dCS DeBussy to dCS Rossini: A comparison review of two DACs Page 2

I decided to reach out to Quick and get the company line on several aspects of the Debussy and Rossini, here is our Q&A:

dCS Rossini DAC Q&A

Rafe Arnott: Having lived with the Debussy for a couple of months, receiving the Rossini and listening to the two back-to-back revealed a significant step up in the realism to tone, timbre, resolution and musical drive, to say nothing of the increase in overall cohesiveness to the playback – regardless of genre. Considering the jump in price, this was in-line with the expectations I had cultivated before hearing the two DACs. Can you talk about the major circuit-topology, transformers, materials, components and chassis differences between the Debussy and the Rossini?

John Quick: “Well, the differences between Debussy and Rossini are pretty big, and they should be; aside from the Rossini DAC being twice the price of Debussy, there are major differences in the hardware and what software we can run on each model.”  

“Debussy represents the distillation of the best features and technologies we developed for our previous-generation processing platform and RingDAC analog board. This platform was originally developed for our previous flagship, Scarlatti, in 2006, and the same processing core and analog board were used in the four-box Scarlatti and Paganini systems, the Puccini CD/SACD Player/DAC, and the Debussy DAC. The Scarlatti and Paganini separates benefitted from multiple power supplies and processing cores, each dedicated to a critical part of the digital playback chain; whereas, in Debussy (and Puccini, for that matter), everything is/was contained in one chassis, meaning everything shares one power supply and one processing core.”  

“Rossini also represents the distillation of the best features and technologies developed for our third-generation digital playback reference system: the four-box Vivaldi, launched in late-2012.  Compared to the previous generation (that includes Debussy), for generation three we redesigned and vastly improved the processing platform, the RingDAC analog board, the power supplies, chassis construction… pretty much everything across the board. In our view – taking into careful consideration what customers are looking for today and what we project they will want for the foreseeable future – Rossini needed to replace both Paganini and Puccini from our previous generation.”

“Working toward that goal, Rossini shares an identical RingDAC analog board with the Vivaldi DAC; network hardware identical to that found in the Vivaldi Upsampler; and twin power supplies like Vivaldi (and in the Debussy’s generation, Scarlatti) DAC; however, unlike trickle-down models derived from our previous flagship, we developed a new, dedicated processing-platform for Rossini that would have enough grunt and storage capacity to allow the single chassis to do more and perform better than any one-box dCS in the past.”

RA: dCS utilizes a significant amount of proprietary technology in their designs, with a production model that seems to see the upper echelon of components implementing and benefitting from the latest innovations of the company’s constant research and development first (clock architecture, FPGA fine-tuning, digital processing platform advances, UPnP software improvements) and the entry-level models seeming to benefit later in the form of “trickle-down” updates, many of which dCS provides free of charge through firmware updates. Is it safe to say the company sees this as a matter of fact, rather than a favor to owners of older, or lower-echelon models?

JQ: “What you describe is pretty much the way we’ve always developed products, and definitely more just how we do things rather than a favour to owners at either end of the affordability spectrum. As far as we’re concerned, the launch of any new dCS product is the start of its product development lifecycle, not the end. Each new generation of dCS products (Vivaldi, Rossini, and Bartok are generation three) begins with a flagship that needs to be at the cutting-edge of state-of-the-art performance; needs to accommodate everything we’ve learned and developed to date; AND needs to have significant processing headroom and storage capacity to allow us to add features and make performance improvements as we continue ongoing R&D.  We scale our flagship back from there in the least compromised way we know how. In some cases, less expensive models actually get new features before our flagship does, but in general because what much of what we do is determined by software, we have great flexibility to continue to offer upgrades over the course of a product’s life.”

RA: Upsampling, clock integrity to minimize ‘jitter’ in digital signal pathways and isolation of delicate analog and digital circuitry seem to play key roles in the overall ethos of dCS’s approach to music reproduction. Can you talk about the company’s core values when it comes not just to fidelity of playback, but to what dCS stands for as a respected entity in the hi-fi industry?

JQ: “I guess you could say dCS is a company of agnostic and objective problem solvers. When it comes to engineering in general, despite what some say, there really is no ‘no compromise’ solution; rather the best know what to do to find a balance with the least compromising trade-offs. Present our engineering team with challenges to improve performance, to get the most out of PCM, DSD, or MQA… and dCS will find the least compromised way to achieve state-of-the-art performance.”

RA: The dCS Network Bridge and Rossini apps have a very similar look, feel and intuitive performance and interface. How important is software app development to the company when there are elephants in the room like Roon in the software segment of the market? Was it key to dCS to design and implement their own app from the ground up as opposed to letting another company do any of the heavy lifting software-wise? And was it because of a difference in how dCS sees the streaming or local library browsing experience? Was it a holistic pursuit for the software to reflect the hardware ethos?

JQ: “From the moment we decided to include a network input on the Vivaldi Upsampler in 2012 we knew that as much as its performance had to be world-class, the interface was equally as important to get right.  We began work on a custom interface then, and we continue to improve and adapt each of the dCS Apps today.  For the same reason we were eager to add support for Roon in 2015-16; it is a world-class interface, and we’re proud to offer dCS owners both options.”

RA: What in layman's terms is the RingDAC™, is it merely a proprietary Trademark name for a dCS researched and developed R2R DAC? What sets it apart from other designs on the market?

JQ: “The dCS RingDAC design and name is patented and trademarked, but it’s actually something very different from a R2R ‘ladder’ DAC.”

“R2R DACs of any design have great intentions to offer less compromise over using off-the-shelf DAC chips that are, by far and large in today’s world, designed for low-power-consuming portable consumer electronic products. As much as that seems like a genuinely wonderful goal, R2R DACs are, in fact, highly flawed.”

The first problem with R2R DACs is that they start off life best-intentioned, but highly imperfect. An ideal R2R DAC requires increasingly inapproachable resistor values that move further from ideal the more complex it gets: the required accuracy of each resistor doubles with each additional bit added to the architecture. Some manufacturers resort to physical modification (“trimming”) of the resistors to get their ideal values closer; others resort to software to make the actual value of each resistor in the ladder closer to the ideal value when processing its contribution to the DAC’s output. The problem with either/any approach to an R2R DAC is that component values change with age and temperature, and this causes processing errors associated to this normal process that are CORRELATED to the music signal; this means that WHEN a component in a R2R DAC changes, there is ALWAYS distortion in the SAME PLACE in the signal, and at the SAME LEVEL.”

“This type of distortion crushes low-level detail, causes high intermodulation distortion (i.e. the ability to simultaneously play two or more frequencies/harmonics that are near, but not at the same frequency is mutilated), and in general guarantees the DAC will sound and measure worse over time.”

The dCS RingDAC is made up of a FPGA-centric central processing core and the RingDAC™ analog board. All incoming digital signals first run through proprietary digital filters (written expressly for each sample frequency supported) that are then oversampled to a unique 5-bit synchronous format in the FPGA processing core. This signal is then presented to the RingDAC analog board through a mono pair of FPGAs that run a proprietary algorithm that “maps” the 5-bit signal to 32 of 48 equally-weighted current sources (per channel) at a rate of 2.822-6.14 million times per second, depending on model/setting and the incoming sample frequency.”

“The summed low-level signal from the output of the current sources is mixed, filtered for out-of-band switching noise, and then amplified by a discrete Class-A output stage to yield an output of 2V or 6V, with an output impedance of under one-Ohm. All stages of the conversion process are known and planned for, meaning as the signal flows from one stage of the process to the next, a minimal number of compromises are made – and most importantly: any variance in parts values/tolerances over time and temperature cause errors that are DECORRELATED to the music signal and are filtered out by the filter/mix stages as noise. This means that, over time, the RingDAC will provide consistent, exemplary performance because the usual suspects that contribute to DAC performance degradation are dealt with up front, by design. There is nothing else on the market that does this, and for this, the reason nearly 30 years later, is why we still base all our designs around the RingDAC.

Data Conversion Systems Ltd
Unit 1 Buckingway Business Park, Swavesey, Cambridgeshire CB24 4AE, United Kingdom
+44 (0)1954 233950

Ali's picture

Thanks for the article, I wonder if you have trying Rossini directly connected to amp and bypasding pre?

Rafe Arnott's picture
If you read this review on the McIntosh MC611 I wrote last week, I get into how I tried several DACs directly hooked up to my mono blocks and via the preamplifier and found I preferred all the DACs through the C2600 pre.