Stereophile January 1996— Systematic: The 2C3D Hologram

This review was originally published in Stereophile Magazine in January 1996.
Read the review below or visit the publisher online at to order a reprint.


America's oldest and largest subjective-review audio magazine. Subscription rates: $35 US (US and Canada), and $75 US (foreign).

To Subscribe, remit all funds to Stereophile, P.O. Box 469027, Escondido, CA 92046-9027,
or call toll-free to charge subscription to your credit card: (800) 444-8908. Outside the US, call (619) 745-2809.



Robert Harley reviews the Avalon Acoustics Radian HC loudspeaker,
Spectral DMC-20 Series 2 preamplifier and DMA-180 power amplifier,
and MIT MI-350 Reference CVTerminator interconnect,
MH-850 Multi-Bandwidth CVTerminator loudspeaker cable,
MIT Digital Reference datalink, and MIT Z-series power-line devices.

Avalon Acoustics Radian HC loudspeaker. Three-way dynamic loudspeaker. Driver complement: Two 9" Nomex/Kevlar-composite cone woofers, one 3.5" Nomex/Kevlar midrange, one I" titanium-cone tweeter. LF alignment: Sealed, Q=0.5. Crossover frequencies and slopes: not disclosed. Sensitivity: 88dB/2.83V/1m. Impedance: 4 ohms nominal, 3.6 ohms minimum. Anechoic frequency response: 34Hz-24kHz ±I.5dB. Recommended amplifier power: 50W-500W. Standard finishes available: Curly Maple, Figured Walnut, Quilted Cherry. Premium finishes available: Myrtle Cluster Burl, Walnut Burl. Dimensions: Inn/ by 19" D by 48" H. Weight: 170 lbs each (net). Serial numbers of review samples: 5290/5291. Approximate number of dealers: Avalon overall, 20; Radian HC, 10; complete Spectral/Avalon/MIT system, 4. Price: $12,500 (standard hardwood) to $15,900 (premium hardwood). Manufacturer: Avalon Acoustics, Inc., 2800 Wilderness Place, Boulder, CO 80301.Tel: (303) 440-0422. Fax: (303) 440-4396.

Spectral DMC-20 Series 2 Preamplifier Inputs: Six line, one phono (phono input optional, balanced input optional). Outputs: Single-ended on RCA jacks, single-ended inverted on RCA jacks, balanced on XLR jacks, tape output on RCA jacks. Attenuation control: 32-step conductive plastic. Balance control: I 5-step with center position bypass. Protection modes: DC offset, oscillation. Dimensions: 19" W by 12.5" D by 2.5" H. Weight: 12 lbs. Serial number of review sample: 200592. Price: $7595 as configured for review (includes DMS-20 power supply, Model 202A Phono Module, Model 203B Balanced Input Module).

Spectral DMS-20 Power Supply for DMC-20 preamplifier Output voltages: Dual ±40VDC at 0.6A, dual ±40VDC at 0.6A, single ±5VDC at IA, single ±10VDC at IA. Noise and ripple: <200mV p-p. Reserve capacity: 180 Joules per supply (±40VDC). Power consumption: 80W normal, 175W maximum. Protection: short-circuit, over- voltage line, transient protected. Dimensions: 19" W by 2.5" H by 12.5" D. Weight: 22 lbs. Price: included with DMC-20.

Spectral Model 202A Phono Module for DMC-20 Series 2 Preamplifier. DC-coupled phono preamplifier. Gain: 30d B, 38dB, 45dB (selectable). RIAA accuracy: 20Hz-20kHz ±0.25dB. Distortion: <0.01% THD and IMD. Noise: 95dB "A"-weighted, referenced to I 00mV at I kHz. Input impedance: 10, 20, 100, 800, 47k ohms (selectable) with I 00pF capacitance. Output voltage: I 7.7VRMS (50V p-p) maximum. Price: adds $700 to price of DMC-20.

Spectral Model 203A Balanced Input Module for DMC-20 Series Two Preamplifier Frequency response: DC-I5MHz, +0, -3dB; DC-10MHz, +0, -0.1 dB. Slew rate: > 500Vps. Distortion: <0.01 % THD and IMD. Input impedance: 10k ohms. Gain: unity or -20dB attenuation, continuously variable.

Spectral Model 201 Balanced Output Module for DMC-20 Series Two Preamplifier. Frequency response: DC-4MHz, +0, -3dB; DC-2.5MHz, +0, -0.1dB. Slew rate: >I000Vps. Risetime: 7Ons. Distortion: <0.01% THD and IMD. Crosstalk: <90dB. Noise: 105dB A-weighted, referenced to I 00mV at I kHz. Output voltage: I .6V RMS, I OOV p-p maximum. Output current: IA maximum/channel, balanced output.

Spectral DMA-180 Power Amplifier. Power output: 200Wpc RMS into 8 ohms (23dBW), 400Wpc RMS into 4 ohms (23dBW), 683Wpc into 2 ohms (22.3dBW). Maximum output current: 60A peak/channel. Static distortion: <0.015%, DC-100kHz, typically 0.009% at 200Wpc RMS into 8 ohms. Dynamic distortion: 0.01% into 8 ohms, 0.015% into 4 ohms with 8-tone cluster test, 500Hz separation. Risetime: <400ns. Settling time: 1.5 µs to -40dB. Slew rate: 600Vps. S/N ratio: 97dB unweighted, 107dB A-weighted. Crosstalk: 98dB below full power into 8 ohms. Input impedance: 100k ohms, Input sensitivity: I .5V. Power consumption:250W quiescent, 1600W maximum. Dimensions: 19" W by 19" D by 6.75" H. Weight: 60 lbs (net). Serial number of review sample: 180422. Price: $7495.

Common to Spectral models: Approximate number of dealers: 2IWarranty: 3 years parts and labor. Manufacturer: Spectral Audio Incorporated, 442 Oakmead Parkway, Sunnyvale, CA 94086.Tel: (408) 738-8521. Fax: (408) 738-8524.

MIT Reference M1-350 Reference CVTerminator.Twin CVTerminator line-level interconnect. Price: $1995/1m pair, $3215/35' pair.

MIT MH-850 Multi-bandwidth CVTerminator loudspeaker cable. Tri-wired loudspeaker cable custom designed for Avalon Radian HC loudspeaker. Price: $8995/8' pair, $14,125/45' pair.

MIT Digital Reference. S/PDIF digital interconnect terminated with RCA plugs. Price: $325/1m, $395/2m.

MIT Z-Stabilizer Mk.11 AC Line Treatment Device. Transformerless AC power-line treatment device, AC filter, and overvoltage protector. Number of AC outlets: one duplex (inserted in parallel with AC line). Dimensions: 17.25" W by 11.375" D by

5.625" H.Weight: 27 lbs. Price: $995 (includes 2m Z-Cord).

MIT Z-Center AC Line Conditioner. AC line conditioner with filtered outlets. Number of AC outlets: 10. Dimensions: 17.25"W by 11.375" D by 5.625" H. Weight: 30 lbs. Price: $1495 (includes 2m Z-Cord).

MIT Z-Iso-Duo Dual AC Line/Ground Isolator and AC Filter. AC line filter and isolation device. Number of AC outlets: 2. Dimensions: 17.25" W by 11.375" D by 5.625" H. Weight: 27 lbs. Price: $1495 (includes Z-Cord).

MIT Z-Cord II AC Line Cord. Double-filtered and double-shielded AC line cord. Price: $175/2m.

Common to MIT products: Approximate number of dealers: MIT overall: 140; Reference products: 18; MH-850 tri-wire set for Radian HC: 5. Manufacturer: CVTL Inc., 13620 Lincoln Way, Suite 320, Auburn, CA 95602.Tel: (916) 888-0394. Fax: (916) 888-0783.

When giving seminars and answering audiophiles' questions, I'm frequently asked why Stereophile doesn't review complete systems. Our readers -that's you-appear to need more information about putting together a system than the mix-and-match approach served by reviewing single products.

I point out that most components from different manufacturers will work well with each other, and that we try to include as much information in the review as possible about what other components the review unit is compatible with. For example, a review of an insensitive loudspeaker with a difficult load impedance will include a warning that a high-powered amplifier with the ability to drive current through low impedance loads is required.

Reviewing complete systems is more difficult than it appears. First, replacing one's entire reference system at one time with the review system makes it difficult to assess the characteristics of the individual components that make up the system. The reviewer loses his frame of reference. One of the first rules of reviewing is to change only one component at a time.

Second, the logistics of correctly setting up an entirely new system on a regular basis are daunting. It can take quite some time to fully tweak a system so that it performs its best-not to mention break-in and settling time. Third, the information gained by reviewing a complete system is likely to be less valuable to the majority of Stereophile readers; most audiophiles are looking to upgrade their existing systems, not start from scratch.

Despite these potential drawbacks, some high-end products can be evaluated fairly only within the system package for which they were designed. The Cello and Meridian systems recently reviewed by Lewis Lipnick and J. Gordon Holt, respectively, are good examples.

Although the components within these systems will work with other equipment, they'll sound their best only when used as intended in an entire system.

There's a compelling argument for designing a music playback chain almost as a single entity. In the hands of the right designers, the individual components can be crafted to work together synergistically. Design aspects left to chance in the single-component approach can be fully understood and accounted for in the complete system.

This approach has been at the heart of Spectral Audio's philosophy since the company was founded 20 years ago. The Spectral products are based on common design principles that, according to Spectral, can only be appreciated when used in an all-Spectral system. Indeed, Spectral designer Keith Johnson spends about half his design time examining interactions between the product under design and the other components in a system.

This integration extends to the interconnects and loudspeaker cables in a music playback system. Music Interface Technologies (MIT) has had a long partnership with Spectral, with some of MIT's designs necessitated by the unusual technical performance of Spectral products. In addition, MIT makes a line of AC power-line conditioning products called the Z-Series. This power- line treatment is reportedly essential to allowing the rest of the system to sound its best.

Spectral and MIT have frequently achieved great sound at hi-fl shows by using Avalon Acoustics loudspeakers. This prompted Avalon to work closely with the two companies to develop a loudspeaker that would be a synergistic match for Spectral electronics and MIT interconnects and loudspeaker cables.

The result of this collaboration is the Radian HC loudspeaker, a special version of Avalon's Radian. The HC was designed specifically for use with the Spectral DMA-180 and DMA-150 power amplifiers. (HC stands for "High Current.") Bruce Brisson created a tripwire loudspeaker cable specifically for the Radian HC, based on the MH-850 Reference series.

Note that Spectral components can be used with other loudspeakers, MIT cables will work with other brands, and Avalon loudspeakers can be driven by a wide range of electronics. The association between these companies doesn't preclude using their products in other systems. Rather, the goal was to create an optimal system in which the various products could perform their best.

The only technical requirement is that MIT loudspeaker cables be used with Spectral power amplifiers. The Spectral amplifiers' ultrawide bandwidth needs the low-pass filter function built into MIT cables to prevent the loudspeaker from potentially being driven by energy in the megahertz range. Of course, the Radian HC version of MIT's MH-850 loudspeaker cable can be used only with Radian HCs.

Next year Spectral plans to introduce a transport using their SpectraLink connection to the SDR-2000 Pro digital processor. The SpectraLink connection will presumably feature separate clock lines to eliminate the need for a recovered clock, and thus lower the jitter in the SDR-2000. The Spectral transport will complete the system.

Spectral DMC-20 Series 2 preamplifier

Spectral DMC-20 Series Two preamplifier: The Series Two is a reworking of Spectral's five-year-old DMC-20 preamplifier. The new model looks like its predecessor and is based on similar design concepts, but contains many refinements and upgrades. Oddly, neither the unit's front panel nor the owner's manual says "Series 2," although a circuit-board cover inside the DMC-20 carries that designation, as does the rear panel.

As configured (with balanced input stage and phono stage), the DMC-20 Series 2 sells for $7595. One of the unit's two slim chassis contains the audio circuits; the other holds the power supply (called the DMS-20). The DMC-20 Series 2's front panel holds four toggle switches for tape/source selection, polarity inversion, mono/stereo switching, and 20dB muting. Three large machined knobs provide volume control, balance adjustment, and input selection. The knobs have a solid and precise feel that combines with the good ergonomics to make using the DMC-20 Series 2 a pleasure.

The rear panel provides a tape loop, five single-ended line-level input jacks, one balanced line-level input, and a phono input. The balanced input and phono stage are options, which the review sample included. The output section has both inverting and non-inverting single-ended outputs, as well as a balanced output. A locking multi-pin jack accepts the DC cable from the power supply.

Inside the chassis, the gain of four line-level inputs can be continuously trimmed over a 20dB range by adjusting small potentiometers. This feature allows precise matching of source levels. Another pair of trimmers adjusts the tape input level. A toggle switch bypasses the trimmer when no attenuation is wanted for the purest sound. When in the bypass mode, a fixed resistor replaces the trim pot. The balanced input module also has a gain trim feature, but operates over a 10dB range. I did all my auditioning with the trimmers in their bypass positions.

The phono-stage adjustments include a three-position gain control (30dB, 38dB, 45dB) and five cartridge loads (10 ohms, 30 ohms, 100 ohms, 800 ohms, 47k ohms). The phono-stage load capacitance is fixed at 100pF.

The DMC-20 Series 2 features an unusual architecture in which smaller amplifier boards are mounted on a large motherboard. The motherboard carries power and ground, as well as conducting control signals from the front panel to the relay switching network. DC power is brought up from the motherboard to the amplifier modules where needed, eliminating the need for long power, ground, and signal conductors on the same circuit board. The preamplifier's point-to-point wiring uses individually terminated MIT/Spectral cables. The polysulfone (a Teflon-derived substance) circuit boards are custom made for Spectral by the instrumentation manufacturer Tektronix.

The DMC-20 Series 2's circuitry is essentially a power amplifier without the output stage. The circuits are based on the designs Keith Johnson has been refining for more than 20 years -designs also used in the Reference Recordings recording electronics. No IC op-amps are found anywhere in the DMC-20 Series 2 -not in the signal path, the DC servo, or even in the power supply. A combination of JFETs, MOSFETs, and bipolar transistors are used, with tight-tolerance pair-matching. Some of these devices are mounted in what look like op-amp ICs. The integrated packaging of four transistors, however, maintains temperature tracking between the devices' junctions. The transistors in these arrays can be more closely matched than transistors in discrete packages. Some of the separate transistor pairs are bound together with small copper bands, again to ensure temperature tracking.

The line amplifier features a differential JFET input stage with a MOSFET cascode. The output drivers are heatsink-mounted power MOSFETs in TO-220 packages that can reportedly source a whopping 1A of current per channel. The output stage "floats" on active current sources, meaning that no ground reference is needed. The class-A, direct-coupled circuit is tightly packed on the line amplifier board for short signal traces. A DC servo prevents DC from appearing at the output, and a protection system mutes the outputs when more than 20mV of DC is detected. A front-panel LED turns red when the unit is in DC protection. The protection circuit, which also guards against shorted outputs, uses an optical coupling scheme that allows it to be outside the signal path and thus sonically transparent.

High speed and low energy storage are design priorities at Spectral. The circuit described is specified as having a bandwidth of 4MHz, a slew rate of >1000V/is, and a risetime faster than 7Ons. The optional balanced input module's bandwidth is 15MHz. Spectral believes that this ultrafast circuitry is essential to good audioband performance. However, because you don't want energy in the megahertz range getting into a power amplifier, the mandated MIT interconnects and cables have built-in low-pass filters.

The Model 202 phono module uses a direct-coupled circuit similar to the linestage, with JFETs at its input. The RIAA equalization stage is a combination of active and passive networks, realized with Teflon-dielectric capacitors and Vishay resistors. The DC servo reportedly took one year of development effort. The price difference between the line-stage version and the phono-equipped model is $700.

Some "balanced" preamplifiers simply convert the balanced input to a single-ended signal with a differential amplifier. The preamplifier then treats the signal as single-ended, converting it back to a balanced signal at the output with a phase splitter. This technique adds two additional stages (the diff amplifier and phase splitter) to the signal path-with no benefit. Truly balanced preamplifiers have four separate signal paths (+L, -L, +R, -R) and a four-element volume control.

The DMC-20 Series 2's circuit is said to maintain the advantages of balanced operation, but requires only a two-element volume control, Spectral stating that four-element volume controls aren't precise enough to provide a sufficiently high degree of common-mode rejection.1 This active balancing circuit is also used in the Spectral SDR-2000 Pro digital processor.

The cool-running DMS-20 power supply uses cascaded discrete regulation of the ±40V audio circuit supply rails. These are re-regulated next to the line and phono amplifier boards with an additional floating shunt regulator. The critical supplies are thus regulated three times -all discretely-with no IC regulators anywhere in the power supply or preamplifier itself. The series-pass elements in the DMS-20 power supply are TO-3 packaged devices on large heatsinks, and those in the DMC-20 preamplifier are in TO-220 packages.

I must comment on the beautiful layout and workmanship inside the DMC20 and DMS-20. The execution appears to be meticulous, with an obvious attention to every detail.

Spectral DMA-180 power amplifier

Spectral DMA-180 power amplifier: The DMA-180 is Spectral's top-of-the-line power amplifier, and has been in production for more than five years. The $7495 amplifier is rated at 200Wpc into 8 ohms, and is claimed to double its power into 4 ohms, justifying its "High Current" designation.

The unit has the classic Spectral look, with a front-panel rocker switch and illuminated Spectral logo. The rear panel holds both single-ended and balanced inputs, along with five-way binding posts for loudspeaker connection. Unusually, the channel marked "left" is on the amplifier's right side as seen from the front, and vice versa. Connecting the amplifier according to the channel designation thus requires routing the left- and right-channel interconnects and loudspeaker cables over each other. I found it simpler just to ignore the markings and connect the amplifier for the cleanest cable routing.

The power transformer is a custom stacked design that consumes the chassis front. Rows of large and unusually deep heatsinks flank the chassis sides.

The circuit contains many innovations, one of which is a layered design in which the signal flows from top to bottom in the amplifier. This technique, called "Vertical Dimension Topology" by Spectral, reportedly reduces electromagnetic-field interactions in the amplifier.

Matched differential JFET pairs were chosen for the input stage for their high speed. Using small-geometry transistors in differential pairs produced a slight noise penalty, but was the only way to make the circuit as fast as it is. The JFETS drive positive and negative polarity sets of cascoded bipolars that convert the input voltages to current. This stage is followed by another pair of cascodes that feed the push-pull driver stage. The fully balanced circuits use a balanced feedback path.

The output stage consists of three pairs of individually powered MOSFET output devices. Each device has its own power supply-a rectifier and 10,000p,F filter capacitor -next to it. This technique is in sharp contrast with the traditional power amplifier design of two or four massive filter caps located in the center of the amplifier and connected to the output devices with point-to-point wiring or busbars.2 Again, Spectral's topology was necessitated by the amplifier's wide bandwidth and fast slew rate. Although the amplifier's intrinsic bandwidth is 1.8MHz, it is restricted to 800kHz at the output terminals. The output devices are apparently biased quite high, judging by the rise in heatsink temperature when the amplifier is idling. The output devices are matched, then individually biased in each unit after burn-in.

An innovative DC servo prevents the direct-coupled design from passing DC to the loudspeakers. This circuit senses DC offset and creates a voltage proportional to that offset. The voltage heats a passive semiconductor element in an IC package, which in turn heats active elements in the same IC package, which then adjusts base-emitter voltages to maintain zero DC offset. By using thermal rather than electrical connection, the DC servo is said to be able to do its job without degrading the amplifier's audio performance.

As with the DMC-20 preamplifier, the DMA-180 is beautifully built inside. The circuit boards are made of the same Teflon-derived material as in the DMC-20, and the parts quality is as good as it gets. The internal wiring is sourced from MIT.

Spectral products are of limited production, and each unit is personally auditioned by Spectral founder Richard Fryer before it is allowed to leave the factory.

Avalon Acoustics Radian HC loudspeaker

Avalon Radian HC loudspeaker: The Radian HC is a modified version of the company's Radian loudspeaker. The HC designation stands for "High Current," indicating the HC was designed specifically for high-current power amplifiers. In fact, the HC's development was spurred by the Spectral DMA-180 power amplifier, the primary amplifier on which the HC was voiced. The HC version of the Radian was also developed in conjunction with Bruce Brisson, who designed the custom, cost-no-object MIT MH-850 triwired loudspeaker cable for the HC.

The review pair confirmed-no, expanded -Avalon's reputation for gorgeous woodworking and meticulous fit'n'finish. The samples were finished in a book-matched Myrtle Cluster Burl veneer, which adds a whopping $3400 to the Radian HC's standard-finish price of $12,500. The standard finishes are Curly Maple, Figured Walnut, and Quilted Cherry; Myrtle Cluster and Walnut Burl are considered premium finishes.

I cannot say enough about the review samples' stunning finish quality. The craftsmanship required to produce these enclosures is extraordinary. The loudspeakers come in pairs in which all the grain patterns match from cabinet to cabinet and enclosure side to enclosure side. The veneer placement is carefully chosen to create the best-looking enclosure possible. Unusually, the veneer is applied after the cabinet is assembled, contributing to the enclosure's seamless look.

The Radian HC is a 48"-tall three- way system featuring the familiar Avalon faceted front baffle. The HC differs from the Radian only in the crossover and internal damping; the drivers and enclosure are the same. The HC was designed to improve the Radian's phase linearity and achieve faster transient response. The musical goals included righter imaging, sharper transients, and greater ambience retrieval. A minor change in the acoustic treatment inside the woofer chambers completes the HC development. The HC version adds $2000 to the standard Radian's $10,500/pair price (with standard hardwood finish).

Realizing the HC's design goals was reportedly possible only after MIT developed the MH-850 tri-wired loudspeaker cable for the HC. As described later in this review, MIT's products are described as transferring an amplifier's voltage and current to the loudspeaker in phase. Without this purported phase alignment in the cable, the HC's crossover refinements would have produced less some benefit, according to Avalon. Designer Neil Patel experimented with different cables, including connecting the DMA-180 directly to the loudspeaker terminals with a few inches of cable. He concluded that only MIT cable could produce a signal at the loudspeaker terminals that was correctly phase-aligned.

The twin 9" woofers are made exclusively for Avalon by Eton. The cones use a combination of Nomex and Kevlar for high stiffness and low mass. Once shipped to the Avalon factory, the woofers undergo further modification. The system's low-frequency alignment, or "Q," is 0.5, meaning it's critically damped. A critically damped alignment produces ideal transient response, with no hangover. Underdamped alignments (a Q of 1 or more can produce deeper extension and a greater sense of low- frequency weight, but at the expense of articulation, detail, and transient response; the woofer continues to move after the drive signal has stopped, smearing dynamics.

The custom-made 3.5" midrange unit, also made by Eton, uses the same Nomex/Kevlar cone material. A 1" MB Quart titanium-dome tweeter, modified by Avalon, completes the driver complement.

The crossover uses large-gauge Litz-wound inductors throughout, and polypropylene or polystyrene capacitors. No circuit board is used; the parts are hardwired together. Avalon considers the crossover frequencies and slopes proprietary, and won't disclose them. Three sets of screw terminals on the enclosure bottom provide the potential of bi-wiring or tri-wiring.

The HC's enclosure is made from layers of MDF bonded together with constrained-layer damping techniques. The front baffle starts life as a 5"-thick rectangular block from which the faceted shape is machined. The facet angles reduce diffraction and aid in the loudspeaker's dispersion. The baffle's slope attempts to provide physical time-alignment of the drivers. The entire enclosure is tilted back, which, along with the faceted baffle, results in almost no right angles in the cabinet.

The enclosure's interior is a maze of damping chambers; the woofer section alone contains 12 separate cavities of different shapes, volumes, and damping -unusual for a sealed design. Extensive bracing makes the enclosure resistant to vibration, a hallmark of Avalon's entire line. Although not particularly large, the Radian HC's density is reflected in its 170-lb weight.

A variety of grille options is available with the HC. The best-sounding choice was also the least visually appealing: a grille frame with felt lining and no grille- cloth. Instead of being covered with fabric, the grille frame simply holds a thick felt panel cut to the frame's shape, with cutouts for the drive-units. This leaves the drivers exposed, along with the felt.

A second option is no grilles at all. You'll see the drivers, but not the industrial-looking felt. The third option is to use the standard grille frames with cloth covering. At Avalon's suggestion, I did nearly all my auditioning with the felt and frame in place. They contend that a large number of their customers willingly sacrifice the HC's appearance for the superior sound offered by the felt. The grille frame is made from injection-molded plastic, and has angled edges that work in conjunction with the faceted baffle structure. Finally, you can remove the magnetically held tweeter covers for better sound.

MIT MI-350 Reference CVTerminator

MIT MH-850 Multi-Bandwidth CVTerminator
loudspeaker cable

MIT Digital Reference
digital interconnect

MIT MI-350 interconnects & MH850 loudspeaker cable: Music Interface Technologies was the first cable manufacturer to use passive networks in their cables and interconnects. These networks are contained within the boxes you see on all of MIT's line-level interconnects and loudspeaker cables.

I'd always regarded this approach with skepticism. Surely the goal of any interconnect or cable is to carry the musical signal with as little effect on that signal as possible. How could adding reactive components to wire be better than a straight conductor? It flies in the face of the conventional wisdom reflected in the name for the perfect and unrealizable amplifier: "Straight wire with gain."

This was the first question I asked Bruce Brisson the day he visited my listening room. He proceeded to give me a tutorial on cables in audio systems, and told me of his research into cable design. The following description is based on what he told me, along with information contained in his technical papers (available from MIT).

Brisson has spent nearly 15 years and lots of money researching the subject of how cables behave. He has developed sophisticated test instrumentation-at a reported cost of more than half a million dollars -to measure and quantify cable behavior. From speaking with him, I was greatly impressed by his depth of knowledge of the subject and dedication to the art, but I can't independently verify his fundamental design philosophy on a technical basis. My knowledge of the subject is derived from standard electronics textbooks that don't begin to describe the phenomena Brisson says occurs in cables. My level of understanding may be inadequate to comprehend the insights of someone who's spent 15 years studying the effects of cables on an audio signal.

I see a parallel in MIT's technical position with that of clock jitter in a digital audio system. Digital audio textbooks make only passing mention of clock jitter, and assume that jitter isn't a factor in a digital audio system's performance. Someone whose knowledge of digital audio was gained purely by textbooks would conclude that jitter isn't important. Only by looking into the subject more closely does one develop an appreciation for jitter's significant role in digital audio performance. I have an intuitive grasp of what happens when a DAC's clock is jittered, something I can't say for an audio signal's behavior when transmitted down a conductor at the level of detail described by Brisson.3

With that background, here's Brisson's explanation of why networks are required in audio cables.

MIT cables are designed around a criterion called "power factor," which reflects the cable's effect on the phase (time) relationship between voltage and current in the conductor. Because power is a product of voltage and current, any change in voltage or current changes the power factor. When the cable's phase angle varies with frequency, the cable conducts more power at certain frequencies than at others, causing a sonic emphasis or de-emphasis of some portions of the audio spectrum. This analysis results in a number representing the percentage of power transported in phase, a characteristic MIT calls "efficiency." A theoretically perfect cable with ideal power factor and flat efficiency of 100% across the audioband conducts all frequencies in-phase, resulting in improved bass, greater clarity, and better soundstaging.

Any cable designed without a phase- alignment network doesn't deliver in- phase power across the band. Consequently the cable's efficiency progressively drops as the frequency decreases. The cable may have 90% efficiency at 20kHz, 50% efficiency at 3kHz, and 10% efficiency at 100Hz. Low efficiency is correlated with an inability to form a soundstage, limiting the soundstage to the music's mid- and high-frequency components.

Because a cable acts as a low-pass filter, it introduces group delay, a form of time-domain distortion. Group delay could, for example, cause the music's harmonics to be reproduced at a slightly different time than the fundamentals. The ideal cable would therefore preserve the phase relationships in the signal and introduce no group delay. Some low-pass filters (Bessel filters, for example) can pass signals with very little group delay and low phase shift.

This is where the networks come into play. The network creates a low-pass filter with more controllable characteristics than the low-pass function introduced by non-terminated cables. Specifically, the network forms a Bessel-like filter that allows the cable to transmit the signal with all frequency components in-phase, and consequently high efficiency across the audioband. MIT's measurements show a flat efficiency curve for their terminated cables.4

MIT's technical papers include three-dimensional energy plots showing the relationship between efficiency and frequency, from which they predict the size and shape of the soundstage.

As for the specific cables under review, the MH-850 tri-wired loudspeaker cable is beautifully made-as it should be for $9000/pair. The spades are made from gold-plated tellurium, and the terminations and workmanship are flawless. The amplifier end of the cable is fitted with spade lugs, along with a small "input terminator" network. The loudspeaker end of the cable has a large metal box housing the terminations, from which three pairs of outputs are provided for tri-wiring. The individual outputs have been separately terminated specifically for the individual frequency bands. Note that the cable's three outputs are full-bandwidth, with the crossovers in the Radian HC's enclosure. The low-pass function is different for each output, but with a corner frequency well above the audio- band. MIT calls this approach "Multiple Bandwidth Technology." A pair of MH850 tri-wired cables weighs more than many small power amplifiers -and costs more than the Spectral DMA-180.

The MI-350 Reference is MIT's top-of-the-line interconnect. The RCA plugs use MIT's patented locking connectors for a tight connection to the jack. The MI-350 Reference has two terminator networks (called "Twin CVTerminator"), one at each end of the cable. The cable itself uses a geometry called "Vari-Lay" that reportedly helps keeps the signal's phase relationships intact. At $1995/1m pair, the MI-350 Reference is the most expensive interconnect I'm aware of-by a factor of two.

The Digital Reference digital coaxial interconnect is said to reduce reflections in the cable and thus lower jitter. Like the analog line-level interconnects, the digital cable is fitted with locking RCA plugs.

MIT Z-Center Power Line Conditioner

MIT Z-Center, Z-Iso-Duo, Z-Stabilizer Mk.II, Z-Cord II:
Much of the work behind MIT's AC powerline treatment products was done by Richard Marsh, who was an engineer at Lawrence Livermore Laboratories for nearly 20 years and has been involved in audio design for quite some time.5

The Z-System AC power-line treatment system and Z- Cord AC cables are designed to keep noise on the AC line from getting into your components. The source of this noise is twofold: the dirty AC line itself, and the other components in your system. A digital processor or CD transport can generate noise that gets into your preamplifier or power amplifier using the AC line as the conduit. A preamplifier plugged into this power-line will be fed this noise along with the AC, where it degrades the sound.

MIT's Z products isolate the various components in a hi-fi system so that noise in one component doesn't invade another. They also filter noise already on the AC line, protect your equipment against voltage surges and spikes, and provide multiple AC outlets for your components.

Most power-line conditioners use series filters to remove unwanted noise above the power-line frequency of 60Hz. Marsh's excellent white paper on power-line noise explains the limitations of series filtering, and shows why MIT's parallel filtering is more effective. The $1495 Z-Center, $1495 Z-Iso-Duo, and $995 Z-Stabilizer all roll off AC line noise with a steep filter that introduces no series impedance between the AC line and the audio equipment. They also don't use the common technique of bypassing the noise to ground with a capacitor, which Marsh shows is fraught with problems.

Looking inside the Z-Stabilizer, I saw two circuit boards, one for over-voltage protection and the second holding a row of 13 large square capacitors and other parts that had been partially potted on the circuit board.

The Z-Center has three isolated banks of AC outlets, with six outlets on one bank and two each on the other banks. Maximum power draw is 1800W from the Z-Center. The Z-Iso-Duo is identical to the Z-Center, but has two banks with two outlets each. The Z-Stabilizer, designed for power amplifiers, has one duplex AC outlet. MIT's Z-Cord Two completes the Z- System AC treatment, with two in-line RF filters to further reduce noise on the line, and double shielding from extraneous noise. The Z-Center, Z-Iso-Duo, and Z-Stabilizer are housed in full-sized chassis, and look identical from the front. Front-panel rocker switches turn the unit on and off, but it's recommended that this switch not be used for powering up or down your entire system. Hospital-grade AC outlets are used throughout the Z-series.

Overall, I was impressed by the straightforward yet insightful engineering in the Z-Series of products. The package appears to be the most comprehensive power-line treatment system available. [Dick Olsher favorably reviewed the earlier Series One Z-series components in December 1994, Hl.17 No.12, p.185.-Ed.]

I gutted my reference system and cleared the racks and listening room to make way for the Spectral/Avalon/MIT equipment. A Billy Bags 5500- series rack held the new Spectral DMC20 Series 2 preamplifier, a Spectral SRD-2000 Pro digital processor (which I reviewed in Stereophile Vol.18 No.5, May 1995), and the MIT Z-Center and Z-Iso-Duo AC line conditioners. A Mark Levinson No.30.5 Reference CD Transport, also on the Billy Bags rack, fed the SDR-2000 Pro via a 1m run of MIT Digital Reference (RCA to RCA). A heavily modified Well Tempered Turntable (Mango Mat, Marigo Well Damped Arm Clamp, Marigo Isolation System and Motor Terminator, Marigo damping dots, and Lary Pederson's totally reworked Well Tempered Arm) played LPs. The cartridge was an AudioQuest AQ7000nsx, with a pair of WireWorld Gold Eclipse interconnects connecting the turntable to the DMC20's phono inputs. The LP front-end was mounted on a Merrill Stable Table. Vinyl accounted for about 70% of my listening time with the system.

A 35' pair of MIT MI-350 interconnects ran from the preamplifier to the Spectral DMA-180 power amplifier on the floor between and behind the loudspeakers. A lm pair of MI-350 connected the SDR-2000 Pro's single- ended outputs to the DMC-20 Series 2 preamp. A 12' pair of MIT MH-850 tri-wired loudspeaker cables connected the Radian HCs to the DMA-180 power amplifier. The entire system was connected with single-ended interconnects of the three companies' choosing.

The power amplifier was plugged into an MIT Z-Stabilizer AC line conditioner via MIT's Z-Cord II. Back at the rack, a Z-Center was plugged into the wall outlet, and a Z-Iso-Duo was plugged into the Z-Center for an additional stage of isolation. The preamplifier was powered from the Z-Iso-Duo's isolated bank, and the digital processor was plugged into the Z-Iso-Duo's second isolated outlets. The turntable and transport were powered from separate isolated banks from the Z-Center. All AC cords to the Z-Stabilizer, Z-Iso-Duo, and Z-Center were MIT's Z- Cords, and AC lines to the audio components were the Z-Cord II.

Fig 1: Spectral DMC-20, line-stage frequency response (right channel dashed, 0.5dB/vertical div.).

Fig 2: Spectral DMC-20, phono-stage RIAA error (right channel dashed, 0.5dB/vertical div.).

Fig 3: Spectral DMC-20, line-stage crosstalk (10dB/vertical div.).

Fig 4: Spectral DMC-20, line-stage THD+noise (%) vs frequency at IV into 100k ohms (right channel dashed).

Fig 5: Spectral DMC-20, set to unity gain, spectrum of 50Hz sinewave, DC-I kHz, at IV into 1001< ohms (linear frequency scale).

Fig 6: Spectral DMA-I80, MIT MH-850 cable, fre- quency response at I W into 8 ohms (top), 2W into 4 ohms (bottom) (right channel dashed, 0.5dB/vertical div.).

Fig 7: Spectral DMA- 180, small-signal I OkHz squarewave into 8 ohms.

Fig 8: Spectral DMA-I80, crosstalk: L-R (solid), R-L (dotted) (10dB/vertical div.).

Fig 9: Spectral DMA-180,THD+noise vs frequency at (from top to bottom at I 5kHz): 4W into 2 ohms, 2W into 4 ohms, and I W into 8 ohms (right channel dashed).

Fig 10: Spectral DMA-I80, spectrum of 50Hz sinewave, DC-I kHz, at 245W into simulated speaker load (linear frequency scale). Note that the third harmonic at 150Hz is the highest in level, at -76dB (about 0.015%).

Fig 11: Spectral DMA-I80, I kHz waveform (top); distortion and noise waveform with funda- mental notched out (bottom, not to scale).

Fig 12: Spectral DMA-I80, distortion (%) vs output power into (from bottom to top at IOW): 8 ohms, 4 ohms, and 2 ohms.

Fig 13: Power resistor, electrical impedance (solid) and phase (dashed) when connected by conventional coaxial cable (2 ohms/vertical div.).

Fig 14: Power resistor, electrical impedance (solid) and phase (dashed) when connected by MIT MH-850 cable (2 ohms/vertical div.).

Fig 15: Avalon Radian HC, electrical impedance (solid) and phase (dashed) when connected by MIT MH-850 cable (2 ohms/vertical div.).

Fig 16: Avalon Radian HC, electrical impedance (solid) and phase (dashed) when connected by conventional speaker cable (2 ohms/vertical div.).

Fig 17: Avalon Radian HC, anechoic response on midrange axis at 50", 37" from floor, averaged across 30 degrees horizontal window and corrected for microphone response, with complex sum of nearfield midrange and woofer responses plotted below 300Hz.

Acoustic Sciences Corporation (ASC) sent a slew of Tube Traps in case they were needed to augment the four ASC Tower Slims and two Tower Stouts I usually use in my room. A large quilt hangs on the wall behind the loudspeakers.

As with any high-performance technology, the Spectral/Avalon/MIT system needed some careful tuning to realize its full potential. I unpacked the components and did a rough setup so that the system could break-in. Avalon's Lucien Pichette and MIT's Joe Abrams arrived the next day to adjust the loudspeaker placement, tune the room acoustics, and set up. The following week, Avalon President and chief designer Neil Patel, MIT founder and designer Bruce Brisson, and Spectral designer Keith Johnson visited my listening room for more fine-tuning.

It was fascinating to watch these designers work and listen to the results of changes they made to the system throughout the day. I had just started to get used to the sound after Joe and Lucien left, and had a good basis for assessing the results of Neil's, Bruce's, and Keith's efforts.

First, Bruce totally rearranged the interconnect and cable routing, twisting the un-broken-in cable into a tight spiral. Bruce paid particular attention to keeping the interconnects away from loudspeaker cables and AC cords. Neil put Tiptoes under the DMA-180 amplifier, moved the preamp's power supply to the floor next to the rack, and installed felt-filled grilles on the Radian HCs (more on these grilles later). He also replaced the 15-amp fuses in the Z products with 20-amp fuses, something the owner's manuals say not to do.

They made minor changes in loudspeaker placement, primarily giving the Radian HCs a little more toe-in. Because the Radian's supporting cones aren't attached to the speaker bottom, the speaker will slide on the flat portion of the cone. This feature made adjustments much easier compared to spiked speakers that must be lifted off the carpet to be moved.

Most of the time spent tweaking was to add, remove, reposition, and rotate ASC Tube Traps. To my existing complement of four Tower Slims and two Tower Stouts we added one Studio Trap, two 36"-diameter Tube Traps, and one 52"-diameter Trap. Additional Tube Traps on hand were later removed when we realized the room was overdamped.

Although I've long been a fan of Tube Traps, they were particularly useful in setting up this system. Not only are the Tube Traps highly effective, but their versatility and seemingly infinite range of tuning (affected by rotating the trap between absorptive and reflective side out) made it possible to precisely control the room acoustics, and thus the some presentation.

If you buy this system, expect the dealer to install and tweak it to the level described here. The four dealers who sell this package are reportedly trained in these setup techniques.

Once the setup was complete, I left everything alone for the review period's duration. The interconnects and loudspeaker cables were left where they lay. According to Brisson, moving a cable disturbs the dielectric break-in. Every component in the system was continuously powered throughout the review period.

I must report a few glitches. One run of the 35' MI-350 was bad. It measured as having continuity with an ohmmeter, but wouldn't pass an AC signal. MIT sent a second set, which needed break- in (all the other cables and interconnects had been broken-in before being sent out). Second, one of the input jacks on the DMA-180 power amplifier worked intermittently; the review sample was a unit that Keith has used on location recording for the past five years.

I decided early on to evaluate the Spectral/Avalon/MIT system as a complete package, not as a collection of components. The overall sound the system produced in my room would be the judgment criterion, with an eye to the system's $72,000 price (including the $8495 Mark Levinson No.31 transport and LP front-end). Coincidentally, my usual reference system has almost the same retail price, including the Levinson transport and LP front-end.6

After the initial setup by Lucien Pichette and Joe Abrams, I had nearly a week to listen to the system before the second team arrived. The sound in my room was fabulous, with a soundstage that was considerably different from what I've heard before from reproduced music. Frankly, I didn't expect the additional tweaking by Neil Patel, Bruce Brisson, and Keith Johnson to significantly improve what I thought was already a great sound.

After the additional day of tweaking, the system went from sounding terrific to sounding utterly spectacular. All the special qualities I'd heard in the previous week's listening were heightened after the system had been fine-tuned and the room acoustics adjusted for the system.

I must start by describing the sound- staging, for that's the aspect of this system's presentation that is most unlike anything I've heard previously in reproduced music. The system threw a gigantic three-dimensional soundstage that was jaw-dropping. The sound- stage's width, transparency, focus, and image specificity were significantly better than any I've heard. The oft-used description of loudspeakers "disappearing into the soundstage" truly applied to the presentation in my listening room.

Not only did the soundstage extend way beyond the loudspeaker boundaries, but the presentation had a tremendous sense of height. Centrally placed images weren't just at loudspeaker level, but sometimes seemed to hang far above the loudspeakers. Michel Jonasz's voice on la fabuleuse histoire de Mister Swing 2292-42338-2, not my kind of music, but a spectacular recording) was decidedly above the loudspeaker plane. The soundstage's ability to extend up produced more a feeling of seeing the music through a huge picture window than through a narrow slot.

Similarly, the system had an uncanny ability to throw images far beyond the loudspeakers' lateral boundaries. I'm not just talking about hearing the recorded acoustic surrounding the presentation at the soundstage edges, but an ability to re-create tangible and solid images at the loudspeaker boundaries and beyond. It took some time to realize that I was hearing, for the first time, a soundstage in front of the loudspeakers that was rectangular, not triangular. Rather than present only centrally placed images in front of the loudspeakers, the system projected images in front of the loudspeakers all along the width, even to the soundstage's farthest edges.

A good example was the layers of percussion on the LP Cascades (Milestone M9109) by the Brazilian group Azymuth. The Spectral/Avalon/MIT system projected some of the percussion that was panned to the outside edges in front of the loudspeakers, with other layers of percussion directly behind the loudspeakers at the far left and right of the soundstage.

Another example was the finger snaps on the Michel Jonasz CD, which seemed to exist in space completely in dependently of the loudspeakers. An example of the system projecting images in front of the loudspeakers at the far left and right of the soundstage was the left-channel rhythm guitar and right-channel horn section on the tune "Wishing Well," from Michael Ruff's Speaking in Melodies (Sheffield CD-35). I heard rock-solid images beyond the loudspeaker boundaries that were just as tangible and real-sounding as the images thrown directly between the loudspeakers. Again, this aspect is unique in my experience.

This extraordinary imaging produced a sense of disbelief similar to what one feels when seeing a magician's illusion; you've seen it with your own eyes, but know that it defies explanation. My wife expressed a similar feeling after a few minutes in the sweet spot: "How do they do that?" she asked.

The soundstage also had an amazing degree of focus, spatial coherence, and image specificity. There was a sense of pinpoint precision both in image locations and image size. The system correctly portrayed image size over a wide range of instruments, from solo flute to the massive chorus in the recording of John Rutter's Requiem (Reference Recordings RR-57CD).

This stunning spatial presentation wasn't apparent on all recordings, only those that contained such information. The system was like a chameleon in its soundstaging, changing dramatically between recordings. Unlike many systems, the Spectral/Avalon/MIT package didn't impose a similar spatial rendering on all recordings. During my time with the system, I was constantly amazed at how revealing it was of the recording. The spatial relationships between instruments and the recorded acoustic were laid bare. Recordings I thought I knew (including purist recordings I'd engineered as well as those I'd made in 24-track studios) were resolved with a spatial precision that was revelatory. One jazz recording I'd made (which was named as one of the 10 best jazz records of the year by CD Review magazine a few years ago) was particularly interesting. I'd used Telefunken tubed mikes in an X-Y configuration over the drums (with only an additional kick-drum mike); I could hear perfectly through the Spectral/Avalon/ MIT system the exact placement of each drum and cymbal in the kit. The mounted toms seemed to emanate from exact points in space, with absolutely no connection to the pair of loudspeakers.

The remarkable soundstaging produced a sense of immersion in the music. It was like being inside the scene itself instead of looking through a picture window. If most systems produce a soundstage that's like looking at tropical fish in an aquarium, the Spectral/Avalon/MIT system was like scuba diving in the Great Barrier Reef.

Despite the system's amazing sound- staging, the ultimate sense of depth- of instruments and hall reflections far behind the loudspeakers-was less impressive than what I hear from the dipolar Genesis II.5s driven by tubed Audio Research VT150s. The Spectral/Avalon/ MIT system had greater width, projection, focus, clarity, and spatial coherence, but not the same ability to make the wall behind the loudspeakers disappear.

I've gone into this detailed description of the soundstage not because I'm a soundstage freak who values this aspect of performance over all else, but because the Spectral/Avalon/MIT system's spatial presentation was unlike any I've heard from reproduced music. Moreover, the astonishing soundstage was a significant factor in my enjoyment of music through the system. By so accurately and convincingly re-creating space, this system made it much easier to slip into the illusion and become more deeply involved in the music.

If the unique soundstaging had been all the system offered, I might have dismissed it as an auditory curiosity, and not regarded it as a vehicle for conveying the musical expression contained in a music collection. Fortunately, the soundstaging was just one aspect of the system's sonic performance that contributed to its musicality. In fact, the soundstaging was just the beginning.

The system excelled in two other important and related respects: transparency and resolution. Transparency is an audio system's ability to pass the musical signal without imposing its own character on the sound. The playback system should be as a clear pane of glass, providing an unadulterated view of the musical event. Departures from transparency can take many forms: tonal colorations, dynamic alterations of the original signal, a thickening of the sound, or timbres overlaid with a mechanical or synthetic character. Transparency also involves the ability to preserve fine detail, which is why transparency goes hand in hand with resolution.

I can say without hesitation that the Spectral/Avalon/MIT system had the greatest transparency and resolution I've heard in reproduced music. The system's clarity and ability to reveal low- level detail were staggering. Every night, as I pulled out record after record, I was amazed at how much information was in my music collection-information previously unresolved. This low- level information could be the fine inner structure of an instrument's timbre; its resolution made the instrument sound more real and less artificial. For example, the percussion instruments on the previously mentioned Cascades LP sounded less like transient impulses and more like wood being struck. Something in the way inner detail was resolved revealed the mechanism by which the sound was created.

This truth in timbre struck me as I listened to the superb new Oregon CD on Chesky (Chesky .ID130). Paul McCandless plays a variety of woodwinds (soprano sax, oboe, bass clarinet) that have been captured (by Bob Katz) with lifelike timbres. The Spectral/Avalon/ MIT system had an uncanny ability to present these instruments with a coherent and totally natural harmonic structure. Frankly, I didn't think this level of midrange clarity was possible with dynamic loudspeakers.

The system's overall tonal balance was flat and neutral, with a warm mid- bass balance, leanish lower bass, and lots of treble air and definition. Although the Radian HC's LF Q of 0.5 produced tight bass, the system had a good sense of power, impact, weight, and projection in the bass. The treble was extremely clean, quick, detailed, and resolving, without being overly bright. At my normal listening height of 36", the system lost a little bit of immediacy. Sitting on a cushion put my ears at 39", which slightly increased the sense of presence. The difference in tonal balance with listening height was much less severe than what I hear with the Genesis, which also needed a higher-than-normal listening axis.

The system challenged the assumption that accuracy and musicality are mutually exclusive. This accuracy vs. musicality argument suggests that an "accurate" audio system is by definition bright, edgy, and unpleasant to hear because, it is also suggested, that is how real musical instruments and the microphones used to capture them sound. A system that draws you into the music, provides many consecutive hours of fatigue-free listening, and generally communicates the music must therefore be euphonically colored. I found the Spectral/Avalon/MIT system uncompromising in its resolution, accuracy, and transparency, yet eminently musical and enjoyable. Although highly resolving, the system was never analytical or fatiguing.

Compared to the Genesis II.5s driven by the Audio Research VT150s, the Spectral/Avalon/MIT system had more warmth and body between about 80Hz and 200Hz, but less weight below 80Hz. Of course, the Radian HCs couldn't match the effortless extension in the bottom octaves provided by the servo-powered Genesis. I sometimes missed the Genesises' awesome extension. The organ on the previously mentioned Requiem disc, for example, didn't have the same depth or power from the HCs. Similarly, the bass-drum whacks on Trittico (Reference Recordings RR52CD) lacked the visceral impact and apparently unlimited dynamics of the Genesis. I'd characterize the Spectral/ Avalon/MIT system's bass extension as adequate rather than spectacular. Note that in a room with two solid sidewalls, the HC's bass will benefit from greater boundary reinforcement than I experienced in my new listening room, which is both asymmetrical and L-shaped.

However, I greatly enjoyed the Spectral/Avalon/MIT system's bass articulation, control, and lack of smearing. Acoustic and electric bass had a tautness and precision that revealed everything the bass player was doing. This LF definition was accompanied by a sense of power and weight that was particularly satisfying. Often, the tradeoff is between bass quality and quantity; the bottom end is either full and lacking detail, or precise and missing weight and power. The system seemed to offer both, with a slight emphasis on bass quality over quantity.

The bottom end also had a stunning dynamic agility. A kick drum's dynamic envelope was reproduced with a sudden attack and equally quick decay, with amazing tautness and lack of overhang. Moreover, the kick drum seemed to pop out of the presentation over the bass line rather than become smeared within it. The result was a satisfying rhythmic drive and cohesion.

This ability to resolve transient attack extended to the rest of the spectrum. The transients' leading edges seemed to line up, with their energy being projected all at once instead of slightly smeared over time. This quality gave the system a top-to-bottom dynamic coherence that made everything sound somehow "correct." Listen to the snare drum on the Michael Ruff disc; it had a tremendous sense of sudden impact and "pop" that infused the music with an upbeat and dynamic quality. This excellent dynamic resolution extended to revealing the music's fine dynamic structure.

Although the system played quite loudly without strain or congestion, the Genesis II.5s reproduced orchestral climaxes with a greater sense of effortlessness. The Spectral/Avalon/MIT system took on a slightly hard edge when pushed, but these limits were above what I would consider a normal listening level.

HDCD8-encoded recordings reproduced by the Spectral/Avalon/MIT system were stunning. The disc From the Age of Swing (Reference Recordings RR 59CD) had an openness, realness of timbre, transparency, and detail that were nothing short of amazing. The brass instruments' individual timbres were maintained during the ensemble playing, rather than degenerating into one big undifferentiated continuum. The background also had a "blackness" that created a greater contrast with the music, and produced a deeper silence between notes.

I've said enough about the system's specific sonic attributes. What I want to stress is how thrilling, compelling, and enjoyable music listening was through this extraordinary playback system. This past month was one of the most musically rewarding times I can remember. Playing records and CDs I thought I knew produced an exciting sense of discovery as I heard their musical nuances and expressiveness fully revealed for the first time. It was the kind of experience that made me pull out record after record into the night.

Finally, I should mention that I've heard this same system twice before, at the WCES in Las Vegas in January 1995, and at the Los Angeles Stereophile show the following April. My listening impressions were consistent for all three setups, although you can't really appreciate what this music system can do until you spend some late nights with it and your record collection.

Spectral DMC-20 Series 2: Starting with the DMC-20 Series 2 preamplifier, I measured an input impedance of nearly 13k ohms at the unbalanced line inputs, and 17k ohms at the balanced input. The 20's output impedance was 100 ohms unbalanced and 200 ohms balanced. DC offset levels were a low 1mV in both channels.

Maximum voltage gain from the line- stage was 19.4dB, an appropriate gain for a full-function preamplifier with a phono stage. Phono gain at 1kHz from phono input to main output was 53dB. With the AudioQuest AQ7000nsx catridge's 0.3mV output level, the system produced a normal listening volume with the DMC-20's level control about halfway up --just a little higher than that needed for CD replay. Note that the 20's balanced output will provide 6dB more gain overall.

The DMC-20's volume control had 0.55dB of tracking error between channels at the 9:00 position, 0.27dB at the 12:00 mark, and just 0.04dB channel imbalance with the volume control at the 3:00 setting. This level difference can be audible; I found myself using the balance control on some recordings.

Both the phono and line inputs had generous input-overload margins. The line input didn't overload at the Audio Precision System One's maximum 13V RMS output level, and the phono input overloaded at a high 180mV at 1kHz, far higher than any cartridge's output level. The DMC-20's output section didn't begin to clip until 31V RMS output, indicating the preamp's ability to swing high voltages.

The unweighted line-stage S/N ratio at unity gain referenced to 1V output was 82dB (left channel) and 79dB (right channel). These figures increased to 94dB (left channel) and 91dB (right channel) when A-weighted. The phono S/N ratio was 65dB in both channels unweighted, and 82dB with A-weighting.

The line-stage frequency response (fig.1) is ruler-flat. Indeed, the DMC-20 Series 2's response goes far beyond the Audio Precision System One's 200kHz bandwidth. Note the channel imbalance, however. The phono stage's RIAA accuracy (fig.2) was also flat, with less than ±0.15dB variation through the audioband. The treble is slightly shelved up in relation to the bass (about 0.2dB), which may contribute to the DMC20's lively sound with LPs. Many RIAA equalization stages are less accurate than that of the DMC-20 Series 2. For this measurement, I used the balance control to achieve left/right level matching.

The line-stage channel separation (fig.3) shows the typical 6dB/octave decrease in separation with frequency due to capacitive coupling. The left and right channels are so closely matched that the traces perfectly overlap. This channel separation is good, but nothing to write home about.

Looking next at the DMC-20 Series 2's distortion, I measured about 0.015% THD+Noise across the audioband (fig.4). This measurement is likely dominated by noise, not actual harmonic distortion. The DMC-20 Series 2's low level of distortion is confirmed by fig.5, an FFT-derived spectrum of the preamplifier's output while reproducing a 50Hz, 1V sinewave at unity gain. The 0dB reference level is 1V. The harmonic distortion components are extremely low in level and few in number. [Though the third and fifth harmonics are noticeable, they are still below the level of a 180Hz power-supply component.-Ed.]

Spectral DMA-180: Moving next to the DMA-180 power amplifier, all the measurements were made with the MIT MH-850 cable; cables without MIT's filter networks could damage the amplifier. The DMA-180 became quite warm after the one-hour-at-1/3-power conditioning period. The heatsinks were too hot to leave my hand on for more than a few seconds, but weren't that much hotter than when the amplifier is at idle, indicating that the output stage bias is set rather high.

The amplifier had an input impedance of 110k ohms, unbalanced, and a much lower 20k ohms, balanced. Output impedance measured a low 0.08 ohms across most of the band, increasing slightly to 0.11 ohms at 20kHz. The amplifier's voltage gain into an 8 ohm load was 26dB, and DC offset levels measured 10mV (left channel) and 19mV (right channel). There was some slow shifting of the DC levels (particularly in the left channel), suggesting the presence of a very-low-frequency noise component.

The DMA-180's frequency response at 1W into 8 ohms and 2W into 4 ohms is shown in fig.6. We usually measure frequency response to 50kHz, which I did with an 8 ohm load (upper pair of traces). For interest, I measured the 4 ohm response with a 200kHz bandwidth to look at the amplifier's wideband response. Although the amplifier is specified as having an intrinsic response to several megahertz (limited to 800kHz at the loudspeaker terminals), the terminating network in the MIT cables rolls off the response sooner, as fig.6 reveals.

Confirming the wide bandwidth, the small-signal 10kHz squarewave response (fig.7) shows a virtually perfect shape, with no overshoot or rounding of the edges. Indeed, the DMA-180's output looked just like the input signal.

Fig.8 shows the DMA-180's channel separation, which measured better than 70dB at 20kHz and met the 98dB channel-separation specification at 1kHz. The unweighted S/N ratio was 77dB (left channel) and 83dB (right channel), increasing to 91dB, A-weighted, in both channels. It's likely, therefore, that the left channel's noise is of a frequency attenuated by A-weighting— such as AC powerline noise.

The presence of an interfering 60Hz noise is confirmed by fig.9, a plot of the DMA-180's THD+N percentage at 1W into 8 ohms, 2W into 4 ohms, and 4W into 8 ohms. The apparent dip in distortion at 60Hz in the left channel (also in the right channel in the 4 ohm traces) means that the measurement is limited by the interfering signal.

THD+N measurements are made by sweeping the source oscillator driving the amplifier up the audioband, and simultaneously filtering the test signal from the amplifier's output with a swept notch filter. This technique removes the test signal, leaving only the amplifier's noise and distortion, which are then plotted as a function of frequency If an intruding noise is present (usually 60Hz AC hum, either intrinsic to the amplifier or introduced externally), that noise gets attenuated along with the test signal by the analyzer notch filter, and consequently shows up as a dip in the distortion trace at the interfering signal frequency.

I tried different grounding connections between the DMA-180 and System One, and also floated the AC ground on the DMA-180, but still got the slight noise. Overall, however, the THD+N was quite low, with only the 2 ohm curves rising appreciably above 0.02%, and then only at high frequencies.

Fig.10, an FFT-derived spectrum of the DMA-180's output while it was amplifying a 50Hz sinewave at 245W into 4 ohms (2/3 full power), shows low distortion levels. No distortion component rises above the -80dB level, making this one of the best-looking plots I've seen. The highest component is the second, followed by the third, then the fifth and seventh, all decreasing in level with increasing order. Fig.11 shows the DMA-180's small-signal distortion waveform (lower trace) with the 1kHz sinewave fundamental above it. The waveshape confirms both the low-order harmonic content and its low level (it is obscured by noise).

Fig.12 shows how the DMA-180's distortion/noise level changes with increasing output power into 8 ohms, 4 ohms, and 2 ohms, with one channel driven. If we define clipping as 1% THD, the DMA-180 clipped at 246W into 8 ohms (23.9dBW), well above the specified output power of 200Wpc. Into 4 ohms, the maximum output power was 372W (22.7dBW), somewhat shy of the specified 400W. With a 2 ohm load, the DMA-180 also fell short of the specified power, delivering 493W (20.9dBW). With both channels driven, the DMA-180's left and right channels clipped at 233W and 235W into 8 ohms (23.7dBW). With a 4 ohm load, the amplifier produced 355W (left channel) and 365W (right channel), or 22.5dBW and 22.6dBW respectively. While this power delivery fails to meet the amplifier's specification, this is still impressive performance, and suggests the DMA-180 will drive virtually any loudspeaker load.

During these maximum-output power tests (including that used to produce fig.12), the AC line voltage measured 117V (8 ohm testing), 116V (4 ohm testing), and 115V (2 ohm testing). Holding the AC line at a solid 120V would likely produce higher output- power numbers-we don't do this because it is not representative of the real- world situation in which amplifiers are used. Still, the DMA-180 is a powerhouse, and worthy of its "High Current" designation. Moreover, the amplifier was unfazed by the full-power testing, and never even blew a rail fuse.

MIT cable: Curiosity about MIT's terminating networks prompted me to measure, with and without the MITMH cable, the impedance magnitude and phase angle of the bench resistor we use as a load when testing power amplifiers. Fig.13 shows how the resistor's impedance changes with frequency when connected to the Audio Precision by a short piece of banana-plug-terminated coaxial cable. Because the resistor is almost a pure resistive impedance, with no capacitive or inductive reactance, its phase angle is 0° up to about 10kHz. The positive phase angle seen above 20kHz is due to a slight inductance.

For contrast, fig.14 is the same measurement, but with a run of MIT MH850. This cable has a higher inductance, seen as the positive phase angle at high frequencies. Whereas the coaxial cable introduced a phase angle of +19.6° at 200kHz, the MIT cable produced a phase shift of +27.5°. The MIT cable's inductance also affects the overall impedance, seen as the slight rise in the impedance magnitude curve above 150kHz. This doesn't tell us much about what's happening inside the MIT terminator networks, probably because we aren't looking at the phase angle over a wide enough bandwidth.

Avalon Radian HC: John Atkinson measured the Avalon Radian HC, providing me with its measured performance after the auditioning was completed. With one exception-see later the MH-850 cable was used exclusively when the Radian was measured.

The Radian HC's B-weighted sensitivity measured 85.5dB/W/m, a figure just below what we'd call average sensitivity. This measured sensitivity is 2.5dB lower than the claimed 88dB/W/m, but we have no information on how Avalon arrives at their figure. [We use the B-weighted figure because published research shows it to correlate very well with a loudspeaker's perceived loudness. -Ed.]

Fig.15 shows the HC's impedance magnitude (solid line) and phase angle (dotted line) made with the MIT speaker cable. The impedance is quite low through the bass, reaching a minimum of 3.5 ohms in the upper bass. The Radian HC's low-to-moderate sensitivity, combined with the lowish impedance through the bass, suggests a high- powered amplifier with the ability to deliver current into low-impedance loads as required. Of course, the HC was designed around such an amplifier, the Spectral DMA-180.

Fig.16 is the same measurement, but made with a generic spaced-pair loudspeaker cable in the Stereophile test lab. Despite the claimed abilities of the MIT cable to optimize the electrical phase behavior, the only differences are a very slight increase in the impedance magnitude peak at 5kHz, a somewhat lower impedance magnitude above 5kHz, and a very slightly more capacitive phase angle in the top two octaves.

The HC's FFT-derived quasi-anechoic response on the midrange axis (fig.17) shows a very flat, smooth response through most of the band, with a slight overall downward tilt in the treble. The dual woofers appear to be a little more sensitive than the midrange and tweeter, with a slightly elevated level apparent. This corresponds perfectly with my comparison above with the Genesis 11.5: " ... the Spectral/Avalon/MIT system had more warmth and body between about 80Hz and 200Hz, but less weight below 80Hz." Although the HC had a full midbass presentation, it was anything but thick and bloated. Indeed, one of the Radian HC's great strengths was its combination of bass quantity and quality. Other features in fig.17 worth noting are the 6dB-down point of 28Hz (referenced to the 1kHz level), and slight peaks at 750Hz and 1.2kHz.

For contrast, the Genesis 11.5 is anechoically flat to 16Hz [which gives an exaggerated low-bass output in all but very large rooms-Ed.]. Avalon claims an anechoic -3dB point of 34Hz, which is confirmed by fig.17. The curve's overall shape of a downward tilt in the treble didn't correlate with my impressions of an airy and detailed treble presentation. Many loudspeakers with a downward tilt seem to have just the right amount of treble. As J. Gordon Holt said many years ago, "Down with flat!"

The HC's horizontal dispersion (fig.18) is textbook perfect, with wide midrange and mid-treble dispersion, and a well-controlled rolloff of the uppermost frequencies with increasing off-axis angle. In JAs experience, loudspeakers with such good lateral dispersion coupled with a fundamentally flat on-axis response always produce excellent imaging and well-defined soundstages.

Fig 19:

The HC's vertical dispersion (not shown) reveals that the HC's tonal balance doesn't change appreciably with listening height as long as you sit with your ears between the midrange and tweeter axes. Sit or stand so you can see the cabinet top, and a suckout appears in the low treble. Sit lower than the midrange axis, and the treble tilts down a little more. I found a 36" listening axis a little low, and preferred a 39" axis, which increased the sense of immediacy and detail.

Looking next at the time-domain behavior, the step response (fig.19) reveals that despite its sloped-back baffle and even when connected with the MIT cable, the Radian HC is not time-coherent, at least on the midrange axis. The tweeter output is the first low-amplitude spike of energy to arrive at the microphone, followed by that of the midrange unit, then later the woofers. All the drivers are, however, wired with the same polarity.

Finally, fig.20 is the HC's cumulative spectral-decay or waterfall plot. The decay is extremely quick and clean, with almost no energy storage. This is exceptional performance, and correlates with my impressions of a clean, highly detailed presentation, with excellent transient performance. The ridge above the audioband is due to ultrasonic ringing from the tweeter, evidenced earlier as the peak at 26kHz in the speaker's frequency response.

From the measurements, it's easy to conclude that the Radian HC is a very well-engineered loudspeaker. The smooth response, superb dispersion, and clean waterfall plot point to excellent technical design.

Fig 18:

Living with the Spectral/Avalon/MIT system has been a watershed event for me as a music lover, audiophile, and reviewer. The system's resolution, transparency, and soundstaging set new standards in reproduced music. Knowing what this system can do has shifted my reference point in judging sound quality.

More important, I found it the most musically compelling and engaging system I've had in my home. I greatly enjoyed using the system for its highest purpose: communicating the music in my LP and CD collections. Every listening night was an odyssey of discovery as the system consistently uncovered newfound musical expression.

At roughly $72,000 for the system as configured (with the Mark Levinson No.31 transport and Well Tempered Turntable), the Spectral/Avalon/MIT package costs a lot of money by any standard. But in this day of $30,000 power amplifiers and $60,000 loudspeakers, $72,000 is toward the low end of the price scale for a truly world- class music playback system.

And world-class it is. I can't imagine someone looking for the ultimate in sound quality and musicality being disappointed by the performance of the Spectral/Avalon/MIT package. I must emphasize, however, that expert setup and tuning the room's acoustics are essential to achieving the level of performance I experienced.

There's only one unpleasant aspect of having this extraordinary system in my home: the day I pack it up for return to the manufacturers.

Note that the Spectral/Avalon/MIT system's price can be trimmed. You could start by getting the Radian HC in a standard finish, saving $3400. A less expensive transport (the system sounded great at the Stereophile show with a Theta Data would knock off another few thousand. MIT's less expensive cables, and less elaborate AC line treatment, would further cut the price. If you don't need a phono stage, the DMC-20 Series 2 is $700 less in the line-stage version. Scaling back the system could cut the price by about $13,000-$19,000 if you already own a turntable or don't need one.

©Stereophile-Vol.19 No.1

Fig 20:



We would like to extend our compliments to Bob Harley for undertaking the formidable task of reviewing the many components that comprise the Spectral/Avalon/MIT/2C3D system. The sheer length and attention to detail of his report clearly demonstrate the thoroughness of his evaluation.

Of all the positive comments in this article, there are two of which we are most proud. A system's ability to change "like a chameleon," depending upon the recording, requires a loudspeaker with minimal colorations of its own. Only a truly neutral transducer like the Radian HC has the ability to produce a three-dimensional holographic image through the presentation of information encoded in the program material. But most pleasing of all was Bob's impassioned rediscovery of his favorite music, late into the night. This is the result we are striving for, and the experience we should all demand from our audio systems.

We would also like to thank John Atkinson for his empirical corroboration of Bob's listening impressions. These measurements assist the audiophile in recognizing that products of this echelon are created with the innovative application of deliberate and rigorous research. Our goal at Avalon Acoustics will always be to bring the listener closer to the intent of the artist, and that essence which is the magic of music.

President, Avalon Acoustics


Our congratulations to Robert Harley for his thoughtful and comprehensive review of our Spectral Reference components in the context of the Spectral/MIT/Avalon music system. We particularly appreciate the special effort made to evaluate our components in the type of optimized system environment which they were designed for. It is no small challenge to set aside familiar references and consider new system approaches with sensitivity and an open mind. Mr. Harley's essay does this and more, addressing some of the most essential issues confronting the High End today, such as accuracy vs. musicality in sonic presentation. We are gratified that he has chosen the Spectral/MIT/Avalon system to illustrate the fact that truly "accurate" music reproduction is inherently musical.

The Spectral design team has always believed that the pursuit of accuracy to the source must take priority over any attempt to design "musicality" into the component itself. Adhering to this more challenging goal in component design led us early on to our "system" design philosophy. Taking into account the myriad factors occurring among audio electronics, cables, and transducers as an interactive system allows a high level of optimization to be pursued. The Spectral SDR-2000 processor, DMC-20 preamplifier, and DMA-180 amplifier, along with the MIT cable and treatment components, represent a highly optimized and refined electronics system. The combination of these components along with such fine transducers as the Avalon Radian HC results in a system that is more than just the sum of individual parts, but a finely tuned music "reproducing" system.

Mr. Harley's lavish praise of the Spectral/MIT/Avalon package serves to support the fundamental correctness of the "system" design approach. That he feels that he has experienced growth as a reviewer and rediscovery of his favorite music validates our many years of development efforts.

President, Spectral


We at MIT would like to thank Robert Harley first of all for his hospitality. He and his wife, Evalee, even put off their vacation to accommodate our crew for the second day of setup. We'd also like to thank RH for the thoroughness of his evaluation. This review left us breathless, not only because of its enthusiasm, but because it illuminated our own experiences when listening to this system.

As RH reported, the 2C3D (two- channel, three-dimensional) holographic system evolved out of the collaboration between Spectral and MIT's R&D teams, working together over many years. The underlying concept was to cooperate to create a linear system rather than try to correct nonlinearities from a previous stage. One evening in 1993, I commented to Spectral's Rick Fryer that we were beginning to get a true holographic image from the system.

We continued discussions that evening over dinner regarding the electrical criteria required to create a sonic hologram from two speakers. We agreed to complete the system in a three- to five-year period. By 1994 we were discussing criteria for speakers that would work with our products to create the 2C3D certified system. It wasn't long before Neil Patel of Avalon indicated that he could build a speaker that would make up the last part of our system.

About four months later Neil returned with the HC Radian, the first 2C3D-certified speaker. In this system, the HC Radian creates an image that not only has depth (to the rear of the speaker), but also (under exacting conditions) pushes well in front of the speakers, and in equal ratio outside their boundaries. This creates a large image around and in front (as well as to the rear) of the speakers, causing them to disappear within the music.

The MIT MH-850 CVTerminator Tri-Wire links the fast Spectral electronics to the individual crossovers of the HC Radians using a systems engineering approach incorporating novel technologies. Spectral's engineers provided MIT with the necessary test data on their amps and preamps. These electronics have extremely fast rise and fall times and require proper damping from the cables interfacing them. Not enough damping and the Spectral equipment will ring, adding artifacts to the music. Too much damping and the energy left over in the cable will cover up the silences between musical notes. Also, because the Spectral electronics have such a wide bandwidth, the MIT cables must condition the rolloff of the amplifier just right, or the amplifier can shut down.

Avalon provided MIT with a pair of HC Radians for measurement. Then we measured each of the HC Radians' three driver/crossover assemblies. Each assembly naturally has some dips and peaks in relationship to each other. This requires that MIT independently controls the amount of energy to each driver and crossover assembly. This was the criterion for the Avalon version of the MH-805 Tri-Wire. This was the final component designed for this cooperative effort which allows the system to create the 2C3D holographic effect.

As RH indicated in his review, setup is everything to this system. MIT has filed for a certification trademark for 2C3D, and is currently working with three certified US dealerships. By the end of January we should have five; by the end of 1996 we expect to have 8 to 10 certified US dealerships. Some of the requirements we look at before certification are: 1) rooms-construction techniques and dimensions are critical; 2) proper electricity and AC treatment, including the Z-System power-conditioning treatment; 3) room-tuning equipment, such as real-time analyzers and spl meters, to aid in tuning the room with the ASC Tube Traps; and, last but perhaps most important, 4) at least one qualified person in the dealership to undergo setup training with MIT. Only 2C3D-certified dealers with this level of commitment will be able to create the 2C3D holographic effect.

We believe that by year's end any consumer in the US will be able to find a dealer within a reasonable distance who can set up and demonstrate this system, and get it to perform as well as it did for Robert Harley. Again, we thank both RH and Stereophile for the thoroughness of their evaluation of the 2C3D system.

MIT-Music Interface Technologies

1 A fully balanced preamplifier must have identical signal paths between phases of the balanced signal. Any deviations in level, noise, or distortion between signal paths introduce non-common-mode (differential) components that are subsequently treated as part of the wanted signal.

2 The McCormack DNA-1 and DNA-0.5 also distribute the filter capacitors (but not the rectifiers) next to the output devices.

3 Brisson has invited me to MIT's laboratory in Auburn, California for an in-depth look at their test instruments and how they quantify cable behavior. Many aspects of cable behavior are, according to Brisson, easily demonstrable on MIT's test equipment. The tight scheduling of this review precluded a visit before this review went to print, but I will take him up on the offer and report my findings in a future article.

4 I would have thought that keeping the cable's capacitance and inductance as low as possible would minimize phase-angle swings, as it's the capacitive reactance and inductive reactances that introduce phase angles above and below 0'. Because the "phase angle" describes the load's capacitive and inductive reactance, a simple inductance-free and capacitance-free resistor has a phase angle of 0' at all frequencies. In a cable with minimum capacitance and inductance, the cable's resistance would dominate, giving it a benign phase angle. See figs.13 and 14 later for phase-angle measurements on a resistor with conventional cable and MIT cable.

5 In my second semester of college electronics, the lab portion of one class involved building an electronic project of the student's selection. I chose a project that my instructor thought was ambitious: a half-octave real-time spectrum analyzer. The project's plans had been published in the September 1977 issue of Popular Electronics. After considerable effort (and lots of learning along the way), I got the thing built and working. Many years later, when I became familiar with Richard Marsh's audio work, it struck me that this was the same Richard Marsh (along with Bob Jones) who had designed the spectrum analyzer and written the Popular Electronics article.

6 This system consists of the Well Tempered Turntable described earlier, Vendetta Research SCP2B phono preamplifier, Mark Levinson No.30.5 Reference Digital Processor, Sonic Frontiers SFL-2 Mk.II line-stage preamplifier, Audio Research VT150 power amplifiers, Genesis 11.5 loudspeakers, AudioQuest Lapis and Diamond X3 interconnects, WireWorld Gold Eclipse interconnects, Tice Power Block AC line conditioner, AudioQuest AC cords, and AudioQuest Dragon II loudspeaker cable.