Emerald Physics CS2.3 Mk II Loudspeaker

In the Emerald Physics room of the Newport Audio Show the year before last I experienced some of the best sound I’ve heard at any show in over forty years as an audiophile, the speakers those under review here. This is what I wrote in my report: “The CS2.3 II offered simply the most precise imaging and best soundstaging I have ever heard anywhere; for once and without exaggeration I could use the word holographic. Dynamic range is extremely wide, bass response subterranean, midrange rich and open. Orchestral music is spectacular, while Sinatra on “Angel Eyes” is palpably present, three-dimensional, and spookily real—also rare at this show, unmistakably a true baritone (it took all of maybe twenty seconds before a smile appeared on my wife’s face). If this thing performs in most rooms the way it did here, and given the quality of these hotel rooms that should be practically anywhere else, it may be just about the best performance per dollar I know in high-end loudspeakers.”

Hyperbole is a liability of show reports because most of the sound at audio shows is so poor that when one hears something good one tends to overreact, not least because negativity, even when warranted, is so dispiriting. But as the review that follows will reveal, apart from the “subterranean bass” bass observation, which is overstated, I stand by that report.

The Design
The speaker is a descendent—sort of—of the original Emerald Physics CS2 that my colleague Robert E. Greene reviewed so enthusiastically in TAS. I say “sort of” because when the company was purchased by Walter Liederman a few years ago, the CS2 became the $5995-per-pair CS2.3 Mk II, which is claimed to share no parts with its forebear and its midrange/ tweeter configuration is markedly different. Nevertheless, the two models do share some basic design principles and philosophy: large stacked woofers vertically mounted on an open baffle, and the use of DSP to assist in crossover slopes and equalization. They differ in that the CS2 crossed over the dual 15″ woofers to a compression-driver tweeter, with no midrange. The CS2.3’s point-source midrange/tweeter is said to improve the speaker’s imaging. The design is unusual enough to warrant describing in some detail its components parts and how they’re meant to go together.

The drivers, though not manufactured by Emerald, are proprietary. Per side they consist in a pair of imposing 15″ woofers that operates below 100Hz and a 12″ upper-bass/ midrange driver with a coaxially mounted, lens-loaded 1″ tweeter (henceforth I shall refer to this as the midrange/tweeter module even though it handles part of the upper-bass). The speakers operate as dipoles up to the transition to the tweeter. Biamping is required and necessitates either four mono amps, two stereo amps, or one four-channel amp. Per side, one amplifier connects directly to the stacked woofers, the other to the midrange/ tweeter module through an outboard crossover box that crosses the midrange to the tweeter at 900Hz. Unlike the woofers and the midrange driver, the tweeter operates in monopole mode (more about the implications of this later).

Also supplied with the system is an outboard digital signal processor called the DSP2.4. (In the original version this was an outboard DSP and equalizer by Behringer, but the current owner and engineers felt that it was too unreliable to be retained and that it wasn’t “truly an audiophile product.”) Although physically very compact—about the size of a cellphone differently configured and thicker—the 2.4 is a powerful and sophisticated device. Connected between the preamplifier and the power amplifiers, it serves two functions: crossing over between the woofer and the rest of the frequency range, and judiciously equalizing to flatten and smooth the response throughout the upper bass and lower midrange, exactly the region where so many floorstanding speakers exhibit the unfortunate “floor bounce” effect that makes them sound too thin or lean. Note that the CS2.3 doesn’t offer full DSP room correction; instead it offers an array of fixed-equalization settings that help to flatten the bass response in a range of listening rooms.

Full DSP room correction is usually done with a microphone and several measurements taken in situ, but the current CS2.3 II went a different route. When Clayton Shaw owned the company, he undertook a program of extensive and detailed measurements of speakers in a wide variety of typical listening rooms so as to determine their bass-response characteristics at various distances from front and side walls. Using these he came up with algorithms to flatten the response curves. End users supply Emerald Physics or the dealer with the dimensions of their rooms (preferably with a diagram) and the desired location of the speakers, and the manufacturer programs the appropriate response algorithm into the DSP box. The owner must also supply the sensitivities of the amplifiers to be used. It is not necessary that the amplifiers be the same, but if they are not, the DSP2.4 must be programmed so that the amplifiers’ signal outputs are identical. I had on hand only two amplifiers of such different characteristics—a Quad 909 and a Zesto Zia—that the good people at Wyred 4 Sound generously lent me a pair of superb ST Series II amps (see sidebar), so the issue of identical sensitivities didn’t arise. Emerald specifies the sensitivity of these speakers as 97dB, so quality of wattage matters much more than quantity. Although the sensitivity is 97dB at 1kHz, the separately powered bass drivers’ sensitivity is 90dB. As for tubes versus solid-state, bear in mind that each speaker has two large woofers operating in free space, so logic suggests amplifiers with reasonably high damping factors, which most of the time means solid-state.

In addition to the DSP2.4 there is also, as noted, an outboard crossover for each speaker array. The outputs of the bass amplifier connect directly to the paralleled bass drivers, but those of the upper amplifier go through the crossover box, which has separate outputs for the upper-bass/midrange driver and the tweeter (the boxes, supplied with all necessary cabling, are each placed behind a speaker on the supporting platform).

As far as setup goes, the 2.3’s manual makes a point of getting the speakers out from the wall behind them. As with, say, Quads and other dipoles or bipoles, one meter is a minimum, and much more is preferred. Being dipole over half their range, the farther the 2.3s are out from the wall the more the first reflection is delayed and thus the more precise the presentation. I set up mine pretty much where my Quads go and even with the assist of measurements couldn’t much improve upon the location. While they don’t need to inscribe an equilateral triangle between yourself and them, each array must be the same distance from your listening position and their axes should be aimed either directly at your head or intersecting slightly in front or slightly behind it. The manual suggests toeing them in a bit. Disregard that “a bit”—I am reliably informed that “a bit” was just obeisance to inexperienced audiophiles who’ve been led astray by the soundstaging über alles crowd. The truth is, you want to sit on axis because that is where the tonal balance is the most accurate, and, make no mistake, the tonal balance of this speaker is among the most musically natural of any speaker I’ve heard regardless of price or design type. It exhibits some anomalies, which I’ll come to later, but far fewer than most speakers I’ve listened to, including the vast majority of those costing tens of thousands of dollars and more.

To Direct Or Not To Direct
Perhaps no other aspect of speaker design excites as much debate among audiophiles and designers as the issue of directivity, that is, wide-versus-restricted dispersion. It is beyond question that, all else being equal, the most accurate reproduction of the source is to be gained from speakers that restrict the dispersion—lateral dispersion in particular—because such restriction tends to involve less of the sound of the room than wider dispersion. But there are many audiophiles who like the illusion of sound coming from beyond the boundaries of the speakers, even though, if you stop to think about it, this cannot be accurate, however pleasing it may be.

There is a tendency among both audio reviewers and audiophiles to treat the terms “soundstage” and “imaging” as if they’re the same thing, but although they are related, they are quite different. When we speak of a soundstage we are talking principally about the apparent width and depth of an aural presentation, how convincingly it seems to present the gestalt of a performing ensemble of whatever size or makeup. I am referring here to music-making for which there is a live reference, not to electronically generated studio recordings as such. Long before Harry Pearson in the early days of The Absolute Sound made the concept of soundstaging a central preoccupation in audio commentary and design, it was obvious that many audiophiles wanted reproduction with a greater sense of spaciousness and size, something that extended beyond the confines of both the box that contained most speakers and the boundaries set by the stereo pair itself. (I should point out that Harry himself did not favor bouncing sound off the walls—on the contrary, though he liked the spacious effect of recordings miked like the old Mercuries, he was quite clear that these were best heard when speakers were placed out of doors. He had a point: no room reflections.) This could be seen in the popularity of such controversial speakers as the Sonab, the Bose 901, and even the Hegeman. And it persists in the practice of many audiophiles who set up their speaker systems so that the drivers fire straight ahead, the idea being to generate more prominent reflections off the side walls, the better to increase the apparent size of the soundstage. For my tastes, this is precisely the wrong approach. The only reason for listening to a speaker well off its axis is if its frequency response is so poorly controlled that that is the only way you’ll hear a reasonably smooth, tonally accurate response. But it is absolutely the wrong way to go about getting precise imaging.

Imaging refers to the ability of a speaker to resolve spatial cues within the soundstage and to locate voices and instruments precisely. When Consumer Reports famously observed of the Bose 901 that it had a tendency for the image to wander about the room—or words to that effect, I don’t recall the quotation exactly—it resulted in a lawsuit from the manufacturer (which, fortunately, it lost). I seriously question the accuracy of CR’s description—even from the poorest speaker I’ve ever heard, I’ve never felt that the image wandered around the room (assuming the stereo pair is connected in polarity)—but the idea behind the criticism was and is a valid one: namely, that the more you draw the reflected sound of a room into the reproduction, the vaguer and less precise the image-resolving characteristics of a speaker are going to be. It is not for nothing the original Quad ESL, almost sixty years after its introduction, or the Quad 63 and its descendants, thirty years after the introduction of the original, still leave most listeners slack-jawed by their imaging precision. Set these things up correctly, seat yourself equidistant from them, and the only way in which you will more precisely located musicians in a soundstage is with binaural recordings over headphones.

But there is a price to paid, or, rather, two prices: one actual, the other psychological. The actual is that the listening window is limited. Please disregard the “head-in-the-vise” metaphor that is often trotted out. I’ve owned Quads and other speakers that require a restricted listening window for years and the vise metaphor is a demonstrable canard. But the window is undeniably very narrow. The psychological loss is of course the “kick” many audiophiles get from hearing images from beyond the boundaries of the speaker system. This always amuses me, because it’s plainly something of a hat trick, an artifact that depends for its full effect upon one’s not being fully involved with the music that is playing. In other words, the effect depends upon one’s being aware of the place of the speakers in the room and of the sound emanating from beyond their boundaries. Moreover, because this effect depends upon side wall reflections that are obviously mixed into the reproduction, there cannot be an accurate replication of the acoustic space of the recording venue. This is another thing that amazes so many listeners about Quads and other speakers of restricted dispersion: how uncannily they render both the acoustical atmosphere and the physical space of the recording venue or concert hall. Again, there’s no mystery about this: what you’re hearing is more of what is on the recording and less of the contribution of your own room.

There is obviously no way to resolve this debate to everyone’s satisfaction, satisfaction itself being a subjective matter. If you happen to have a very pleasing room, you probably enjoy having some of its sound mixed into that of the recording and reinforcing it. Then, too, unlike, say, Peter Walker, you may not want your sound system to be a window onto the concert hall— you might want something more dramatic or interventionist to make up for the lack of the visual component of music making. “Realism,” after all, is not reality; it is merely a construct that expresses an impression of reality. A bad recording accurately reproduced will sound like a bad recording; but a bad recording can be inaccurately reproduced and perhaps sound not so bad or even good, according to one’s taste. Another way of saying this might be that speakers which image precisely almost always render the soundstage accurately as well, that is, according to what has been captured by the microphones and recorded, but they don’t aggrandize it in any way. As in so many things in audio, you pays your money. . . .

The Sound
I don’t believe it’s necessary for me to run through the usual litany of recordings to describe the sound of this speaker. Let me refer you, once again, to the description from my show report with which I opened this review and suggest that, depending on what you’re used to listening to, fairly little in your experience may prepare you for the CS2.3 II’s extraordinary imaging capabilities. All kind and manner of orchestral, choral, opera, and other big material is reproduced with a thrilling sense of ease, authority, and truthfulness. As I’m writing this I am listening to the splendid recent recording of Otello by Muti and the Chicago Symphony Orchestra and Chorus. The spectacular opening in a thunderstorm through the drinking song that becomes a violent brawl conjures the whole world of Shakespeare’s tragedy as expressed in Verdi’s brilliant score. A universe and sensibility away from this are the quartets of Beethoven as essayed by the Tokyo Quartet in its second complete cycle, which I reviewed during the evaluation period. As with large orchestras, the impression of the four musicians palpably, holographically present, is all but peerless, instrumental timbre true and the tonal balance convincingly real. The Christmas Revels, a pageant staged for recording and based on a live presentation, was brought fully to life in my room, so viscerally you could sense the size of the venue, diagram the movements of the many performers, and hear the hall.

Part of the reason behind this impression of life and lifelikeness is the open-baffle configuration. There is no boxiness to the presentation because there isn’t any enclosure. The baffle that holds all the drivers is wide enough at the bottom to provide the necessary reinforcement for the bass so that the presentation sounds realistically full. Indeed, this is one of the characteristics of this speaker that I like most: the rich lower midrange and full mid-to-upper bass, precisely the spectrum from which orchestras and jazz ensembles derive much of their strength and power. This is one speaker that can render music not just loud but powerful in the way you actually hear it in a good concert hall. I remarked in my show report that—much to my wife’s great pleasure—Sinatra actually sounds like the baritone he is, not the tenor wanna-be that we get from so many speakers, especially those narrow-baffled floorstanders with their bass drivers raised off the floor. I get so tired of the current trend in speakers toward a sucked-out upper bass/lower midrange in combination with a projected presence region and a rising top end that makes everything sound unnaturally crisp, overly articulated, and—well, why use a lesser word?—unbeautiful.

Nor is it just male singers who benefit from an accurate presentation in this region. Put on Ella Fitzgerald in her prime and you realize that while hers is a light voice it’s also got some real body and weight to it. Yet saying that, really light voices, such as the young Barbara Cook on the fabulous original-cast of Candide, are presented that way. Or try Mary Costa on the original soundtrack of Disney’s animated feature Sleeping Beauty for a clear, crystalline voice.

Several musician friends of mine who’ve listened to this speaker have absolutely fallen in love with its sound the way they typically do when they hear Quad ESLs, Harbeths, LS3/5as, Gradients, early Spendors, and a small handful of other brands known for natural, musical tonal balances. (As I’ve had occasion to observe before in these pages, when music lovers buy speakers like these, they keep them for a long time—often a lifetime.) In addition to the other virtues I’ve cited, one big reason my musician friends like these new Emeralds is the top end, which is smooth, sweet, and very natural, but very much not à la the current mode. At first listen, you might even think it a bit dim sounding, but to this I have two responses. First, listen again and then recall what, say, a triangle really sounded like the last time you were at an orchestral concert. One of my listeners—a seasoned audio professional listening to the famous Royal Ballet anthology from Harry Pearson’s Super Discs list—made a special point of noting this when he heard these speakers. One recording I made sure I listened to was Boulez’s DG performance of his Pli Selon Pli with all its high percussion, which the CS2.3 reproduced excellently and very persuasively. (Boulez the composer is not to everyone tastes, to put it mildly, but this piece is a very appealing mid-century classic and often very beautiful in a kind of aural equivalent to Paul Klee-Picasso-like way.) Second, make sure you’re listening on axis. If you’re not, then the top will sound very dim indeed (more of this further on).

Lest I give the impression the CS2.3 II is perfect, it isn’t. Nothing is. So let’s start with that tweeter. The lens loading has the desired and desirable effect of directing the response so that you don’t get much sidewall reflection. But this also results in a considerably less than uniform power response. In other words, if you sit much off the axis, the extreme highs fall off precipitously and in places can seem to disappear entirely (especially if your room is well upholstered). Then, too, lens loading affects the radiation pattern in other ways. You can hear this most clearly on spoken word. There are two speeches by John Wayne on the soundtrack from The Alamo that were recorded outdoors and that are good tests for coloration. The CS2.3 II does not pass these with flying colors, sounding a little closed in and curiously “hooded.” This effect is easily heard by contrast with, say, Quads, which famously exhibit none of it: Wayne’s voice sounds literally as free and open and present as if in the room or, rather, as if you were transported to where he is. Other recordings that demonstrate this effect are Derek Cooke’s voice in analysis of Wagner’s Ring Cycle and almost any test record with an announcer on it. When I queried Mark Schifter, Emerald’s Engineering and Design Consultant, about this, he remarked, with a candor that is genuinely refreshing, “Yes, that’s true. We can’t duplicate Quads in this regard.” It’s hardly necessary to add that most other speakers can’t either.

I am at a loss to explain why these effects don’t show up on music, including solo voice, but they don’t, or at least not nearly to the same degree, whereas they do so clearly on speaking voices, especially speaking voices outside of musical contexts. Probably it has to do with all the emotional and psychological associations we bring to the experience of music, which is one reason among several why Alan Shaw of Harbeth does not use music when it comes to the final voicing of his speakers—neither did Peter Walker of Quad—but instead recordings of people, typically family members, with whose voices he is intimately familiar. Then, too, there are the differing radiation patterns of the 12″ midrange and the 1″ tweeter, plus the fact that the tweeter is monopole, the midrange and woofers dipole. Whatever the case, you will not hear these speakers at anything approaching what they are capable of if you listen much off axis.

I said that my show-report observation about “subterranean bass” was overstated. This is true. The bass from these speakers is impressive down into the thirties, and it can be very powerful. But much below that and all you get is a sense of foundation but little in the way of definition and clarity. This is to be expected. As one speaker designer put it—he shall go nameless—“Dipole bass may be very good, but the trouble is, there isn’t very much of it.” Hence Emerald’s heroic measures: two 15-inch woofers per side plus judicious equalization and DSP room correction and separate amplification. Schifter told me that one of his customers, an industry professional who likes to listen to hard-driving rock, is crazy about the speaker but wondered if there was anything to be done to improve the very-deep-bass situation. The candor of Schifter’s reply was again refreshing: “Not without otherwise ruining the speaker. Our advice is that you buy yourself a REL.” This was offered in all seriousness: the people at Emerald are great enthusiasts for REL subwoofers. Mate the CS2.3 II with a REL (or a pair) and you’ll have a true full-range speaker system that for sheer bass extension virtually nothing on the market can touch.

Finally, one last caveat, inasmuch as I am making comparisons on the highest possible scale: while the CS2.3 II is very accurate, revealing, and lifelike in its presentation, I would not recommend it to the detail-is-everything contingent. Nor does it exhibit the last several degrees of see-through transparency of Quads, some other electrostatics, and a very few of the finest dynamic systems. Saying this, let me add that it is in no way deficient in these areas—indeed, is quite superb—but if that chair squeak in the back row is more important to you than the correct timbre of a violin or a trombone, then you might want to look elsewhere. Otherwise, nothing, regardless of price, is state of the art in every aspect and particular of performance. (I must add here that the vast majority of dynamic speakers, including most of the super-expensive ones, with transparency that approaches Quads, also have a tipped-up tonal balance that I personally find antipathetic.) This speaker excels in so many areas, including those that are central for the truthful reproduction of music in the home, that they warrant the highest recommendation. When you factor in their price—$4800 a pair in the current configuration (but see sidebar)—then their value is quite off the charts. And even if you’re not in the market for speakers, I’d advise you try to audition a correctly set-up pair just to hear what truly precise imaging in its holographic sense really is all about.

SPECS & PRICING

Type: Three-way, four-driver, active dynamic loudspeaker
Impedance: 4 ohms
Sensitivity: 97dB 2.83V at 1m (midrange/tweeter); 90dB (woofers)
Frequency Response: 20Hz–22kHz -3dB (with DSP correction) to target curve
Dimensions: 8″ x 51″ x 2.75″
Weight: 78 lbs.
Price: $5995

EMERALD PHYSICS
(770) 667-5633
emeraldphysics.com