more spacious and spread out than what we would have with typical, forward-facing speakers in similar locations. Part of the reason has to be that the first-arrival signal from the system is considerably weaker than the reverberant-field response, even when sitting close to the speakers. It would be much weaker than what we would get even from very wide-dispersing conventional designs.

Bose also markets, and Gary uses, attractive floor stands for the speakers that get them a couple of feet off the floor. Obviously, this is good for decent soundstaging. However, the stand also puts the system at a distance from the floor that will create boundary-related cancellations in the bass range. Typically, the height would result in a mild dip at about 100-125 Hz, and if the systems are a foot from the front wall there would be an additional, mild cancellation at about 350 Hz.

Fortunately, these dips are far enough apart that unless one or both speakers were at a distance from the side walls equal to the distances to the floor or front wall, the sonic consequences would be minimal. This kind of placement is unlikely in most rooms that would be suitable for the 901 systems. In addition, the Bose active equalizer has a bass-correction feature that should be able to partially compensate for these artifacts. Combining proper placing with an intelligent use of the active equalizer (meaning that one should at least have a decent RTA) should result in reasonably smooth response throughout the mid-bass region.

It should also be noted that another wild card with Gary's particular setup is that his main pair of 901s are positioned somewhat differently from what Bose recommends. Actually, the positioning is considerably different, because in this case the speakers are about 5 feet from the front wall. They are centered about 10 feet apart, are about 5.5 feet from either side wall, and are about 19 feet from the listener's ears at the prime listening position. Needless to say, this puts the listener well into the reverberant field at just about all audible frequencies. Also needless to say, this kind of placement is only going to be workable in really large listening rooms.

I should also briefly note here that Gary's second set of 901s is used for surround/ambiance effects. They are hung from the ceiling well above and several feet behind the main listening couch, with the eight main drivers aimed at the room's rear wall. I listened to the full system with and without the surround speakers in operation, and although they contributed a subtle degree of ambiance to some of the movies, as well as to standard musical programs, I had them bypassed for a fair amount of my music listening. I also indicated to Gary that they would probably do a better job of simulating both concert-hall and movie-theater acoustics if the multiple driver arrays in each box were aimed at the side walls, rather than the rear wall.

Getting back to the primary pair of 901 systems up front, the result of their somewhat unique positioning is that we have a strong front-wall bounce that is delayed about 11 milliseconds behind the initial signal from the forward-facing driver. This is still not enough to radically separate the two signals with musical program material, but the angle of the wall-aimed signals does allow for secondary reflections to subjectively widen and slightly deepen the soundstage. The effect was clearly audible with this arrangement.

That was the case with nearly everything I heard on the speakers. The imaging was not as precise as what one would find with typical, forward-facing systems, but in a room this large that has the speakers 19 feet from the listener, I would not expect pinpoint, "hi-fi sound" imaging from any speakers short of big theater-horn jobs. On the other hand, the soundstage was considerably more expansive than what we typically find with conventional speaker systems located in somewhat smaller rooms.

Some of the first-class material I listened to (much of this was on the Delos Engineer's Choice II sampler) that sounded terrific on certain other systems have auditioned sounded a bit too diffuse and spacious for my taste on this one (remember, my taste runs towards spacious sound to begin with). With smaller-scaled program material, I certainly did miss some of the details I hear when I sit closer to a good pair of tight-imaging speakers.

On the other hand, certain other releases seemed to come alive when played on this system. For example, the Columbia reissue of Brubeck's Time Out CD sounded better on the 901s in Gary's room than with any other speaker/room combination I have heard.

Interestingly, it was with music of this kind that the system sounded at its best. Jazz combos done in small, rather acoustically dead environments took on an almost live-music feel, as if the group was in the room, or perhaps in a somewhat larger studio environment, right along with the listener.

With larger-scale works, however, the room's acoustics sometimes tended to work against the recorded acoustics. With smaller-scale works that were done in large spaces, I again felt that the room tended to dominate too much, with detail being somewhat masked.

Still, the sound was not the disaster that some Bose-haters would have predicted. Indeed, most of the time it was really quite good.

In addition to listening, I also took some room-response measurements with my trusty SA-3051, and did so while using the same kind of 20-second-averaging, moving-microphone technique as I use when I measure in my three, admittedly smaller listening rooms back home.

Dealing with the low frequencies first, my initial measurements indicated a substantial bass plateau between 50 Hz and 200 Hz, with a slow roll-off below 50 Hz and a return to midrange levels at about 25 and 250 Hz. I am certain that all of the bass inflation was because of the way Gary's pair of JBL subwoofers (placed against the front wall, roughly as far apart as the 901 systems) and the Richter Scale crossover were interacting with each other and with the room. I am not sure which component was at fault, but I am sure that the fault did not involve the 901 systems. Indeed, when we disconnected the subs and crossover and ran the Bose systems and their active equalizer full bandwidth, the midbass was quite flat (+/- 2.5 dB from 40 Hz to 500 Hz, actually), with a mild peak at about 30 Hz, and then a very rapid roll-off below that point.

This is actually extremely good performance over that range, and when I listened to a few of the recordings that had sounded a tad bloated with the Bose/JBL combination, they sounded considerably more detailed and open with the 901 systems operating by themselves. The bass did not have quite the same impact at times, and probably things would have been a bit more impressive if the output below about 80 Hz had been elevated slightly by means of a genuinely good subwoofer/crossover system. However, the overall sound was notably clean, especially considering the listening distances involved.

Of course, there is more to sound than bass and midbass, and so I was particularly interested in the room-response readings of the systems throughout the midrange and treble. Here, things were not quite as smooth as they had been from 500 Hz on down. The curve displayed a 4 dB dip at 630 Hz, a 5 dB peak at 1.6 kHz, and a 5 dB dip at 3.15 kHz. Above that frequency, the response smoothed out to a very flat +/- 1 dB between 5 kHz and 12.5 kHz. Further up still, the response dropped off rapidly, at 16 kHz, the reading was 9 dB down.

OK, let me first say that the so-called lack of treble that some claim exists with the 901 is pretty much a myth. Yes, the response at 16 kHz was way down. However, there is not a lot of music up there, and at 12.5 kHz, where a lot of tweeter-equipped systems are beginning to sag, the 901 pair's room response was only 1 dB down. That is really pretty good, although the forward-facing driver had to be beaming pretty significantly at that frequency, too.

The dips and peaks between about 500 Hz and about 4 kHz were not examples of the smoothest midrange response I have ever recorded, but I have also seen other, otherwise decent systems that were no better. The dip at 630 Hz was no big deal, but the peak at 1.6 kHz was fairly significant, and lent the system a slightly forward quality. The sloping downward above that peak and the resulting dip centered at 3.15 kHz were almost certainly the result of the loss of radiation efficiency with the 4.5-inch drivers, and the return to flatness above the peak, between 5 kHz and 12.5 kHz was certainly an example of the active equalization in action.

Interestingly, that 3.15 kHz dip resembles the kind of power-response notching typically seen with midrange/tweeter or woofer/tweeter systems that have their crossover points near that frequency, and its existence is there for similar reasons: a loss of radiation efficiency as a driver begins to beam.

With typical woofer/tweeter or midrange/ tweeter systems, a dip of that size, followed by a return to flatness somewhat higher up, would be the result of the midrange or woofer rolling off at the top of its range, with the tweeter cutting in a bit higher up and restoring uniform dispersion and flat power. With the 901, the dip is the result of the same kind of loss of radiation efficiency with the single and ganged 4.5-inch drivers above about 2 kHz, and the restored response uniformity above the dip is due to the active equalization. What we see in either case are different solutions to the same problem.

Ironically, if a group of knowledgeable enthusiasts were to look at the room-response curve of the systems without knowing what those systems were, I would guess that most would judge it to be the curve of a three-way system, with a woofer/mid crossover at about 630 Hz and a mid/tweeter crossover at about 3 kHz. Interesting, to say the least.

I found three deficiencies in Gary's system. Interestingly, none is related to the 901s or the way the front pair was positioned.

First, for the best in home-theater sound this package needs a center-channel speaker. Indeed, even fairly directional speakers, with excellent imaging, and a toed in placement arrangement need a center-dialogue fill for movie program material. This was quite apparent when we watched a movie on the system. (The Sony projector was spectacularly good, by the way.) A new age is upon us, people, and center channels are a part of that new age, even for music listening.

Second, the system needs a good subwoofer or subwoofers. Those JBL units, working in combination with the Richter Scale crossover just do not make it, in my opinion. A sub that is only strong to 40-50 Hz is not a real subwoofer.

Third, the 901 systems doing surround-channel duty need to be oriented so that most of their sound is bounced off the side walls on either side of the listening position, rather than the rear wall. I think that this would result in a much better sense of theater and concert-hall spaciousness.

Oh, yes, the overall system needs a DVD player and a Dolby Digital (and probably DTS) decoder. Otherwise, the room/speaker combination and companion video hardware form a very nice package, indeed. --HF

Comments from Gary Eickmeier: Many thanks to Howard for taking the time to come down here and listen to my speakers. We both learned a lot. I have just a few comments to add to his write-up.

First, I would just note that I have no stock in the Bose Corporation, and the use of the Bose 901s was not the main point of inviting Howard here. The fact is, the Bose 901 is just the closest commercially available speaker to my theoretically ideal speaker, in terms of radiation pattern. I have built a couple of prototype speakers of my own design, similar to the 901 but with wider frontal radiation pattern, but they are for experimental purposes only, and I have not been able to make them sound as good as the 901s yet. At the same time, I did wish to show that the 901s certainly do not sound as bad as the high end press and other nay-sayers would have you believe: quite the opposite, in fact.

But the main thing that I wanted to show was how to correctly deploy this type of speaker in a properly designed listening room to achieve a type of imaging that is different in kind from most other speakers. It is very easy to get it wrong with a speaker with this much reflected sound incorporated in its design, and that is usually what you read about in the bad press that it gets. Because of the great amount of reflected sound, the stereo image is formed as an aerial image in the region behind, between, and to the sides of the speakers, rather than just a left to right, speaker-to-speaker type of image. You get the impression of individual instrumental images forming as solid points in space right there in the room with you.

The key to getting it right is speaker positioning. A speaker with this much reflected output forms its imaging as a summing localization with all of its reflected images, not just the direct sound from the actual speakers. To understand this, you make a drawing of the positions of the speakers and all of their images on the other side of the reflecting surfaces nearby. I call this an image model drawing, and it shows graphically what you are really hearing. In my speaker positioning scheme, I place the speakers 1/4 of the room width out from the front wall, and also in from the side walls. For a 20-foot-wide room, this means 5 ft out and 5 ft from side walls. What this accomplishes is to space the speakers and all of their images exactly evenly from each other. What happens if you violate these rules is even more interesting, and I believe counterintuitive. You would think, for example, that moving the speakers farther from you would increase depth of image. But if you do this, you move them closer to the front wall, bringing the reflected images of the speakers closer to you and compressing depth. Pulling the speakers toward you and away from the wall increases perceived depth of image. Similarly, you would think that moving the speakers wider apart would increase width of image and spaciousness, but just the opposite happens. If you move the speakers wider, you move them closer to the side walls, and the side images come closer to the speakers, decreasing width and collapsing spaciousness.

This entire concept of using the reflected output of the more omnidirectional speakers to help form the frontal sound image is so different and so foreign to what most of us have learned that I wrote an entire paper on the subject. It is an attempt to rewrite stereo theory to incorporate reflected sound. The title of the paper is "An Image Model Theory for Stereophonic Sound" (preprint #2869, presented at 87th AES convention, 1989 October 18-21, New York).

As a final note, I would just say that my theory really just brings us full circle to fairly standard audiophile practice in speaker positioning, but with somewhat greater understanding about why it works that way. Most audiophiles put their speakers on stands and pull them well out from the walls, but most advice on exactly where to put them has been based on frequency response models such as the effect on comb filtering or bass reinforcement. My theory says that you should position speakers for imaging, not for frequency response. My ideas run contrary to most advice on room treatment as well. You remember the one about finding the first reflection points with the mirror technique, then putting Sonex or curtains at those points to kill those reflections? In my room, I did that, but left the mirrors there to encourage those reflections! I do this because I want hard, specular reflections so that the reflected images will have the same frequency response as the actual speakers. I am proud to say that this reasoning has been (independently) confirmed by acoustical consultant David Moulton (http:// moultonlabs.com), who has used these techniques to build some very successful monitoring rooms in recording studios.

My thanks to Howard also for bringing his analyzer and doing a far field response curve for me. We found a couple of inexplicable frequency response anomalies, and some problems with my subwoofer response. Howard suggested getting some Hsu TN1220 subs, which I am looking into. It was a great visit, and I wish we had more time.