"In-Wall Speaker Test, Part 1"

Article by Keith Yates

Because in-wall speakers are difficult to review, no magazine has ever put them through a rigorous laboratory test. Until now.

Contributing writer Keith Yates has undertaken the tremendous task of individually installing, scientifically measuring and subjectively auditioning nearly two dozen in-wall speakers. In addition to his 16 years of experience as a high-end home theater and critical listening room designer, Keith is an acoustical designer, educated at the University of California, Berkeley.

We invited 20 manufacturers to send him a pair of in-walls–preferably their best–for this two-part series. We instructed Keith to pull no punches, and are confident that this series will be beneficial to both consumers and custom installers. –Ed.


The low sonic standards to which in-walls are routinely held is something of a puzzle: One would think that the entry of such hi-fi icons as KEF, B&W, Apogee and others into the in-wall market a few years back would have invited magazines to judge these speakers against the standards set by conventional, free-standing speakers.

When I agreed to undertake this review, I made it clear that I would subject the contenders to a battery of tests, measuring each against the same performance standards used for conventional freestanding and bookshelf designs, then to disclose the results without mincing words.

Informed of the rigorous performance yardstick I’d be employing, manufacturers responded in ways that said a lot about the mixed state of the genre. Several opined, dismissively, that in-walls, including their own, would “never equal even a cheap mid-fi standalone speaker,” while another allowed that “in-walls get a free ride, and shouldn’t. Good design produces good sound, whether the reproducer is in a wall, on a shelf, or standing on the floor.”

It was surprising how many held the view that consumers are using in-walls for background music only—this despite the fact that there are in-walls selling, often briskly, for up to $3,000 per pair, sometimes employing the same drivers as their freestanding siblings, and that many music and movie lovers are choosing in-walls to replace conventional freestanding speakers in settings, including media rooms and home theaters, where music is far from a background-only affair.

A very substantial part of any loudspeaker’s performance is dictated by its enclosure, and the in-wall’s enclosure is typically built of nonacoustically-enginered materials by someone outside the direct supervision of the manufacturer—an A/V installer, maybe a carpenter or wallboard hanger, maybe the end-user or even his or her kid or neighbor. The soft wood studs, gypsum board and thermal insulation used are often seen (erroneously) as hopelessly inferior to the more “engineered” materials used in the manufacture of the better freestanding speakers. Not even the enclosure’s internal volume is generally known for sure.

If its enclosure isn’t properly designed and executed, even a state-of-the-art in-wall will likely disappoint. Rattles, air leaks, mixed-up phasing, droning cavity effects, wall boom, and acoustic bleedthrough to adjoining rooms are surprisingly common, even in “magazine-quality” installations. But even getting the enclosure right isn’t enough: Selecting the most acoustically appropriate location for the speaker, and then orienting it, are steps often given lip service, but they seldom get the follow-up all speakers, in-walls included, deserve. Many manufacturers are therefore reluctant to lavish their best engineering talent and highest-tech materials on a category that would appear doomed to mediocrity, or worse.

None of these hurdles excuses the prevailing ho-hum attitude about in-walls. Similar excuses can be made for any component in the high-fidelity playback chain. A well-heeled audiophile may connect his Spectral preamp with cables that cause frequency response droops, set his Meridian processor for the wrong surround setting, hook up his Mark Levinson power amp out of phase, and spike his new Wilson speakers to a wood subfloor that produces a muddy, tympani-like drone, yet these possibilities don’t dampen the manufacturers’ enthusiasm for delivering the very best products they can. The in-wall has historically been a disappointing performer because of equal parts installer ignorance and manufacturer complacence. Clearly, the end-user deserves better, and sometimes, as we’ll see, he or she actually gets it.

The Test
After saving my TEF meaurements [see sidebar at end of this installment and “Testing In-Walls“] to computer hard disk, I brought each speaker/wall under test into my living room, placed it against the wall, fitted it with acoustical “ears” to minimize edge diffraction/radiation problems, and then evaluated it on a playback system assembled to realistically represent what an end-user might use to get the most out of top-of-the-line in-walls. A Pioneer PD-F1004 100-disc CD changer’s optical output fed a Soundstream DAC-1 digital-to-analog converter, which drove a Soundstream C-2 THX preamp (set to stereo bypass mode) which fed a Soundstream DA-2 THX 200-watt per channel power amplifier. Interconnect and speaker cabling was by Madrigal and Soundstream.

The speaker under test was compared to a professional control room monitor (Genelec 1037A), a more modest consumer freestanding satellite/subwoofer system (Triad System 7), and one or more other in-walls under test. I based the listening sessions on a core of 10 music selections, ranging from Renaissance music to current rock’n’roll [see sidebar], occasionally supplementing them with additional tracks from other familiar discs—opera, lieder, jazz, blues—when I felt it valuable.

Using pink noise and an SPL meter, I typically matched playback levels to within one decibel. The tests were structured, but informal: I went back and forth between selections and speakers as suited me. With hardwood floors, area rugs, lath and plaster walls and ceiling, the room is slightly more reverberant than the average living room of similar volume. Although I had absorptive panels on hand to reduce the room’s reverberation time, I generally did not use them, as I wanted to audition the speakers under conditions most typical of real-world non-purpose-built listening rooms.

This first installment in the survey focuses, in alphabetical order, on the 10 models that are most affordable and readily available, while the second installment will look at 10 more esoteric offerings from specialty firms.

Atlas Soundolier AS-83A
This flagship model is one of the few three-way designs in the survey, featuring an 8-inch polypropylene woofer, 4-inch polypropylene midrange, and 1-inch soft-dome tweeter. Tweeter level is adjustable via a 3-position switch on the crossover board. The 83A claims a sensitivity of 88 decibels (dB) with a 1-watt input, measured at 1 meter, power handling of 100 watts, and a frequency response of 45-20kHz. The asymmetrical layout of the three drivers would suggest that the speaker’s coverage would be different both laterally and vertically.

Patrick, my installer, reported that the unit was well packed and easy to install, achieving a good seal with the wall, though he noted that the frame/bezel was not as sturdy as some of the competition.

The first unit tested developed a buzzing distortion when given input in the sibilance region (2-5kHz). A second unit was substituted with no further problems encountered. The overall tonal balance of the AS-83A was quite forward, with a prominent treble and reticent bass. This was heard immediately on the Cecilia Bartoli track, where the mezzo’s sibilants (s’s and sh’s) were emphasized, almost as if she had a slight lisp, and her piano accompaniment took on a somewhat clangy character. A similar clanginess was noted on the Pogorelich piano recording of Scarlatti sonatas, which would have benefitted from a fuller, weightier tonal balance. Mary Black’s voice on the folk-pop “Still Believing” showed some nasal coloration, but the most serious hindrance to this track’s success was the inability of the AS-83A to nail down the bass line. This reticence in the lowest three octaves also marred the Sting and Chris Rea tracks, which would have sounded more believable with a more vigorous contribution from the woofer. On the Christopher Parkening recording of classical guitar favorites, the speaker lacked the characteristic body and “woodiness” of Parkening’s instrument. Turned up loud, Rea’s voice on “Auberge” and Hiatt’s on “Riding with the King” became shouty, although overall the speaker seems fully able to handle the manufacturer’s 100 watt power rating.

While clearly not a contender in the best-of-the-bunch race, Atlas Soundolier’s top model offers good power handling, and may well prove satisfactory for users with relatively absorptive listening environments and a preference for a forward, “etched” presentation.

On-axis measurements confirmed the pronounced treble peak, heard as excessive sibilance, in the 2-7kHz range, with another peak at 14kHz, as well as a bass-shy woofer (averaging some 5 dB lower than the tweeter) with a falling response from 100 Hz down. The speaker’s average off-axis response is somewhat smoother, suggesting that it will likely give its best when listened to at some distance away, allowing the reverberated (off-axis) sound a chance to dominate the presentation. The 88 dB sensitivity specification would appear to be realistic.

Boston Acoustics 381
The 381 is a two-way design based on an 8-inch copolymer woofer and a 1-inch dome tweeter made of a silk-like material called Kortec. The 381 is said to be moisture resistant enough to be safely used outdoors. Claimed frequency response is 48-20kHz, with a 90dB/1 watt/1 meter sensitivity rating.

Patrick reported that the unit was well packed. Longer screws than those provided were necessary to mount the speaker in the double-layer gypsum board wall, though once the 381 was mounted it achieved a good fit and seal.

The 381 didn’t take long to establish itself as a contender for best-of-class award. From the very first demo track, a late 15th century Spanish dance-like piece, the 381 sounded clean and well balanced, not just for an in-wall speaker, but for most any speaker in its price range, in-wall or not. The Boston’s smooth, natural handling of vocal material was especially welcome after testing several in-walls with sibilance problems and nasal colorations. The speaker’s performance on the Prokofiev track also earned it big points for staying articulate and sure-footed through orchestral passages that reduced other speakers to fuzzy, bleating messes.

While the 381 couldn’t muster the “slam” to propel the rock tracks into party orbit, the bass that it did offer was crisp, firm and tuneful. On Chris Rea’s “Auberge”-one of the two crank-it-way-up tracks-the drums stayed tight, driving the tune forward while the horns sounded appropriately sassy without going coarse or gritty. The speaker seemed so free of typical in-wall honk-like colorations-which usually show up conspicuously on the piano and acoustic guitar selections-that it could usually pass as a competent bookshelf model rather than an in-wall.

The test data showed the 381 projecting a commendably smooth on-axis frequency response, remaining within a +/- 3 decibel window from about 80Hz on up to 19kHz. The designers were evidently well aware of the importance of off-axis behavior in typical residential settings: Other than a 3-4 dB droop in the 2-3kHz range, the 381’s averaged off-axis response was smooth all the way to 10kHz or so. The on-axis and averaged off-axis curves would suggest that the 381 would likely sound good in a variety of residential settings, whether acoustically dry or reverberant (“wet”), and at a variety of distances away. The 90 dB sensitivity spec would appear to be a bit optimistic; 88 dB would seem a fairer description.

Energy EAS-6.5
Energy’s top in-wall model, the EAS-6.5 features a 6.5-inch diameter woofer cone of a resin/glass bead matrix material called Spherex and a 1-inch cloth dome tweeter. Claimed frequency response is 45-22kHz +/- 3dB, and sensitivity is given at 87 dB/1 watt/1 meter.

No glitches were encountered in unpacking or installing the EAS-6.5. A good seal was achieved without any special effort on the installer’s part. While the supplied installation instructions were adequate, no “owner’s manual” as such was included.

The Canadian-based Energy created a favorable sonic impression straightaway, with a smooth, clear performance from the midrange on up that made extended listening sessions a happy and not infrequent occurence. As with most in-walls, there was no bass in the usual bone-shaking, pants-flapping demo-room sense. Indeed, compared to the leading contenders-including, pointedly, the Polk AB805 and Boston Acoustics 381-the 6.5 is voiced slightly on the light side. The Irish folk-pop tune “Still Believing” from Mary Black could have used a little more oomph from the woofer to get the rhythm moving, and Chris Rea’s voice did not quite show its characteristically dark hue. Christopher Parkening’s virtuoso guitar work was full of searing clarity, without much blur between the rapid cascade of notes; my only quibble was that it, too, needed a touch more contribution from the woofer to get the instrument’s rich woodiness just right.

The measurements show the EAS-6.5 to be a classic case-study of how to get an in-wall right. The on-axis response is especially flat above about 1.5kHz, fitting within a tight +/- 2 dB tolerance. The peak at about 1.3kHz is thankfully narrow, and from there on down to 600Hz the response stays pretty much on course. There is a shallow dip in the octave beginning at middle C (about 260Hz); this may well be a subtle contributor to the slightly forward tonal balance previously noted. Off-axis, the 6.5 is one of the better-behaved speakers in the test, with only a slight dip around 2kHz to mar an otherwise by-the-textbook response out to 10kHz. The 87 dB sensitivity spec was confirmed. In sum, the Energy is clearly the work of engineers experienced in what it takes to get loudspeakers to sound right in real-world settings.

The KEF CR200Qr takes an 8-inch polypropylene woofer and 1-inch polymer dome tweeter into new territory, literally, by placing the tweeter at the acoustical center of the woofer. The intended benefit is to be able to provide a coverage pattern that is uniform in all directions from the speaker-up, down, left and right. Indeed, the round grille and bezel mean that the speaker has no visible orientation. KEF rates the CR200Qr’s response at 45Hz to 20kHz +/-2 dB, with a -6 dB point at 35Hz-quite deep for a speaker of its size. Sensitivity is given as 88 decibels with a 1 watt input, measured at 1 meter.

The CR200Qr instilled confidence in my installer right out of the box. All parts were well packed, and the mounting frame is a sturdy metal affair, unusual for a compact in-wall. He needed longer screws than those provided to mount the speaker in the double-layer gypsum board wall, but then got a good fit and seal without trouble.

On all but a few track selections, the KEF showed an unmistakable tonal personality, characterized by what some might call “added presence” and others simply “glare.” The Scarlatti piano sonatas had an uneven, almost piercing quality, while the Prokofiev orchestral track sounded edgy and congested during the more complex passages. The backing vocals on the Mary Black tune had a rough, shouty quality, and the lead vocal on the Counting Crows track was unflatteringly nasal, as if the singer had suddenly developed sinus problems.

A few tracks seemed to make it through the CR200Qr relatively unscathed: John Hiatt’s “Riding with the King” rocked along without incident, and the Cecilia Bartoli selection managed to sound rather fine on the vocal element, though the piano accompaniment came across somewhat less believably. An interesting thing happened when I walked through the living room while Sting’s “Mad About You” was playing through the KEFs: Moving off-axis-roughly 45 degrees laterally and 45 degrees vertically-Sting’s voice changed quite noticeably. Whether due to the speaker itself or an idiosyncratic interaction with my living room, this KEF did not live up to its maker’s reputation for smooth, neutral tonal balances.

While several decibels louder than competing models with the same power input, the KEF’s on-axis frequency response is far from ruler-flat, fitting a +/- 8 dB window instead of the +/- 3 dB claimed. The overall peakiness in the response would suggest that listening in an acoustically dead room-or simply moving in close to the speaker in a more reverberant room-would only serve to emphasize the aforementioned tonal problems. However, the alternative-listening in a live room or just moving further away from the speakers-will emphasize the off-axis components and, as my averaged off-axis response curve shows, this may not be preferable to the on-axis component. (The individual off-axis curves, not shown, confirm the essential similarity of the left, right, up and down components at 30 degrees, and again at 60 degrees.)

My suspicion is that KEF is too talented an organization to leave these puzzling deficiencies unaddressed for very long. In a rare twist, the 88 dB sensitivity specification provided would seem conservative; 89 or 90 dB would seem a fairer description of the speaker’s ability to turn a single watt of amplifier power into sound pressure.

Klipsch IW-250
Klipsch’s flagship in-wall utilizes a pair of 6-inch woofers and a single tweeter with a 2-position level switch. Performance specifications were not included in the boxes, though elsewhere Klipsch lists the speaker’s frequency response at 40Hz to 20kHz, +/- 3 dB, and a sensitivity of 93 dB with 1 watt at 1 meter. The manufacturer’s installation instructions note that the speakers “will accommodate either vertical or horizontal mounting.”

Patrick put the IW-250 near the top of his list: The packing was excellent, the mounting clips strong and rattle-free, and the fit of the driver baffle near perfect.

The listening sessions revealed the IW-250 to lie just outside the top rank. The speaker got off to a good start: The percussion on the early Spanish Renaissance disc was relatively free of tubbiness; Bartoli’s voice was smooth and devoid of fuzz or tizz; and the Pogorelich performance of the Scarlatti sonata showed that the IW-250 could handle the piano without making it excessively brash or tinkly. The speaker was marked down, albeit slightly, on the pop and rock vocal tracks (Mary Black, Sting, John Hiatt and Chris Rea), where it lacked bass punch and the kind of smoothness through the lower midrange that made the Polk and Boston models so seductive. Of the remaining selections, only the Prokofiev proved troublesome, with the IW-250 adding a subtle honk to the sound of violins and violas. Except for the bass deficiency already noted, the speaker showed good dynamic “snap” on all selections; this together with its efficiency marks it as an attractive option for party environments.

The on-axis measurement shows a slight “V” shape to the response, with the point of the “V” placed almost exactly at 1 kHz. Even so, response tolerance stays within +/- 4 dB from 55Hz all the way out to 20kHz. I analyzed the eight individual off-axis curves to test Klipsch’s claim that the speaker can accommodate horizontal or vertical mounting. The only conclusion I can offer the reader is that the speaker may be able to accommodate it, but good sound can’t. The vertical and horizontal characteristics of the Klipsch (and, indeed, all speakers in this survey except the KEFs) are sufficiently dissimilar that they cannot be considered “interchangeable.” Like the on-axis “V,” the averaged off-axis response shows a depression around 1kHz, though somewhat broader and deeper. This may be evidence that the IW-250’s designers were aiming for essentially the same sound off-axis as on. If so, they have succeeded, with the result that if you like the speaker’s somewhat non-flat balance when you’re up close directly in front of it-say in a dealer showroom-you’ll probably like it when you get it into your somewhat different home environment.

Part of MTX’s “Pointe” series, the MS-8 features a tweeter than can be freely pivoted up to about 30 degrees left, right, up and down. This “pointability” should enable the speakers to be enjoyed sonically even when the physical layout would appear to put the listener well off-axis. The tweeter itself is a 1-inch titanium dome affair; the 8-inch woofer is formed of a translucent polypropylene. The MS-8 carries a frequency response rating of 36-22kHz, a highish 91 dB sensitivity spec, and a 200 watt peak power handling rating.

Patrick found the MS-8’s “wall vise” mounting system easy to work with, producing a rock-solid fit in the wall. The grille made a good, rattle-free fit.

A glimpse of the MS-8’s tonal personality was gained from the first selection auditioned, the “Danza Alta” from El Cancionero de Palacio, where the shawm (an ancient predecessor of the oboe) had a decidedly “a” (as in “cat”) coloration. Cecilia Bartoli’s recital of 18th century Italian songs brought the same coloration to her otherwise radiant mezzo-soprano, and a little clangy effect marred her piano accompaniment. The next selection, a Scarlatti sonata by pianist Ivo Pogorelich, did not suffer as much; the slight clanginess, in fact, produced an effect that was still sufficiently piano-like to be believable.

It was becoming clear that the MS-8 did not have as neutral a tonal balance as some of the other speakers in the survey. On the Mary Black cut the effect was to bring out the backing vocalists; on the John Hiatt it was to add a nasal sort of highlighting to his voice; and on the Sting it was to help set his voice apart from the backing strings, though the latter took on something of a canned sound. The Counting Crows and Chris Rea cuts were brought off rather credibly overall, although the MTX did not anchor the bass quite as firmly as the very best of the field.

Analysis of the MS-8’s measurements after the listening sessions were concluded showed good correlation with my journal notes. The most prominent deviations from flat on-axis response occur at three locations: 1) around 165Hz, where the woofer is pumping out 4 to 5 decibels more sound than in the 1kHz region; 2) the octave from 2.5kHz to 5kHz, where response peaks again; and 3) in the 14-18kHz area, where the titanium dome’s resonance soars nearly 10 decibels above an eyeballed “average” response in the midrange. Averaged off-axis response appears considerably smoother (partially a byproduct of the 8-curve averaging process), though it follows the same basic contour as the on-axis response, indicating that the designers probably had a specific tonal balance in mind and set out to achieve it both on- and off-axis. At 91 dB, the MS-8’s sensitivity spec is not met; 86-87 dB would appear warranted.

Niles HD-650
The HD650 is one model down from Niles’s top-of-the-line HD-800. The HD-650 features a long-throw 6.5-inch polypropylene woofer and a ferrofluid-cooled 1-inch dome tweeter. For long-term resistance to corrosion, Niles’s HD series features powder-coated aluminum grilles, stainless steel hardware, and other touches that embolden the company to term them “100 percent weatherproof.” The speaker baffle has a knockout for an optional infrared sensor for connection to a larger IR-based control system. Niles rates the speaker at 88 dB/1 watt/1 meter, with a frequency response of 45Hz to 21kHz, +/- 3 dB.

My installer found the HD-650 very well packaged, with three inner boxes clearly labelling the bracket kit, frame and grille, and main speaker baffle. The frame mounted easily in the double-layer gypsum board wall. The input connectors accommodated the 12-gauge speaker wires well. Patrick’s only quibble was getting the driver/baffle assembly to seat in the wall, which took some doing as the bezel tended to bow whenever any one screw was tightened down before the others were similarly tight. The supplied installation and operation guide is attractive, informative, and well written and organized.

On listening tests, the HD-650 projected a light, forward tonal balance that lacked the body and fullness needed to make most program material believable. This is not to say that the woofer was lethargic: In some parts of its range it proved rather energetic, imparting a tubbiness to the percussion on the El Cancionero de Palacio disc and a mild, hollow coloration to the Mary Black selection. Despite the long-throw woofer and ferrofluid tweeter—refinements which, all things being equal, would suggest a better-than-average ability to track music’s dynamic swings—the HD-650 sounded somewhat dynamically inhibited. There was a tendency for busy crescendo passages to coarsen and blur together, especially noticeable in the backing vocals on the Mary Black track.

Like most of the breed, the HD-650 lacked the sheer wallop necessary to make the John Hiatt and Chris Rea tracks rock. When pushed hard, the speaker took on a shouty, hollow character, making vocalists sound as if they were hollering at the end of a small tunnel. At all playback levels cymbals were emphasized, though this feature did not always detract from musical enjoyment. At normal playback levels the Prokofiev, Counting Crows and Sting selections proved enjoyable.

The on-axis frequency response shows a broad emphasis in the octave between about 150 and 300Hz (probably responsible for that honk), a notch at 500Hz and a peak at 9kHz, though it manages to stay within a +/- 4 dB window from 45Hz all the way out to 20kHz. Off-axis, the speaker shows the same broad hump in the upper bass/lower midrange, though from about 400Hz to 10kHz the picture is remarkably smooth, suggesting the HD-650 will be more “room-friendly” than most. The on-axis measurements confirmed the speaker’s 88 dB sensitivity rating.

Parasound CS/T-280
The CS/T-280 is Parasound’s best full-range in-wall in a familiar configuration: an 8-inch polypropylene woofer operating from about 40Hz up to 2.4kHz, and a 1-inch dome tweeter from there on out past the upper limits of human hearing. The tweeter is formed of titanium, and appears to be substantially different from the titanium unit in the MTX MS-8. Sensitivity is quoted at 89 decibels, presumably produced from an input of 1 watt (i.e., 2.83 volts at 8 ohms) measured 1 meter away.

My installer was struck by the Parasound’s structural similarity to the top-of-the-line MTX unit. The frame, bezel and grille assemblies appear to be identical. As with the MTX, Patrick thought the Parasound’s frame not quite as robust as a few competing models, though it proved easy to install and adjust for a good fit and seal.

On audition the 280 proved to be a worthy competitor, with no obvious idiosyncrasies, though as a package it was eclipsed, overall, by the Polk, Boston, Energy and, though only by a hair, the Speakercraft. The Parasound couldn’t quite manage to give the piano its full due, showing a trace of boxiness and less than a full measure of tonal weight and body. Sibilants were slightly overdone on the female vocal tracks (Bartoli, Black), while the Chris Rea rock number showed a somewhat unnatural, pinched quality on the vocal line. The Prokofiev selection amounts to a torture test for in-walls, and although it elicited some hollow, cavity-like sounds from the 280, it clearly did not humiliate it.

The tracks that showed the Parasound to best advantage were Mary Black’s upbeat “Still Believing” from Babes in the Woods, and Francisco de la Torre’s “Danza Alta” instrumental from the El Cancionero de Palacio collection of songs from late 15th and early 16th century Spain. On both tracks the Parasound sounded competitive with the very best models in this 10-speaker survey, with an attractively tidy, airy presentation.

The 280’s on-axis frequency response falls within +/- 3 dB from 64Hz to about 16kHz. The peak at 18.5kHz lies outside the range of hearing for most adults, though a few listeners, particularly young females, may find it audible. Like the Speakercraft 6.5 MIT, the Parasound peaks at about 89 dB at 150Hz, with a downward tilt to nearly 500Hz, where it turns back up again to produce a small peak at 800Hz and another around 1.4kHz. Overall, the on-axis responses of these two well-designed products prove remarkably similar throughout the critical 200Hz to 2kHz region. Off-axis, the similarity is maintained up to about 8kHz, where the Parasound begins to ramp up and the Speakercraft ramp down.

I suspect that the Parasound’s unusual off-axis peak in the highest audible octave amounts to a deliberate attempt to add a little “airiness” to off-axis listening; after all, most listening, even foreground listening, occurs off axis, where most speakers have a somewhat dark tonal balance. The 89 dB sensitivity spec was not met; 86dB would seem merited. Despite its trademark high-frequency peakiness, the 280 is a competent all-around performer, and will probably prove to be especially successful in moderately-damped rooms where listening is done well off axis and the listener prefers the previously mentioned airiness.

Polk AB-805
Like the Klipsch model tested in this survey, Polk’s AB-805 is its maker’s top-of-the-line offering, and similarly relies on a pair of 6-inch woofers separated by a 1-inch dome tweeter. And like the Klipsch, the Polk’s driver complement and layout make for a distinctive, elongated appearance when mounted, as intended, vertically. Polk lays ambitious claims for the AB-805’s frequency response: 30Hz to 25kHz, though no tolerance window (like +/- 3 dB) is provided, meaning the spec is, er, nearly meaningless. Sensitivity is given at 91 dB, presumably with 1 watt at a 1 meter measuring distance.

Patrick awarded the Polk’s installer-friendliness with an “A,” noting the heavy-duty, gold-plated five-way binding posts, and a confidence-building feel that was something like holding a fine camera.

It was difficult to fault the Polk in the first four tracks played; my notes became especially cryptic, with many references along the lines of “no problems,” “even tonal balance,” “smooth” and “clean without being tinkly.” It was on the fifth track, the Irish folksinger Mary Black’s jumpy number from her Babes in the Woods disc, that my response changed from detached admiration to genuine enthusiasm. The bass line was firm and lively, Black’s voice was smooth and believable, and even those often-problematic backing vocals were handled with no-sweat aplomb.

More than any other speaker except perhaps the Boston Acoustics 381, the Polk demanded to be compared not just to the other in-walls in the survey, but to the freestanding speakers I had on hand: the twice-as-expensive Triad System 7 satellite/powered subwoofer system, and the 20-times-as-expensive Genelec 1037As. While the AB-805 fell far behind the Genelec, it was instructive to hear how close it mimiced it from the lower midrange on through the treble. Going head-to-head against the Triad sat/sub system, the Polk managed to stay competitive in all areas except ultimate loudness (it grew coarse only when pushed past its 150 watt rating) and bass response, where it proved no match for the Triad’s self-powered 12-inch woofer. Excellent instrumental definition was heard on the “Romeo and Juliet” orchestral track and, similarly, the Chris Rea “Auberge” cut, which begins with some around-the-house sound effects, slowly unfurling into a riveting rock ‘n’ roll showpiece. The only track which produced noticeable coloration was Sting’s “Mad About You,” where the upper bass range sounded a little thick.

Barely contained within a +/- 4 dB window from 50Hz on up, the AB-805’s on-axis frequency response was not quite as smooth as the Energy’s or Boston’s, both of which fit within +/- 3 dB constraints from roughly 70Hz up. The Polk follows the same basic contour as its closest competitors: Broad hump in the upper bass, then a smooth downramp of about 5 dB to 450 Hz, then a few modest ups and downs out to about 3 kHz, where the three speakers begin to show stronger differences. Specifically, the Polk heads north, to a +5 dB peak centered at 4kHz, while the Boston heads south for a few decibels to 4.5kHz, and the Energy makes a bee-line due east, remaining relatively flat out to 13kHz, where it begins to ramp up to its final peak at 15kHz.

The most salient feature of the Polk’s on-axis behavior is its sheer level, generally 2 to 7 decibels higher than the other models in the survey. The practical effect will not be significant at low playback levels, but users who tend to crank it up may want to consider that the Polk will require about one-half to one-fourth as much amplifier power to produce the same playback level as the other speakers in this survey.

The Polk’s averaged off-axis response (taken from 30 and 60 degree measurements left, right, up and down) shows a tonal balance rivalled in smoothness only by the Boston and Energy units. It is probably no accident that these three, which ended up at the top of the pile in subjective listening, all share the same core features: Relatively flat on-axis response, and a smooth, gently downward tilting off-axis response. For those who expect 100-plus decibel output, the Polk’s additional advantage in efficiency will likely cinch it as the best of the lot. The manufacturer’s 91 dB sensitivity spec is probably over-cautious; the on-axis curve suggests that 92 dB would be justified.

Speakercraft 6.5MIT
The “MIT” in Speakercraft’s product designation refers to “mirror image technology,” whereby the tweeter is positioned above the woofer and over to one side. Claimed benefits from this asymmetrical arrangement include improved imaging resulting from lower diffraction and reflection artifacts from the baffle. The woofer itself is a 6.5-inch polypropylene unit; the tweeter, which can be aimed, like the MTX’s, is a ferrofluid-cooled 1-inch soft dome design. Frequency response is given as 45-20kHz, +/- 2 dB; sensitivity is specified at 91 dB/1 watt/1 meter.

My installer found the Speakercraft to be the easiest speaker he’d ever installed, and, although he didn’t require it, he found the inclusion of the manufacturer’s phone number on the installation instructions a nice touch.

On audition, the 6.5MIT consistently gave a fine account of itself. Like the Polk, Boston and Energy models, the Speakercraft didn’t do much serious damage to music, and what distortions it introduced were generally innocuous. The Scarlatti song by Cecilia Bartoli was properly luminous and free of “grit,” though the piano accompaniment was a little lighter in tonal balance than, say, the Energy. The Mary Black tune was handled gracefully, even when turned up, something that caused most competing speakers to grow congested and/or shrill. The John Hiatt number was surprisingly strong, showing good dynamics and a welcome lack of the irritating “shouty”character that blighted some competitors’ performance. A commendable delicacy in handling instrumental textures was heard throughout the Prokofiev track whether the orchestra was playing at a murmur or swelling up in a gripping crescendo. The Chris Rea tune rocked fairly convincingly, the bass drum showing good snap (though not much weight), with very little overhang.

The 6.5MIT tended to stay clean and free from congestion and tonal problems even when pushed hard. Overall, the speaker was just a half-step behind the Boston, Polk and Energy models and, like them, it can rival the performance of many freestanding speakers in its price range. This is high praise, indeed.

While the 6.5MIT didn’t meet the manufacturer’s tight +/- 2 decibel response tolerance across the full frequency range quoted, it did manage to stay within that window from about middle-C (about 260Hz) on up to 12.6kHz. Accommodating the full 45-20kHz spec required relaxing the tolerances to +/-6 dB. The 91 dB sensitivity spec turned out to be decidedly optimistic, with 87 dB a more realistic figure.

The speaker’s on-axis response is generally quite attractive and free of idiosyncrasy, as is the averaged off-axis behavior, which shows a gentle roll-off (roughly 2 dB per octave) from about 150Hz to 2.2kHz. From 2.2kHz it turns and ramps up about 8 decibels to about 5kHz before resuming its downward pattern.

By this point the reader will have gathered that there is a basic formula at work here that consistently results in success in real-world listening rooms: Develop as flat on-axis performance as you can (esp. between about 200Hz and 2kHz), and combine it with a smooth, very gently downward-pointed off-axis averaged response. Like the Boston, Polk and Energy models that only slightly outpaced it, the 6.5MIT is evidence that Speakercraft’s engineers are working the formula quite successfully.

Sidebar: Test Setup

Each speaker was mounted in a four-foot high wall formed of two-by-four wood studs set 16 inches on center, and sheathed with a single layer of half-inch gypsum board on the back side and two layers on the front. The two front layers were separated by a paper-thin, sticky constrained-layer damping sheet (Omni-brand dB-Rock) to control drum-like wall resonances. The 2,436 cubic inch cavity-a volume typical in residential wall construction with requisite fire-blocking-was filled with conventional R-11 fiberglass insulation.

Each speaker was installed by Patrick Calderone, a professional A/V installer based in Northern California, per manufacturer instructions. All speakers were checked for good “seating” in the wall, as well as for buzzes, rattles and air leaks. Both gypsum board layers and contrained-layer damping sheet were changed whenever the cut-out requirements for a speaker did not closely match the hole left by the preceding speaker tested.

To get the most accurate ground-plane measurements possible, each speaker/wall system under test was set flush into the ground, speaker aiming skyward, in a quiet, rural setting—my backyard. The joints between the ground and wall edges were filled and smoothed to essentially eliminate secondary radiation. If the speaker had a level control it was set to the “flat” or “0” position. If, as in several cases, the speaker’s tweeter was pivotable, it was aimed straight ahead and left there.

An ACO Pacific instrumention microphone and a Techron TEF20HI connected to a PC were used to generate, capture and display test signals. The test signals were amplified by a Soundstream DA-2 power amplifier set to deliver 1 watt at 8 ohms (2.83 volts) and then routed to the speaker under test via 12-gauge low-oxygen speaker cable. Each speaker was tested with the microphone at 2 meters away directly on-axis, as well as at 30 and 60 degrees off-axis laterally, and at 30 and 60 degrees up and 30 and 60 degrees down. Lateral responses for those speakers with non-vertically aligned drivers were taken at 30 and 60 degrees both to the left and to the right. Six decibels were added to the curves to show the plots as if they were derived at 1 watt/1 meter.

I exported the on-axis and off-axis response files to WaveMetrics’ Igor Pro scientific analysis and graphing program, where I averaged all eight off-axis curves for each speaker at each of 4,096 data point to provide a composite picture of how the speaker distributed its energy into the environment. (This is a critical element of the sound of in-wall speakers in particular, due to the fact that off-axis energy predominates the sonic presentation in most casual residential environments.)

Prior to taking frequency response measurements, an energy-time curve (ETC) test was run for each speaker to verify the accuracy of the test setup. Reflections off structures, etc. received within the 58 millisecond measuring time window were never higher in level than -40dB, and typically registered about -60dB out to 120 milliseconds, enabling an extraordinary level of performance detail to be captured, stored and analyzed.

I’d like to thank the following individuals for their help in establishing what amounts to the most rigorous regimen ever used for comparative testing of in-wall speakers: Dr. Floyd Toole, director of research at Harman International; Don Keele, technical editor of Audio magazine; and Farrel Becker of Techron, Inc.

SIDEBAR: Software Used

The following compact discs were used in the core listening portion of the tests:

Francisco de la Torre: “Danza Alta” from El Cancionero de Palacio: 1474-1516, performed by Hesperion XX, Jordi Savall, director. Astrée.

Alessandro Scarlatti: “Già il sole dal Gange” from If You Love Me: 18th Century Italian Songs, performed by Cecilia Bartoli, mezzo-soprano, and György Fischer, piano. London.

Domenico Scarlatti: “Sonata K.20 in E major,” Ivo Pogorelich, piano. Deutsche Grammophon.

Mary Black: “Still Believing” from Babes in the Woods. Gifthorse/Curb.

John Hiatt: “Riding with the King” from Riding with the King. Geffen.

Sting: “Mad About You” from The Soul Cages. A&M.

Sergei Prokofiev: Introduction to “Romeo and Juliet, op. 64,” Kirov Orchestra, Leningrad, Valery Gergiev, conductor. Philips.

Counting Crows: “Mr. Jones” from August and Everything After. Geffen.

Chris Rea: “Auberge” from Auberge. Atco.

Federico Torroba: “Castles of Spain” from A Tribute to Segovia, Christopher Parkening, guitar. EMI Classics.