The Fifth Element Page 2
Repeatedly pressing Status scrolls through displays showing: the Input selected and that input's sampling rate; a resettable log of digital "overs" for the Left and Right channels; and the M1's current operating temperature, in degrees Celsius or Fahrenheit. Pressing Filter shows the number of the filter selected and enables selecting a different one. The Aux button selects between the Aux jack's accepting a digital-audio input or an external word-clock input. The Display button controls the brightness of the display, and Enter executes whatever selection is displayed.
The M1's exterior design is uncluttered, calm, and assured. The front panel is black with white lettering and red displays. The case top, bottom, and sides are a silvery champagne-like color and are milled from solid billet, though not as one unit. Ventilation slots are cut into the top and sides, through which can be seen red LEDs on the various printed-circuit boards, even when the unit is set to Standby. Fit and finish are first-class. The M1 does not give up to Mark Levinson or Lexicon anything in the sweepstakes of design and build quality. Indeed, if you squint, perhaps you can even see a little of that industrial-design DNA. I can't recall any component with as satiny a front-panel finish.
The layout of the rear panel is symmetrical and neat. The M1 is dual-mono, with the (filtered) IEC power inlet, master power switch, and digital input jacks in the center, and the left and right analog audio outputs to the left and right sides. The M1 has both RCA single-ended outputs and fully balanced (true dual-differential) outputs on XLR jacks (pin 2 hot). The single-ended and balanced outputs are separately buffered and isolated. Furthermore, each balanced output has a recessed screw by which its output level can be adjusted in a range from +8 to +22dBm. Of the four digital inputs, all but the optical are transformer-isolated, each with its own high-quality transformer. (The optical input is electrically self-isolating by its very nature.)
Brian Zolner says that the design brief for the M1 was that it should sound "fast," "revealing," "open," and "spacious." He says that the two years of development after they already had a working DAC circuit were devoted to arriving at creative solutions to the inherent problems of "Red Book" CD playbacksolutions that would not involve just putting someone else's off-the-shelf answer in a new chassis with a new name on the faceplate.
Zolner thinks that the fundamental problems of "Red Book" playback were for the most part avoidable, but that those chances were thrown away decades ago. He believes that the 44.1kHz sampling rate was chosen for reasons of convenience, in that Sony already had a huge installed base of its U-matic ¾" broadcast video recorders. Once the Sony pro video recorder was chosen as the backbone of the recording system, the 44.1kHz sampling rate was inevitable, given the other choices that had already been made. (Other examples of expedience in the creation of the Sony-Philips digital-audio system: The S/PDIF standard relies on ubiquitous and inexpensive 75-ohm coaxial video cable, and the ubiquitous and inexpensive RCA phono plug.)
Sony's original professional digital-recording setup consisted of the PCM-F1 processor, and a professional video recorder as the storage device. Zolner believes that if the sampling rate had instead been 64kHz, things would have been much better long ago. Even the Stockham Soundstream system's 50kHz sampling rate would have been better.
Despite all that, Zolner claims that there really are "enough" data in "Red Book" audioif everything else is as good as it can be. According to him, the chief problems in using "Red Book" data are: the challenges of digital-filter design (given the necessary narrowness of the stopband between the highest frequency to be reproduced and the even-higher frequencies that must be cut off), the presence of jitter, and how those first two factors impact analog-stage design, with the most important goal being slew rate or risetime.
Thinking Inside the Box
For its overall design and engineering approaches, Bricasti relies on "Get rid of all that other stuff" and "The sound is only as good as the sound of the worst component part." Bricasti's M1 DAC is not in a features competition. The concern was that each additional feature would add complication in areas such as power supply and grounding, detracting from pure audio performance. So there is no volume control (digital or analog), no headphone jack, and no remote control. Nor does the M1 convert sample rates (though it does oversample).
Although the 24-bit chip the M1 is based on, the Analog Devices AD1955, can handle DSD data, the M1 is a PCM-only converter. Zolner believes that the number of potential users (especially consumers) with access to a genuine DSD datastream is vanishingly small. Likewise, the M1 lacks a USB input, because Bricasti regards USB primarily as a convenience feature for casual listeners. (And in many jurisdictions, serious listeners can acquire Halide Design's USB-to-S/PDIF converter for less than the sales tax on the M1.) The M1 does have selectable digital filters, but its 24-bit oversampling is fixed and nondefeatable.
Bricasti paid attention to the usual details, such as clean printed-circuit layouts, good grounding schemes, and high-quality parts. The M1's circuit boards are made from Arlon, a material said to be more flexible than fiberglass, and therefore more self-damping of vibrations.
The first major engineering challenge Bricasti addressed was jitter. The means chosen was to use the "coolest" approach to clocking, and to implement it with the shortest physical distance between the clock source and each digital chipset. The "coolest" approach, according to Zolner, is Direct Digital Synthesis (DDS) of a phase-locked-loop (PLL) clock signal, which places the clock-signal generator only millimeters from the Analog Devices DAC chip.
Because the Bricasti M1 is a dual-mono design, each channel has its own AD1955 stereo DAC, operated in differential-output mono for better dynamic range. A separate circuit board holds a large SHARC chip that performs such housekeeping tasks as synchronizing the clocks, and providing for the selectable digital filters and the display.
Next was digital-filter design. Bricasti chose the AD1955 chip not so much for its own sound, but rather because it can accept an external clock signal and, more important, outsource its digital-filter function. The M1 provides a total of seven digital filters. Filter 0 is the stock filter inside the AD1955 chip. Filters 16, of Bricasti's own devising, reside on the SHARC chip and include low-delay and half-band designs (though not a "minimum phase" option). The owner's manual states that these filters are constructed at very high frequencies, then downconverted to be applied to the digital-audio signal.
The last major engineering challenge Bricasti faced was the design of the analog stage. The M1 has three separate power supplies with multiple stages of regulation: one for the digital board, and one for each dual-mono analog board. The analog-output section is all discrete, and of Bricasti's own design. Zolner believes that many DACs have slower-than-optimal risetimes because a fast risetime makes the unpleasant sound of a jittery signal even more obvious.
Although Bricasti does not publish a jitter measurement for the M1, Zolner claims that once you get jitter vanishingly low, you can speed up the circuit's slew rate with no ill effects. Bricasti claims a slew rate of 200V/µs (measured before the output buffer). Bricasti claims the resulting improved transient response makes the soundstage deeper and wider, and the harmonic structure of the music more coherent. Dialing in the final recipe was an iterative process simultaneously involving several variables. At that depth of design, says Zolner, "There is no 'absolute.' It's all about 'deltas.' You give something up to get something else. All that matters is whether, at the end of the day, you have won or lost."
Stupendous Nordost Silver Shadow
I did most of my listening with a 2m length of Nordost's truly exceptional Silver Shadow S/PDIF cable, terminated with RCA plugs ($750). The Silver Shadow is a light, thin, flexible micro-monofilament cable for which Nordost claims an impedance out-of-tolerance of less than 1%. One of the challenges of making an S/PDIF cable is approximating as closely as possible the ideal loading of 75 ohms, so that there are no (or the smallest possible number of) internal reflections to corrupt the timing of the digital datastream.
Without question, $750 for an S/PDIF cable is a thrust far into the territory of diminishing returns. But the Silver Shadow justifies its existence with an amazingly easeful and detailed sound, more authoritative bass, and a far deeper soundstage than my cheap'n'cheerful (ca $20), commercial-quality, true-75-ohm S/PDIF cable from broadcast supply house Markertek.
I primarily listened with the Vivid Audio B-1 loudspeakers (since sent on to JA for full review in the October 2011 issue), Luxman's MQ-88 40Wpc tubed stereo power amp, and Luxman's C-600f solid-state preamplifier, all hooked up with Cardas Clear signal and speaker cables. Because my early-production sample of Luxman's D-05 did not have an S/PDIF output (TosLink only), S/PDIF sources were a borrowed Meridian Sooloos Control:15 music server (heavenly) and Musical Fidelity's M3 CD player ($1500).