Digital Processor Reviews

"Roberta Flack has Lou Gehrig's disease and can no longer sing," Béla, my visiting friend, yelled from the guest bedroom. The news rekindled in me the same mixture of sadness and foreboding I get every time I read about the decline or death of an iconic musician who rose to fame during my aeonian youth.

The next day, Béla announced he was about to use my Audeze LCD-X headphones and AudioQuest DragonFly Cobalt to play Flack's "The First Time Ever I Saw Your Face," which he'd ripped to his laptop. "Honey," I replied, "forget about your computer. Let's listen to this in the music room. I've seldom heard that song since the early '70s, when I played it through crappy components."

Soon we were seated in front of a system comprising a Stromtank S 2500 Quantum MKII battery-powered regenerator, Roon-equipped Innuos Statement NextGen music server, D'Agostino Momentum HD preamp and Progression M550 monoblocks, and Wilson Alexia V loudspeakers.

Instead of my reference Rossini Apex and Clock, I was using a dCS Vivaldi Apex DAC ($46,500), in for review. Together with a Vivaldi Upsampler Plus ($25,500) and Vivaldi Master Clock ($19,500), the Vivaldi Apex DAC had been singing in my system for close to three weeks. I thought I had a fair sense of what it can do.

Searching Roon revealed two remasterings of "The First Time Ever I Saw Your Face." The first, on The Very Best of Roberta Flack, issued in 2006, was streamable in 16/44.1 MQA. The second, on First Take, was released in 2020 and is available in 24/192 MQA. What self-respecting audiophile could resist comparing the two?

Within seconds after hitting play on the 2006 remaster, what I thought would be a lovely opportunity to wax nostalgic morphed into something far deeper. The first few bars of the song grabbed us like nothing else we'd listened to over the past 10 days. Flack's complete calm, unwavering focus, and unapologetic intimacy took our breath away. The soundstage was wide, the silence profound, the presentation pristine. The beauty of Flack's voice and passion, enhanced by John Pizzarelli's guitar, Ron Carter's bass, and Ray Lucas's drums, transformed the music room into a holy sanctuary. Toward the end of the first verse, right before "To the dark and the endless skies," I rose long enough to turn off the lights. We sat together in silence, barely breathing. When the song ended, the only words I could utter were, "Let's listen to the 2020 mastering."

The tighter and better-defined double bass, the mesmerizing delicacy of Pizzarelli's guitar, and fine detail that was lost in 2006 when the tape hiss was filtered out gripped us even more.

I thought I had already discovered everything the Vivaldi DAC could deliver. Yet here, on a simple track from which I expected little more than lovely singing, it had opened our hearts and transformed a visit from an old friend, dampened by dual COVID diagnoses, into a rare opportunity for spiritual communion. We sat in awe of music's magic.

A Vivaldi Apex trinity: DAC, Upsampler Plus, Master Clock
Released in 2012, the Vivaldi DAC was the first product in dCS's new line of two-channel D/A converters, which today includes the Vivaldi Apex, Rossini Apex, Bartók, and Lina headphone system. The Vivaldi was always intended as a state-of-the-art multibox system in which one box was devoted solely to D/A conversion; upsampling, streaming, file playback, and CD/SACD transport functions have always been relegated to other Vivaldi boxes.

I spent a very good year with the original Vivaldi DAC—I moaned when it went back to dCS—and had it long enough to review its major (2.0) upgrade. I vividly recall the differences between Rossini 2.0 and Vivaldi 2.0. The Vivaldi had a virtually boundless soundstage with weightier images and stronger, better-defined bass, and it delivered musical texture palpably. The trio arrangement of "Wachet auf, ruft uns die Stimme" ("Awake, calls the voice to us") on Yo-Yo Ma, Chris Thile, and Edgar Meyer's Bach Trios (24/96 MQA, Nonesuch/Tidal) left me marveling at the complex overtones of Ma's cello, the warmth and weight of Meyer's double bass, and the contrast between the initial pluck of a string and its mellower resonant decay. That timbral transition may not be audible distinctly in live music (footnote 1), but it is captured by the microphone. On a fine system, it can bring us closer to the music and the artist.

During my time with the pre-Apex Vivaldi DAC, I didn't have the Vivaldi upsampler, so I was limited to auditioning PCM files at their native resolution, while the Rossini upsampled them automatically to DXD or DSD: user's choice (footnote 2). Nonetheless, 16/44.1 on the Vivaldi sounded more detailed and involving than PCM upsampled to DXD on the Rossini; the Vivaldi's realism was uncanny.

This limitation was now removed, as John Giolas, dCS's vice president of sales and marketing, sent me the Vivaldi Upsampler Plus. I would finally be able to hear files at the same resolution via the Rossini Apex and the Vivaldi Apex system. The Apex revision had raised Rossini's bass response to Vivaldi levels; would the Vivaldi Apex's bass be even better? What other improvements might it offer? dCS also sent the Vivaldi Master Clock. I could have made do with the Rossini Clock—the sonic differences between the two are reportedly small—but pairing the Rossini Clock with the Vivaldi Apex DAC would have required changing clock cables every time I transitioned from 44.1kHz and its multiples to 48kHz and its multiples. Since I wanted to switch quickly between files so that impressions were still fresh, the Vivaldi Master Clock was essential.

Together, the two-piece Rossini Apex DAC/Clock and the three-piece Vivaldi Apex DAC/Clock/Upsampler filled my eight-shelf, double-width Grand Prix Monza rack to capacity. There was no room for a transport. No big deal. I haven't played silver discs at home for over a year. Why bother when I can stream and store millions of files and rip CDs losslessly? SACDs are still viable, whether in stereo or multichannel, but one can just as easily play native DSD64 files, or the PCM files from which many SACDs are derived. I see no reason to continue to acquire and play CDs beyond the desire to collect, feel, and hold—legitimate motivations but not musical ones. I would rather hear and feel the greater texture, color, nuance, and soundstage realized with hi-rez files.

Hales expounds on linearity
Considering the component prices of the Vivaldi Apex system, some readers may turn green with envy, red with anger, or some sickly amalgam of the two colors. Add the cost of the three aftermarket power cables, five or six clock cables, at least one set of dual AES3 cables, USB and Ethernet cables, and equipment supports and some readers may require medical intervention.

Others will note the identical specs for the Rossini Apex and Vivaldi Apex and be skeptical that they sound different. They do.

To find out why they do, and more, I Zoomed with Giolas, Managing Director David Steven, and Director of Product Development Chris Hales. We began with fundamentals. Rather than rehash the copious information about dCS's analog Ring DAC technology and Apex upgrade I included in our October 2022 Rossini Apex review, I asked the men to summarize what they thought is most important about the Apex hardware upgrade.

"The Ring DAC, Apex, and dCS's philosophy are about achieving linearity or neutrality," Hales said. "You can look at our goal in a number of different ways, but it's basically to reproduce the original recording as faithfully as physics permits.

"The whole Ring DAC architecture is very much about linearity, especially at low signal levels, where a lot of other DAC architectures lose out. That's where the signal starts fading away and you hear some unpleasant artifacts."

Hales began the Apex Project during the early stages of the COVID lockdown. Forced to work at home, he brought some test equipment with him to see what he could accomplish. Convinced that dCS had done as much as it could via software updates to improve linearity, he decided to consider possible limitations in dCS's analog circuitry. He never had a specific product in mind—he did not visualize Apex in his head and then try to figure out how to get there—but when he thought he'd found something worth pursuing, he put it on a board so that everyone could listen.

The resulting upgrade, which included reconfiguring and enhancing many components in the main Ring DAC, adjusting component layout on the Ring DAC circuit board, and installing an all-new analog output board, significantly lowered the noisefloor and reduced second-harmonic distortion by more than 12dB.

"With every product that we make, our team's objective is to create a sound design that measures as we need it to be," Steven said. "Then we listen to it. During [what became] the Apex process, Chris developed three or four prototypes. Each time, we listened to see if we connected to the music more. That's what we're talking about when we contrast Vivaldi to Rossini. As hard as it is to explain, Vivaldi seems to have more emotional impact."

For my Rossini Apex review, Giolas said, "The linearity of our DACs is so [much higher than] the industry norm, we've had to create our own test equipment to measure it." I asked Hales if John Atkinson would be able to detect what he and his team measured during the Apex project.

"I believe he has the Audio Precision 555, which is the best machine on the market," he replied. "Still, some of the harmonics we examine remain below the residual of that machine. So, he can get closer than most people, but he still won't be able to measure the whole story.

"A DAC receives a digital code and outputs an analog voltage. Whatever the output voltage is for a code of 1, we'd expect the output for a code of 2 to be exactly double and the output for a code of 4 to double again. Ideally, if you were to plot the output voltage for each input code on a graph, all the points would fall on a straight line; they would be linear. In reality, however, the points won't form a perfectly straight line. Depending on the system, the points may form a curve of some sort, or the line may have jumps in it or one or more points that are out of line.

"The most immediate effect of this deviation is that it adds harmonics to a signal, effectively altering its timbre. Imagine dragging your fingernail back and forth across the face of a perfectly flat mirror. If it is perfectly flat, your nail glides smoothly over the mirror and makes no sound. But if there's a scratch on the surface, every time your finger goes over the scratch, it makes a noise. For one back-and-forth movement, you get two clicks, at twice the frequency of your back-and-forth movement. So, a second harmonic has been created.

"Things get even worse when there are two signal frequencies, as there almost always are with music. This nonlinearity creates both harmonics (i.e., signals at exact multiples of the input frequency) and signals at the sum and difference of the two frequencies. This is really bad news! Musical instruments generate their own harmonics in exact multiples of their fundamental frequency—it's one of the ways we differentiate one instrument from another—so any harmonics generated by nonlinearity are at least musically plausible, whereas sum and difference frequencies most definitely aren't. Consider a note at 100Hz and another a perfect fifth above it, at 150Hz. Here, the sum frequency caused by nonlinearity would be at 250Hz, which is neither a multiple of 100Hz nor of 150Hz, and so would sound quite unmusical.

"There's a whole lot more to nonlinearity. Some cases sound much worse than others. The D/A process is particularly susceptible to some of the most objectionable. Regardless, the basic principle holds: Nonlinearity will create frequency components not present in the original material. Nonlinearity is something anyone who strives for neutrality in their equipment will try to avoid."

Multichassis yet holistic
Hales surprised me by saying that while the DAC core of the Vivaldi and Rossini Apex is "remarkably similar"—they use the same analog board—Vivaldi's digital system is "completely different." The Vivaldi Apex contains an earlier generation FPGA (field programmable gate array), a highly flexible logic device. Although Vivaldi's FPGA requires more supporting electronics and firmware, the Vivaldi DAC has more space internal than the one-piece Rossini (which incorporates streaming and upsampling functions). This allows greater flexibility in transformer positioning, component isolation, and what can be done with I/O and the control board itself. Moreover, the Vivaldi's control board is bigger than the Rossini's, allowing critical components to be more distant from "noisy things."

"Vivaldi's hardware represents a much more ambitious approach to D/A conversion than the Rossini's digital processing platform," Giolas said. "Vivaldi's multichassis approach, with its different power supplies, enables us to dedicate more real estate to the task at hand, exercise far more control over the interactions of different parts of the circuitry, and perform more sophisticated processing in the D-to-D portion of the upsampler. The Vivaldi DAC can do more in the analog domain and more signal management. Dozens and dozens of little things add up to the performance advantage of Vivaldi over Rossini."

"What's unique about dCS is our very holistic approach to electronics," Hales said. "A lot of tech guys will say, 'Oh yeah, ... it's all about the power supply or speed or current or whatever.' But it's not. It's not about any one device; it's about everything. You need everything working as well as you can make it for a product to perform optimally. So, rather than focus on one particular thing, we try to make everything as good as we can."

Steven elaborated on that holistic approach in the design of the Vivaldi and Rossini Apex DACs. "The challenge was to translate the technology in Vivaldi's four boxes into Rossini's much smaller footprint, driven by a single FPGA, without constraining the product or dialing it down. Vivaldi's multibox architecture gave us a bit more freedom with positioning and isolation. Chris was able to filter out a lot of the things that happen when you have lots of PCBs doing different jobs in a single box. This is why the DACs sound similar yet different."

Acknowledging that Vivaldi is a more complex product, Hales said, "As much as I'm a great believer in keeping things simple, to do something better you sometimes have to make it more complicated. The Ring DAC is a perfect example of this approach. No one designed the Ring DAC because they wanted to make it simple and elegant; we created the Ring DAC because it is palpably better than other approaches. That's where the sophistication comes in. If something needs to be more complicated for a better result, that's what we do. We don't just take the line of least resistance, however attractive that often is."

In summation, Steven said, "If you trace the performance of the initial Vivaldi system to where we are now, there's light years of difference. I'm really proud that we're able to offer Apex to existing dCS owners. It would have been really easy to spin out the upgrade as a new product with a new name, but we feel strongly that people invest in dCS because they intend to own it for a long time. So, if we can offer upgrades to existing units and support customers on their musical journey, that's best. We hope to roll out Apex in Bartók in 2023 as well."

---

Footnote 1: Often it is, though. It depends on the performance (including the size of the ensemble and the complexity of the music), the venue, and where you are sitting.

Footnote 2: There are two exceptions to the Rossini Apex DAC's and the Vivaldi Upsampler's automatic upsampling of file resolution (to DXD, DSD, or DSDx2, according to user preference). DSD remains in the same resolution as the original file, and MQA files are never upsampled beyond their ultimate MQA resolution due to stipulations in MQA's protocol. I have always favored DXD upsampling over the two other options. This remains unchanged with the advent of Apex. What has changed, after much coaxing from John Giolas, is my preference for the Map 3 mapping protocol over Map 1. I've also returned to the 2V output setting because D'Agostino has determined that it works best with my reference Momentum HD preamplifier.