Sonic Frontiers Transport 3 CD transport & Processor 3 D/A processor Page 2
Processor 3: stretching the limit
The Sonic Frontiers Processor 3 is a two-box affair featuring a fully differential, balanced topology in both the analog and digital domains. The large main chassis is attached to the smaller (though still hefty) outboard power supply via a 6½' detachable cable. Like the T-3's, the P-3's front panel is easy to read from a distance, displaying locked sample frequency and selected digital inputs.
The power supply is a beefy, straightforward design with a large potted transformer feeding six stages of regulation within the power-supply chassis. Several of these primary regulators are, in turn, cascaded to other stages within the main unit for a total of 14 regulators serving the processor. All but two on the analog output board use Linear Technologies' low-noise version of the popular 317/337 three-pin devices.
The main chassis contains three large circuit boards partially overlapping one another above the floor of the unit, as well as a fourth, vertical board assigned to front-panel duties. Of prime importance for the prospective buyer is the manner in which these boards are laid out. With the emergence of higher-resolution digital formats, I feel it would be folly to plop down a large sum for a processor that wasn't easily updateable, regardless of how good it makes CDs sound. SF obviously had this in mind when designing the P-3, and has carefully configured the principal boards to anticipate new developments. For example, the front-panel board already supports display of 88.2kHz and 96kHz sample rates.
The digital input board contains six interface options, selectable via front-panel buttons, including counterparts to all of those found on the T-3, plus TosLink. All input signals, other than that from the I2S-E interface, are routed through an AES21-C002 encapsulated input receiver for recovery of the clock and data.
Also on this board is the now-ubiquitous PMD-100 HDCD digital filter, but with a wrinkle. Much has been made of the 6dB of attenuation required by Pacific Microsonics to be applied to all non-HDCD discs for level matching with HDCD-encoded software. A DAC designer can choose to implement this signal reduction in either the analog or digital domains, though each technique has potential compromises. The P-3 I reviewed bypassed this attenuation altogether, and did so while eliminating any extra resistors or relays in the signal path. A simple jumper is located just before the PMD-100 chip. Therefore, if you find it annoying to have to adjust the volume on your preamp when switching between HDCD and non-HDCD discs, you can easily engage the specified gain-scaling for regular CDs, in this case applied in the digital domain with a resultant 1-bit loss of resolution.
The second and largest circuit board runs across the entire front of the processor and is made of Arlon, an expensive Teflon/Fiberglas hybrid said to be ideal for delicate high-speed signal-processing applications. Chris Johnson says that it is one of the largest Arlon boards ever made. The other two dual-sided boards are of FR-4 glass epoxy.
Four separate stages of regulation are found just prior to the two dual 20-bit UltraAnalog D20400-A DAC modules, providing a discrete stage of isolation and 5V power-supply conversion for each internal DAC, each of which, in turn, serves a separate channel phase. These DAC modules are UltraAnalog's best, but with a major enhancement applied within the P-3. Most other versions of the D20400-A DAC incorporate the critical current-to-voltage (I/V) conversion circuitry within the DAC module, usually a high-speed op-amp from companies such as Analog Devices or Burr-Brown. In this instance, UA designed an outboard I/V conversion module made up of some 75 discrete surface-mount components per channel. According to Chris Johnson, "this novel circuit has significantly better low-level IM-distortion artifacts than their previous integrated I/V stages." Johnson also credited much of the P-3's reported ultra-low noise floor to the new I/V design.
After voltage conversion, the signal goes through analog reconstruction via a discrete, third-order, passive LCR filter for each channel phase. It's a point of pride for Johnson that SF was able to implement this new reconstruction filter with minimal high-frequency attenuation at 20kHz by using a variation of a Chebyshev design.
The final board of interest is chock full of discrete circuitry supporting the four 6922 dual-triode tubes—double the tube count of the SFD-2 Mk.II—that form the heart of the processor's analog output stage. Each channel phase has one dual triode with its internal sections connected in parallel, effectively resulting in eight tubes within four bottles for a 3dB reduction in noise and a halving of the output impedance. Furthermore, the output stage, a cathode follower with solid-state current sources, eschews the coupling caps of SF's previous processors. Instead, it employs a direct-coupled design via surface-mounted servos on the bottom of the board, resulting in a largely frequency-independent output impedance, as first seen in SF's highly touted SFCD-1 CD player. These servos are accompanied by a "failsafe" circuit that ensures that no DC can pass to the outputs in the event of servo failure. UltraAnalog's Richard Powers said that "you can even yank a tube while the unit's on without any DC surprises." Rather than test this claim, I took his word on it, and suggest you do as well!
The P-3 also has a shunted mute relay that remains out of the signal path, and a second-order output low-pass filter to attenuate RFI from riding down the interconnects. (By the way, if your preamp lacks the ability to set different gain levels for various sources, be very careful when switching to the P-3 from any other input. The unit outputs a healthy 3V from its unbalanced RCA jacks, and a whopping 6V from its pair of balanced XLR connectors!)
All in all, a very impressive piece of engineering, and ample proof that the SF/UA team pulled out all the stops to build the quietest and best-sounding digital processor they could, regardless of price.
Means & Methods
My experiences with the Processor 3, followed much later by the Transport 3, took place in two distinct phases. The initial stage began in the fall of 1997, after an early-production P-3 arrived at my Honolulu apartment. At the time I was in the midst of an extended sabbatical from these pages as I oversaw the design and construction of my new home, into which I finally moved this past May. (My empathy is extended to all who've been though that mind-numbing experience.)
Eager as I was to experience the P-3 in its best light, I immediately connected its 13W3-I2S-E connector to a Muse Model 5 transport. Unfortunately, I could get no signal lock, even though the Muse unit was equipped with a fully functional 13W3-I2S output. (SF itself had not yet released a transport with an I2S-E option.) My dismay increased after I discovered, through phone calls and e-mail to both companies, that each was using a different and incompatible version of the I2S transmission interface! (See Sidebar, "I2S: Two Don't Always Tango.")
Despite this early setback, I had no problem at all adjusting to the P-3's extreme clarity, dynamic dexterity, and wide-band performance when using the AES/EBU and S/PDIF interfaces from either the Muse Model 5 transport or a modified Theta Data II. In any event, while I was steeped in house-building, most of my time with the P-3 was spent simply enjoying a vast array of music rather than in-depth analytical listening.
Phase Two began after I'd installed the astounding new quad-amplified Audio Artistry Beethoven Grand speaker system, with its all-active crossover, in the substantially larger listening room of my new home. This house has background noise somewhere on the order of Grant's Tomb, and, combined with the remarkable resolution and openness of these speakers, revealed subtle sonic variations between components that were simply buried at my previous digs. Accordingly, my discussion of performance issues will be restricted to my most recent experiences of the products in my new home. The picture was completed a few days after HI-FI '98 in Los Angeles, when the Transport 3 arrived and I could get down to business.
Despite the overall convergence of absolute sound quality between most competently designed transports and processors, sonic differences do exist between otherwise well-made top-tier gear. However, these differences tend to be very small. With this understanding, you can interpret my often liberal use of analogy in the proper context as I attempt to convey those portions of my experience that will most likely be translatable to your own. Therefore, while the following characteristics are shared to one degree or another by a number of other front-line digital components I've heard, Sonic Frontiers' Transport 3 and Processor 3 stood out from this select crowd in the areas of performance outlined below.
Sonic Delight: the Processor 3
The overall sound of Sonic Frontiers' Processor 3, using one of its standard non-I2S interfaces, was actually closer to that of SF's SFCD-1 (Stereophile, Vol.19 No.6) than to the SFD-2 Mk.II. The latter model had two principal shortcomings in its otherwise excellent performance: I often found its bass rendition a bit too bloomy and diffused, and its perspective was a little forward for my taste—even though that forwardness often made for some spectacular dynamics and rhythmic drive.
In contrast, the P-3 not only exceeded the large-scale dynamic articulation of the SFD-2 Mk.II, but did so with notable refinements in low-level dynamic resolution, and with a more relaxed, though not too distant, spatial perspective closer to that of the SFCD-1. I suspect that this fundamental improvement owes much to the P-3's ultra-low noise floor. This ability to so effortlessly paint each instrument with just the right amount of dynamic shading within the larger context of the tempo's ebb and flow was a hallmark of the P-3, and often made listening to even mundane CDs an involving experience.