Dunlavy Audio Laboratories SC-IV loudspeaker
For many audiophiles, them's fightin' words. We've all heard speakers with flat frequency response that don't sound nearly as good as the measurements would imply. We've also heard speakers that, while far from exemplary in the measurement realm, nevertheless contrive to sound like music. We pride ourselves in having genuine 24-karat ears that are superior to any instrument (footnote 1). And while most speaker designers make extensive use of computer-based measurement systems such as MLSSA, it's often said that speaker design remains as much an art as it is a science.
Ah, but there's measurement, and then there's measurement. Designer John Dunlavy specifies a "full" set of "credible" measurements. He measures everything about a speaker's behavior, paying particular attention to the impulse response, measured anechoically at the typical listening distance of 10'. He points out that accurate impulse response implies flat frequency (amplitude) response, but not vice versa (footnote 2). Thus, a speaker with flat frequency response (on-axis or integrated over various angles) may be quite inaccurate when it comes to reproducing an impulse signal; ie, the outputs of the different drivers may be arriving at different times.
Dunlavy argues that pulse coherence is essential for the speaker to be considered accurate. He draws an analogy between a cheap lens, with marked spherical/chromatic aberration that blurs the image, and a speaker with a lack of pulse coherence, resulting in blurring of the sonic image. For the past year, Dunlavy has been conducting research to further develop practical implementation of the pulse-coherence concept; his new "Signature Collection" speakers from Dunlavy Audio Labs (DAL) are the result (footnote 3). Dunlavy has been in the speaker business for more than 25 years, and holds a number of patents in a variety of areas of electronics design, so his credentials are certainly impeccable.
Yet it must be said that, no matter how good the theory, how impeccable the designer's credentials, and how impressive the measurements, a speaker may still disappoint in subjective terms. The proof, as always, is in the listening.
Description & design
It definitely takes two people to unpack and set up a pair of SC-IVs. The 6'-tall, 180-lb speakers, which are obviously well out of the minimonitor class, are likely to visually dominate all but very large rooms. However, the SC-IV's proportions are unusually pleasing, and the speaker is mounted on a stand/plinth in a manner that gives the impression that it's "floating" in the air, so the overall effect is not as massive as might be expected.
The review pair was finished in flawless "architectural-grade" oak veneer. Removing the grille revealed a symmetrical arrangement of drivers that those familiar with Dunlavy's earlier Duntech designs will recognize: a centrally located fabric-dome tweeter, flanked by 5½" cone midrange units, with 10" woofers at the top and bottom. The tweeter and midrange are sourced from Vifa, with modifications (including damping of the tweeter's back wave) made at the factory, according to DAL's specifications. The tweeter is relatively inexpensive, but Dunlavy claims that, when modified, its performance exceeds that of the most exotic and expensive units on the market.
The Morel woofers feature long-throw voice-coils. The tweeter and midrange drivers are in a sealed compartment that does not extend the full depth of the cabinet, leaving the rest of the cabinet (including the space behind the tweeter/midrange compartment) to serve as a common enclosure for the woofers. The 1"-thick MDF cabinet is heavily braced and filled with foam (Dunlavy believes that overdamping the woofer produces the most realistic bass response). As with Duntech speakers, there is extensive felt treatment around the tweeter and midrange drivers (footnote 4).
Generically, the rather complex crossover falls into the minimum-phase, first-order (6dB/octave) category, but Dunlavy says it cannot be accurately described in terms of conventional slopes and crossover points. There's also a network to notch out the tweeter's resonant frequency and conjugate load matching to produce an easy impedance load. The SC-IV is set up for single or bi-wiring. The binding posts are extra-thick, which makes them incompatible with most spade lugs, and they're difficult to tighten properly. DAL's Quality Assurance process involves the testing of every component, subassembly, and assembled speaker.
The listening room itself is on the small side (16' by 14' by 71/2'), but has been extensively acoustically treated with the new "Slim" ASC Tube Traps in the corners behind the speakers, CornerTunes and EchoTunes in the usual positions (top corners and the middle of the wall near the ceiling, respectively), and RoomTunes along the side walls to control the early reflections. The floor is covered with a thick carpet and underpadding; there's also a 4' by 6' shag rug hanging on the wall behind the listening area.
The SC-IVs come with a user's manual that contains a sort of mini-treatise on speaker design, with many useful suggestions on how to optimize speaker setup in the listening room. One principle I very much agree with, and which I've followed for a long time, is that speakers should be separated so that the listening angle is at least 60°. Dunlavy points out that the common audiophile practice of setting up speakers along the room's short dimension with the listening area in the middle of the room results in a narrow soundstage and extensive peaks and nulls in bass response. He suggests that the most even frequency response and the widest soundstage are produced by setting up speakers along the room's long dimension, with some acoustical treatment on the back wall. This is essentially the speaker setup I routinely use in my listening room.
I ended up placing the SC-IVs in close to the same location I'd set up the last three speakers I've reviewed—Acarian Alán IV (Vol.16 No.2, p.136), Unity Audio Signature One (Vol.16 No.5, p.119), and Unity Audio Pyramid Signature (Vol.16 No.12, p.205)—and where I normally have my own Quad ESL/Cizek MG-27 combo. The angle subtended by the speakers was about 75°, with the speakers placed unequal distances from the back and the side walls. My listening chair puts my ears at a height of about 38", which is very close to being on the tweeter axis. The vertical listening angle was not particularly critical, the rolloff in the highs being gradual, with a surprisingly good stereo image apparent, even when I was standing up. DAL recommends listening distances from 8'–25'—I made the minimum listening distance with only about 6" to spare! I experimented considerably with toe-in angle, from the straight-ahead position through pointing right at the listening chair, and even cross-firing to a point in front of the listening chair.
The optimal toe-in angle turned out to be a little more than halfway between the straight-ahead and the pointing-at-the-listening-seat positions. With the speakers pointed at the listener, the focus of imaging was actually slightly more precise, but the soundstage was not as wide.
A coupla tweaks
Unlike most high-end speakers, the SC-IV does not come with spikes. When I questioned John Dunlavy about this, he expressed skepticism about the likelihood that spikes would have an audible effect, given the speaker's mass. Well, spikes may not be equally beneficial on every surface, but on the thick carpet of my listening room, four spikes under each speaker produced a noticeable tightening of the entire bass and midbass region, with some improvement in focus in the midrange. I obtained good results with the original Tiptoes, German Acoustic spikes (less so with their cones), and, best of all, Michael Green's new AudioPoints.
Footnote 1: The cartoon, by Elwood H. Smith, that illustrated J. Gordon Holt's series of articles on audio terminology (Vol.16 Nos.7–9) has one listener, with bananas sticking out of his ears, saying, "Measures well," while the other, whose pencil and pad identify him as an audio reviewer, exclaims, "Aggressive, analytical, arrggh!"
Footnote 2: The MLSSA waterfall plot is, in fact, calculated from the impulse-response measurement.
Footnote 3: Dunlavy retains partial ownership of Duntech, the company he founded, but he's no longer involved with them at the design level.
Footnote 4: John Dunlavy holds the patent for this method of diffraction control.