dCS Bartók DAC with Headphone Amplifier Review
Editor’s Note: Having realized that online privacy is officially dead, @austinpop has decided to write his articles under his own name, Rajiv Arora, in the byline. Rajiv tells us more about himself in the bio at the end of the article. Having read it, I now understand some of his crazy interests!
dCS (Data Conversion Systems Ltd.) has a long history as a high-end digital audio brand, with products like the Vivaldi and Rossini, built on their FPGA-based Ring DAC™ platform. While these products have garnered much admiration and critical acclaim, they inhabit the upper end of the price spectrum. The Bartók is a significant step for dCS, bringing elements of its Rossini network streaming DAC to a (relatively) more approachable price point, while breaking new ground with a bespoke class A headphone amplifier.
I’ll be honest - I approached this review with trepidation. We are, after all, talking about a component that costs $15,000 MSRP! When I started writing reviews on Audiophile Style, I wanted to write about components that I myself would be interested in buying as part of my audio journey. Did I envision writing about a $15k DAC? No I didn’t.
But the dCS Bartók is not just a DAC. Fully configured, it is an entire music system. It is a DAC, a preamp, a streaming music endpoint, and a headphone amplifier. As such, it could replace a digital streaming headphone system comprising multiple boxes with one elegant solution. The prime target consumer - although by no means the only one - is a high-end headphone enthusiast who consumes digital audio from a variety of local and remote sources.
Which brings us to me. I am a high-end headphone enthusiast who consumes digital audio from a variety of local and remote sources. As a reviewer, I’ve written extensively about DACs and high-end headphones. As a forum participant, I am the steward of a long-running thread here on Audiophile Style on optimizing the digital streaming chain. Viewed in this light, how could I not review this device!
And so it came to pass that one sunny summer afternoon, my review unit was hand-delivered to me by a smiling John Quick, General Manager, dCS Americas. Other manufacturers might ship out a unit for review, but not dCS. John wanted to personally ensure the unit was fully operational and optimally configured in my setup. We had budgeted a whole afternoon for setup, but partly because of my geeky and admittedly OCD preparation, and partly due to the Bartók’s simple and intuitive interface via the Mosaic iOS app, it took no more than 10 minutes for us to be up and running. We spent the rest of the afternoon alternating between listening to music, and audio geek-talk, where John filled me in on dCS’ approach and history building digital gear.
Technology and Features
dCS has decades of experience in D/A conversion, upsampling, digital filtering, power supplies, clocking, isolation, analog stages, and amplifiers. Like other DACs that I’ve found interesting, they eschew chips and implement digital logic on FPGAs.
The Ring DAC™
The centerpiece of all dCS DACs, from the entry-level Bartók to the flagship Vivaldi, is the Ring DAC™ platform. This platform is both proprietary and unique, so what follows is just a high-level overview.
The platform comprises a Control Board and an Analog Board. The input stream first traverses the Control Board, which applies the user’s selection of upsampling and digital filters to it, followed by noise shaping to render to an intermediate format for D/A conversion in the Analog board. This intermediate RingDAC format is a 5-bit, 2.8224 or 3.072 MHz stream, based on the sample rate family of the input.
The Analog Board contains a pool of 48 equally (unitary)-weighted current sources per channel that are switched at a rate corresponding to the incoming data. The key component in the Analog Board, and a critical element of the entire architecture, is the RingDAC mapper. For each 5-bit sample, the mapper selects a different subset of the 48 unitary current sources, with the number of sources corresponding to the sample value. dCS stress that this is more than just a pseudo-random selector, incorporating additional selection criteria to minimize noise and distortion. Indeed, on the upper-range Rossini and Vivaldi, users can select between multiple mappers. The outputs of these current sources are summed and filtered by discrete analog components, and this signal is then amplified by an electronically floating, discrete, fully-differential Class-A transistor output stage.
dCS developed the RingDAC mapper as a way to overcome the deficiencies of conventional DACs. According to dCS’ Chris Hales:
“...ladder DACs are inherently prone to non-linearity, but Delta-Sigma types are prone to problems caused by timing errors and switching noise. The Ring DAC occupies the “sweet spot” where the benefits of high speed and multi-bit intersects. Note that even with the tightest practical tolerances, a ladder DAC cannot come close to the linearity of the RingDAC.”
With ladder DACs, resistance values can drift over time. This leads to nonlinearities that produce harmonic distortion. While unitary current sources can also drift from their specification over time, this is what the Mapper mitigates. Since a different subset is chosen for every sample, the noise caused by drift is uncorrelated to the signal, resulting in much lower distortion.
I asked John Quick and Chris Hales about other aspects of the technology, and here are their responses.
Rajiv Arora: What is dCS' design approach on clocks - both internal and external clocks?
John Quick: Our clocking architecture is designed to be flexible nearly as much as it is to be accurate. All our clock circuits have always been discrete and employ top grade VCXO crystal oscillators, one specifically for each base rate family, 44.1 and 48. The crystals are graded against a reference under temperature extremes at the factory in an environmental chamber, where any variances are recorded and then individual crystals are matched and optimized with software. In our DACs the clocking circuit can act as the master reference to a connected component (master mode), it can be told to prioritize an external reference (wordclock mode), or it can run regular S/PDIF, S-DIF, or AES-EBU signals through a dual PLL to clean up the signal’s incoming timing reference (audio mode).
We began to employ external master clocks due to our experience in the pro arena, where there are generally numerous clocks running simultaneously, thereby making their use mandatory; though our engineers questioned their efficacy in relatively closed consumer systems, there is both objective/measured and subjective/listening evidence that they can have a profound effect on the sound. The theory why the dedicated, external reference makes a difference has to do with the fact crystal oscillators are electro-mechanical devices, so they are especially sensitive to their physical and electro-magnetic environment. It may be the hammer method, and you can absolutely spoil the performance of an external reference with crummy cables or improper support, but the performance gains are there to be had!
Rajiv Arora: What is dCS' design approach with power supplies?
John Quick: The PSUs in all our products since we introduced Scarlatti in 2007 have been a combination of linear and switching supplies. After the over-specified mains AC transformers, we establish DC PSU rails for all device sections requiring dedicated, regulated supplies with individual, proprietary switching power supplies that are all synchronized together (and to the audio clock!) for the lowest noise possible. Local regulation of all analog circuitry is then done with linear PSUs - this includes the RingDAC analog board AND the separate HP amp in the BHD version of the Bartók DAC. In this way we’re able to utilize the strengths of both PSU approaches: linear (low noise, fast transient response), and SMPS (low-power consumption, wide input range while maintaining efficiency).
Rajiv Arora: What is dCS’ approach to input isolation?
John Quick: Around 2014 the design team spent a significant amount of time analyzing why the USB and Ethernet inputs on the first-generation Vivaldi Upsampler sounded different from the traditional S/PDIF and AES-EBU inputs (i.e. a dCS transport playing into the Upsampler was significantly better than external sources playing the same, bit-perfect file).
Using FFT analysis they determined the nature of the noise that was making its way in from the various computer peripheral generic PSUs and less-than-ideal grounding schemes. The team experimented with various means of isolating our gear from these bugaboos (optical, electrical, and galvanic), and they settled on a combination of mostly electrical isolation components that first went into Rossini upon release in 2015, then Vivaldi Upsampler and DAC v2.00. As a result, the network, USB, and AES inputs are all effectively galvanically isolated by design.
Rajiv Arora: What makes the Bartók’s headphone amplifier special?
Chris Hales: What is special about the headphone stage: It starts with the dCS analogue output stage and maintains that level of performance with true class A operation through balanced and unbalanced outputs. Care has been taken to ensure that it’s optimized to support any pair of headphones regardless of efficiency or impedance. High Impedance headphones require lots of voltage but not much current, Low impedance headphones vice versa, so to cope with both the amplifier needs to be able to deliver lots of voltage and lots of current (though not simultaneously) cleanly.
Having decided that we wanted the amplifier to work in Class A, the trick was how to achieve this without ridiculous amounts of standing heat, which was achieved by the biasing arrangement. As expected, the amplifiers have very low output impedance to fully control, particularly low impedance, headphones. Each Channel has two identical amplifiers driven in opposite phases. With Balanced headphones they’re both used, with unbalanced only one of them.
The power supply for the headphone section is completely independent from the other power supplies and consists of two switching regulators, both synced to the clock, of course. Additional linear regulators supply the common mode-rejecting input stage and another set for the amplifier driver stage. The amplifier is essentially floating from the rest of the system ground to ensure no current is fed back into the DAC ground to cause problems and for EMC reasons.
A fortuitous benefit of operating in class A is that the current load on the power supply is constant, irrespective of drive/load so there is no signal related ripple on the power supplies which improves linearity and crosstalk.
Incidentally, the amplifier has DC protection to avoid damage to headphones in the case of a fault or “dangerous” input conditions, current limiting to protect against short circuit loads and thermal protection to protect the amplifier and transformer against abnormal load/drive conditions.
The Bartók implements network streaming using a hardware module called the Mosaic Processor. Mosaic is dCS’ product name for a collection of hardware and software modules that provide a unified way to access digital music, and to control the product. I’ll discuss features and the control software in later sections. dCS use a modified Stream Unlimited S800 module to implement the Mosaic Processor. Since 2017, dCS have been writing their own firmware for this module, which has allowed them to optimize the sound quality and reliability.
The Bartók’s chassis is built from aerospace-grade machined aluminum, wrapped over an internal folded metal frame. A lot of attention has gone into mechanical isolation. dCS use internal acoustic damping panels to reduce sound-degrading mechanical vibration and magnetic effects.
The full specifications for the Bartók can be found on the product page here. Here are some I want to highlight:
- Ethernet and USB inputs: up to 24/384 PCM, up to DSD128 (DoP)
- 2x AES/EBU and S/PDIF (coax, BNC) inputs: up to 24/192 PCM, up to DSD64 (DoP)
- Dual AES/EBU mode (for dCS components): up to 24/384 PCM, up to DSD128 (DoP or dCS-encrypted DSD)
- S/PDIF (Toslink) inputs: up to 24/192 PCM
- 2x word clock inputs, 1x word clock output (BNC)
- DXD or DSD upsampling, choice of 6 PCM filters and 4 DSD filters
- Full decoding and rendering on USB and Ethernet inputs
- Rendering only on other inputs
- Single-ended RCA and balanced XLR analog outputs
- Full-resolution volume control, outputs can drive power amplifiers directly
- Balanced 4-pin XLR and single-ended 6.3mm outputs
- 1.4W RMS into 33Ω, 0.15W RMS into 300Ω at the single ended (SE) output
- Balanced output is 4x power output of SE
- Output impedance (balanced output): 0.1Ω
Configurable output levels for full-scale input:
- 0.2, 0.6, 2 or 6V RMS on analog outputs
- 0, -10, -20, -30dB on headphone outputs
- Control either via front touch panel, or a Mosaic control app on iOS and Android
Glossary of Terms
There are no standards for terms when it comes to music streaming. I’ve adhered to a consistent convention throughout this review. Here are the terms I’ll be using:
- Music computer: the hardware device on which music player or server software runs.
- Music player: monolithic software applications that combine library management and output to the DAC all in one. Examples are Stylus, or Roon Server.
- Music server: for distributed software, the library management and streaming source. Examples are: Roon Core, minimServer UPnP server.
- Renderer or endpoint: the distributed software component that receives music streams from a music server, and outputs to the DAC. Examples are Mosaic UPnP renderer, Roon Bridge (a.k.a. Roon Ready endpoint).
Form and Function
The Bartók makes its presence felt from the moment you unbox it. At 36.8 lbs and 17.5” x 17.0” x 4.6,” diminutive it is not! Despite its size and its weight, the chassis’s high quality is immediately evident. The silver finish on the “aerospace-grade machined aluminium” panels is immaculate, and while the chassis is an unassuming cuboidal box, it’s obvious that a lot of attention has gone into mechanical isolation. This thing is built like a tank, and everything - from the knob to the buttons, the connectors and ports - screams quality.
The rectangular high-resolution digital display on the front left fascia has a clean modern look that I found very attractive. Next on the front panel is a row of small, elegant buttons for control. I did not use these buttons very much, but when I did, the clarity and resolution of the display was outstanding. On the far right are headphone outputs - both single-ended 6.3mm and balanced 4-pin XLR - and a volume knob.
I primarily used the dCS Mosaic Control app (available on iOS and Android) on my iDevices. dCS released this app at the Munich High End show, which was just a few weeks prior to my receiving the review unit. I’ve seen some really clunky apps with streamers and streaming DACs before. This is the exact opposite. The Mosaic Control app has a similar minimalist, modern user interface as the physical controls, and did not have a single error or glitch during operation. Given it had only been out for a month when I used it, this was impressive. This app serves multiple purposes, as the following screenshots illustrate.
It provides access to the controls on the unit, far more conveniently than the front panel buttons.
It controls the Mosaic processor on the unit, allowing you to configure streaming sources, and navigate and play music.
The Mosaic processor is chock full of features. It implements a UPnP renderer, a Roon Ready endpoint, as well as streaming clients for Tidal, Qobuz, Deezer, Spotify, and more. Also notably, the underlying Stream Unlimited module supports streams up to 24/384 PCM and DSD128 over DoP. This is a welcome change from the ConversDigital modules used in so many other DACs, that annoyingly constrain Ethernet inputs to 24/192. No such limitation here!
Indeed, the Mosaic processor is so full featured that there are multiple ways to get at the same content over the network. Consider my library of local music, which resides on a Synology NAS. I run a UPnP server (minimServer) on my NAS. With the Bartók, I could access my library in a variety of ways. Here’s a few:
- Using the UPnP controller in the Mosaic app to navigate minimServer on my NAS, and direct it to stream a selected track to the UPnP renderer on the Mosaic processor on the Bartók.
- Using Audirvana on a Mac, accessing my library through file sharing (SMB), and directing the playback stream to the UPnP renderer on the Mosaic processor.
- Using the Roon app on my iPad to direct the Roon Core on my music computer, accessing my library through SMB, to stream music to the Roon Ready endpoint running on the Mosaic processor.
Similarly, Qobuz or Tidal content could be delivered either via Roon, or directly with the built-in Qobuz/Tidal clients in the Mosaic processor.
As I found, these delivery mechanisms all had subtly different sound quality too, so this made for a lot of listening experiments.
With a product as capable and multifaceted as the Bartók, I found it best to first lay out the key questions that my review was trying to answer. Proceeding from that, I then developed an evaluation plan, or testing strategy, to answer these questions.
The key questions that the Bartók raised for me were:
- How does it perform as a traditional DAC?
- How does performance compare over my inputs of interest: USB vs. Ethernet?
- How much does the quality of the upstream chain help or harm its performance?
- How does the internal headphone amp compare to my reference Cavalli Liquid Gold amp?
- How does the internal headphone amp perform with various high-end headphones across the spectrum of impedance and sensitivity?
- How does performance vary with music software and protocols?
- How does the Bartók compare to other DACs - from cheaper, to similarly priced, to more expensive?
Review System Topology
To aid the reader’s understanding of the various configurations tested in this review - which can become quickly confusing with a product this functionally capable - I’ve shown below a pictorial representation of the two systems used in this review. The first is my own headphone-based system, which I consider the primary, as this is where the bulk of my listening was done.
The second is my friend Eric’s system. Eric lives in the same city as I, and he shares my passion for high-end audio enough to graciously allow me to test gear in his system. This is incredibly useful, as not only does it allow me to experience the components under review in a speaker based system, it allows me to experience them in another system, period.
To enable you to listen to the same tracks that I did, I have created a public playlist on Qobuz USA. This playlist includes the tracks mentioned in this review, as well as others I listened to in the course of this evaluation. Please note that in some cases, the Qobuz track will only stream at 16/44.1, whereas I may have used a local hi-res version. Still, this gives you a sense for the music I listened to for evaluation.
You’ll notice that many, if not most, of the tracks are not audiophile-grade demo-quality recordings. I don’t buy DACs only to listen to demo tracks. Most users buy gear to listen to the music they love, so this is how gear should be evaluated, in my opinion.
From the first few notes, it was obvious this was going to be an enjoyable review. Music through the Bartók had a rich, expansive, solid, and natural character. I initially listened on my baseline configuration, where the Bartók was driven via USB from my music computer running Euphony OS, in ramroot mode, using the Stylus music player. I used the balanced headphone output driving the Empyrean.
A few days after John installed the unit, my college-going daughter came by for the weekend and wandered into my listening room. After remarking on how the Bartók was the “biggest-ass” piece of gear she’d seen, she said: “Ooh, we should listen to Alaska on this!” Maggie Rogers (Now that the light is fading, Capitol, 16/44.1) isn’t an artist I’d know if I didn’t have young daughters. A few seconds into the song, her eyes widened, and she got “the look” we audiophiles know. I could hear why. Like most indie pop songs, this wasn’t exactly an audiophile recording, but the dynamics were electric, and the somewhat woolly, thudding rhythm came across with satisfying intensity. Rogers’ moody vocals and her high notes were clear and natural.
Suryodaya (Yarlung Records, DSD128) is an Indian classical fusion album by Robert Gupta (of LA Philharmonic and Street Symphony fame) and Badal Roy, a tabla virtuoso. This album was recorded on analog tape and mastered to DSD. On Raga Redux, Roy’s tabla builds from whisper soft to a rollicking crescendo. The Bartók’s black background and low noise floor tracked these dynamics wonderfully. The texture and sonority of Gupta’s Stradivarius came across with impressive realism.
I did a lot more listening to the Bartók over the next few weeks, which just confirmed my initial impressions of its character - big, dynamic, open and natural. It had that elusive ability to pull you in and connect to the music in a deeply emotional way.
Of course, at this price point, outstanding and excellent sonics are just table stakes. To really gauge the quality of the Bartók, I needed to compare it to my reference DACs on hand. I’ll get into comparisons in a later section. First, I explored the sound quality in the many operational modes the Bartók supports.
dCS provide a choice of DXD and DSD upsampling, and filters F1 through F6 for PCM content, and F1 through F4 for DSD content. Which of these sound best really boils down to personal preference. For my ears, I preferred DXD upsampling, F2 for PCM, and F1 for DSD.
I also confirmed that the Bartók can fully decode MQA content from Tidal. By default, MQA streams automatically engage the M1 filter, but unusually, and unlike most other MQA DACs, the Bartók allows the user to pick a PCM filter of their own choice, even with MQA content.
Headphone outputs: balanced vs. single-ended
I compared the single-ended and balanced headphone outputs of the Bartók. This was a no-brainer - the balanced output sounds significantly better. dCS tell me that the balanced output can deliver 600mW RMS into 300Ω, compared to the 150mW of the SE output. How did that translate to sound quality?
The main area of improvement was in dynamics, but other areas like image size and tonal richness were improved too. I strongly recommend balanced cables for headphones with the Bartók.
A Perfect tool: the Transparent Ultra headphone cable kit
To aid in all the headphone switching, I recently received, and was fortunate to have on hand, the Transparent Audio Ultra headphone cable kit. Transparent sent this to me to aid in my reviews, and I have to say, this kit was invaluable!
The Transparent Ultra cable system comprises up to three elements - the 3m cable itself, to which are attached headphone leads with the appropriate terminations for each headphone via a Transparent-designed custom coupler, and on the other end if needed, an adapter from 4-pin XLR to a variety of amplifier terminations. What this allowed me to do was to equip all my headphones with their own leads, and then attach the one I wanted to listen to to the coupler in a matter of seconds. Perfect for the audiophile with multiple headphones! I could discern no auditory degradation from the couplers in the chain, and the sound quality of this cable is better than any of the other cables I’ve owned.
Listening with different headphones
High-end headphones can be finicky about the amplifiers driving them. Since one of the value propositions of the Bartók is to drive headphones directly, without the need of an external headphone amp, I tested the Bartók with 3 headphones in my possession, each with its own quirks:
- HD800 (SD mod) - 300Ω, 97 db estimated (1mW/1kHz). The HD800 is known to be temperamental with amps.
- Meze Empyrean - 32Ω, 100 dB (1mW/1kHz). The Empyrean is easy to drive, but fell just outside the threshold of recommended minimum impedance (33Ω). dCS assured me this would not be an issue, and in fact the output impedance of the balanced output is a mere 0.1Ω.
- Abyss AB-1266 Phi CC - 42Ω, 88 dB (1mW/1kHz). The Abyss has much lower sensitivity and so can be difficult to drive.
With all that setup, the listening results were anticlimactic - in a good way. The Bartók was simply unflappable. It drove all 3 of the headphones with panache and without quirks. I could drive even the current-hungry Abyss to the loudest I wanted with plenty of gain to spare.
Comparison with Cavalli Liquid Gold external headphone amp
Now that I’d established the Bartók could drive all of my headphones with ease, the next question was how did it compare to an outstanding dedicated headphone amp like my Cavalli Liquid Gold?
Osmo Vänskä’s album Sibelius: Kullervo - Kortekangas: Migrations (BIS, 24/96) has a less-common version of Finlandia with male chorus. On this track, the difference between these amps was small. The Cavalli amp portrayed a bigger soundstage, better dynamics, sweeter mids and more natural voices. On the other hand, the Bartók’s internal amp had better transparency, more focus, clearer micro-details, and quicker transients. There seemed to be a real benefit to having the amp be adjacent to the analog output stage of the DAC.
In one use case, the Bartók internal amp was clearly better, and that was with the Abyss headphones. As I reported in my High-End Headphone Roundup: Meze Empyrean, Focal Utopia, Abyss AB-1266 Phi CC, the Cavalli did not drive the Abyss particularly well. Not so with the Bartók, which brought out the best qualities of the Abyss - tonal neutrality and a powerful low-end.
I actually value the transparency and coherency of the internal amp highly, although there is no denying the additional oomph and dynamics of the Cavalli with the HD800 and Empyrean. All in all, I would call this a wash, and declare the Bartók’s internal amp to be outstanding, and the Cavalli’s equal. Of course, there will always be better and more-expensive headphone amps, but you would need to get into the very high end of headphone amps to surpass the Bartók’s internal amp. The fact that this amp adds just $1500 to the price of the DAC-only configuration makes it an absolute bargain.
Effect of the Upstream Chain - USB
DAC manufacturers, especially as we go up in the price spectrum, work diligently to render their DACs immune to the barrage of noise flowing from the upstream chain. As described in the Technology section, dCS falls into this category.
Which brings us to the question: how much does optimizing the upstream matter to the Bartók? In this section, I’ll evaluate the USB input. In the next, I’ll look at the Ethernet input.
The listening experiment was to compare 2 digital chains, both using Roon Server as the music player software:
- The optimized chain shown in the system topology diagram, and
- An unoptimized chain comprising a MacBook Pro laptop with mains-connected SMPS charger, driving the the Bartók directly with an AmazonBasics USB cable.
Short answer: the optimized chain sounded better. However, the magnitude of the improvement was much smaller than I’ve heard with other DACs I’ve reviewed in my system.
The improvements in the optimized chain are similar to what I’ve heard with previous DACs: a reduction in harshness and glare, a more holographic (bigger, wider, deeper) image, an increase in clarity and detail, and deeper and better articulated bass.
Clearly, dCS’ efforts to isolate noise at the inputs has paid dividends. Yet, it is not 100% immune to the upstream, simply because that is an incredibly difficult problem to solve. What it does mean is that the Bartók sounds excellent, even without heroic upstream USB optimizations.
Effect of the Upstream Chain - Ethernet
This listening experiment through the Bartók’s Ethernet input, compares (please refer to the system topology diagram):
- The optimized Ethernet path, connecting the Bartók’s Ethernet input to a port on the SOtM switch, with
- The unoptimized Ethernet path, connecting the Bartók’s Ethernet input directly to my router, using a generic Cat 6a cable.
Here again, the results were consistent with my USB optimization findings. The improvements with the optimized Ethernet path via the SOtM switch were: less harshness, a bigger image, more clarity, and better bass articulation.
But again: the magnitude of the improvement from the optimized path was quite modest, and smaller than what I have heard with other Ethernet DACs.
The conclusion with Ethernet is the same as with USB: the Bartók sounds excellent, even without heroic upstream optimizations. Of course, any optimizations you do make will still help sound quality!
Effect of music software
Given that the Bartók supports so many modes of operation, did the choice of music software matter to sound quality? I looked at this in the context of both the USB and Ethernet inputs.
When the Bartók is driven via the USB input, all the music software runs on the music computer. I ran Euphony OS, booted into “ramroot” mode, which is the best-sounding operating system (OS) I have found to date. Running on this OS, I compared the following music player software driving the Bartók:
- Roon Server
- Roon Core + StylusEP player
- Stylus (Stylus is supplied with Euphony OS)
As with other all other DACs I’ve tried, Stylus was by far the best sounding on the Bartók, followed by Roon + StylusEP, with Roon Server dead last. Roon has unmatched library management and UX (user experience), but in terms of sound quality, it seriously lags other players. The Roon + StylusEP combo tries to combine the best of both worlds, by retaining the Roon Core for library management, while using StylusEP (the player portion of Stylus) to interact with the DAC. This is a good compromise, but StylusEP simply cannot do the optimizations that the Stylus server does to deliver its outstanding sound quality.
Bottom line: the best sound quality over USB to the Bartók was when running Stylus music software on the music computer.
When the Bartók is driven via the Ethernet input, music software is distributed, with the music server running upstream, and the endpoint or renderer running on the Mosaic processor on the Bartók. Since this processor supports multiple software endpoints, I compared them for sound quality. For local music residing on my NAS, I compared:
- The Mosaic Roon Ready endpoint, with Roon Core on the music computer, to
- The Mosaic UPnP renderer, with UPnP server (minimServer) on NAS.
Similarly for streaming music, I compared Qobuz streaming on:
- The Mosaic Roon Ready endpoint, where Roon Core accesses the Qobuz stream, to
- The Mosaic Qobuz application on the Bartók directly streaming from Qobuz.
In both local and streamed cases, the Mosaic applications - i.e. the UPnP renderer and the Qobuz application - delivered better sound quality than their Roon equivalents. Clearly, dCS has optimized the firmware of the Mosaic processor as well as the endpoint software running on it, and the end result is notably better sound quality.
If you are a Roon user, but are willing to forgo the rich user experience of Roon for the best sound quality, the Mosaic applications are a better sounding alternative.
Comparison of Ethernet and USB inputs
This section addresses the perennial question users have: which input sounds best? Of course, it is impossible to evaluate an input without also considering what is upstream, so one has to look at the aggregate of the entire chain. This also means there is no single answer - it is system-dependent.
In my system, I decided to compare the best possible performance I could extract with USB and Ethernet. This meant comparing:
- The optimized USB hardware chain, running Stylus as the music player, with
- The optimized Ethernet hardware chain, with Mosaic UPnP renderer, minimServer on NAS.
In all the Ethernet DACs I’ve had in my system to date, this comparison has not even been close: the USB chain has always been significantly better. Not so with the Bartók! This is the best I’ve heard an Ethernet input sound relative to USB.
In the end, I still found the USB input to have a modest advantage, and I attribute this to a combination of my custom music computer, Euphony OS and Stylus software, and other USB optimizations as shown in the topology diagram.
So what does this mean? While the USB input does provide more opportunity for upstream optimization, both in the hardware and software chain, the sound quality with the optimized Ethernet chain is not far behind.
Users who want the utmost simplicity can just attach their Bartók to their home network and still have an excellent-sounding, full fledged music system. That’s powerful stuff!
Comparisons with other DACs
Now that we’ve established how to get the best out of the Bartók, let’s get into the meat of this review - the comparisons with other DACs. As is my practice, I like to do direct head-to-head comparisons with as many relevant DACs as I can get my hands on. I set them up in my system with careful matching of levels. I did many of these comparisons both on my primary system, as well as the speaker-based system.
Let’s look at the competitive lineup. All prices listed here are US MSRP.
First, the reference DACs: my reference until a few months ago had been the Ayre QX-5 Twenty ($7650 with USB only, $8950 adding Ethernet). My current reference “DAC” is actually a combo of the Chord Hugo M Scaler ($4795) and TT 2 ($5495). Late in the review period, I was able to experience the Bartók’s older and much more expensive sibling, the Rossini DAC ($23,999).
It’s important to keep some things in mind in the following sections. First, every DAC featured sounds excellent in isolation. There are no mediocre DACs here. It needs to be emphasized that any differences in sound quality I describe are relative, and do not change the inherent quality of these DACs. Finally, at this price and quality point, we are well into the diminishing returns of the price-performance curve. Whether the sonic differences I describe are worth the price differential is not something I’ll get into, as that is a personal calculation for every buyer to make.
While I’ll indicate my preferred DAC in each comparison, you as a prospective buyer would be ill-advised to just take my word for it. My intent here is to describe the character of each piece, but every DAC in this section is worthy of an audition in your own system before making any purchase decisions.
Comparisons with Ayre QX-5 Twenty
The Ayre has been a favorite DAC of mine, and a reference for quite some time. I consider its particular strengths to be: dynamics, bass, resolution, and refinement. I compared both the QX-5 Twenty and the Bartók via their USB inputs, both in my primary system, driving the Cavalli amp, and on the speaker-based system.
On Vladimir Jurowski and the London Philharmonic Orchestra’s delightful new release of Mahler Symphony No. 4 (LPO, 24/96), I was reminded how cruel high-end comparisons can be. On the last movement, Sehr Behaglich, it was very quickly apparent the Bartók had the advantage. Sofia Fomina’s voice was more emotionally engaging and tonally refined. In the pianissimo passages, the Bartók’s greater resolution for micro-detail was evident. The QX-5 Twenty seemed to have a brighter tinge and higher noise floor. On crescendos, the QX-5 Twenty retained its penchant for physicality, but the Bartók not only matched it, but pulled ahead on dynamics.
I find a lot of great music from the peerless Album of the Evening thread here on Audiophile Style, including this one: Live at the Olympia - June 27, 2012 (Jazz Village, 24/44.1). Every track is a gem! On Blue Moon, the Bartók better captured the ambience of the venue, and portrayed a bigger, more dynamic soundstage. Jamal’s piano sounded more like a piano, and the notes had more coherence and attack. All the instruments - the piano, the double bass, the drums, etc - had more heft, texture, and dimensionality. Bottom line - the Bartók just conveyed the piece more realistically.
Of course, I need hardly remind readers that the Bartók’s MSRP is almost 70% higher. In high-end audio, you usually get what you pay for!
Comparisons with Chord Hugo M Scaler + Hugo TT 2
The Chord Hugo M Scaler and Hugo TT 2 DAC combo - or HMS/TT2 - is my current DAC reference, so I spent a lot of time comparing and contrasting the Bartók to it. MSRP-wise, the HMS/TT2 is cheaper, but when you factor in the cost of additional cables and power supplies, the cost differential between the combo and the Bartók shrinks considerably. Readers may ask - wouldn’t the Chord DAVE (MSRP $10,800) or the HMS/DAVE combo (MSRP $15,595) be a better comparison? Sure - those would be very interesting, but I did not have a DAVE available.
Also interesting to me is the allocation of resources. The Bartók is an integrated unit, and resources have been allocated in whatever manner dCS saw fit. Since the HMS/TT2 is a combo of separates, we know about 45% of the cost is allocated to the M Scaler, a component I called a game changer in my recent review. How does this distribution of resources manifest in sound quality?
Both the HMS/TT2 and the Bartók sound their most transparent when driving headphones directly, so this is how I assessed them in my system. For speaker listening, I used the balanced analog outputs to the preamp. First up, some speaker listening.
Listening to the title track of Herb Alpert’s classic album Rise (Herb Alpert Presents, 24/88.2), these two DACs presented a study in contrasts. The Bartók impressed with a large, dense soundstage, where Alpert’s soaring trumpet, the marimba, and the piano all had a rich, natural tonality. I’ve heard this song when it was originally released on vinyl, and many times since, but I’ve never heard it sound as dynamic as it did with the Bartók. The HMS/TT2 combo had different strengths. Instruments had more texture. Especially on the marimba, cymbals and brushes, transients were cleaner, with more micro-details coming through.
Simon Rattle’s farewell concert with the Berlin Philharmonic of Mahler’s Symphony No. 6 was an excellent performance, and beautifully recorded for release on the Berliner Philharmoniker label (24/192). In the Scherzo movement, both DACs were able to draw me into the sheer beauty of the melody. The Bartók had the wider and denser soundstage, while the HMS/TT2 placed instruments with more precision in space. Front to back depth was on par with both DACs. On crescendos, the Bartók was the more thunderous, while on dense passages, the temporal coherence of the HMS/TT2 shone through, making instruments easier to follow even after they receded from the foreground.
This study in contrasts continued when I switched to headphone listening. On Laudamus te from Mozart: Great Mass in C Minor (Philips, 16/44.1), Sylvia McNair’s captivating voice was positioned with spooky precision by the HMS/TT2. Despite the late-80’s vintage of the recording, both DACs did an excellent job resolving the dense textures of the harmonies. The Bartók again had the edge on sheer dynamics and instruments were slightly better fleshed out.
Rounding out the comparison was this gorgeous recording by the Anima Eterna Brugge ensemble of one of my favorite pieces of chamber music, Schubert’s Octet in F Major, on the album Schubert, Berwald - Chamber Music (Alpha, 24/96). From the first notes of the opening movement, you can tell how dynamic and expansive this recording is, and it came through authoritatively on the Bartók. The HMS/TT2 placed each ensemble member’s instrument with precision, and conveyed more of the timbre of the strings and winds. The Bartók excelled at the growl of the cello and double bass.
And so it went. On recording after recording, the Bartók and the HMS/TT2 showcased their respective strengths. The Bartók led on density, dynamics, physicality, and a wonderfully natural presentation of instruments. The HMS/TT2 had the edge on coherence, transparency, cleaner transients, instrument placement, and voices. Anyone fortunate enough to be considering these pieces for their system, especially as a high-end headphone DAC/amp, needs to audition both.
I usually try not to waffle in these comparisons, and declare the winner to my ears. In this case, all I can say is I am keeping the HMS/TT2 combo and have sent the Bartók review unit back to dCS - but not without a deep pang of regret. It’s that good.
Comparison with dCS Rossini
I’ll admit it. I did this last comparison mostly to satisfy my curiosity - and because I could. You see, I’ve never been this far up the price/performance curve before. Oh sure, I’ve heard expensive gear at shows, but not in my own home. The Bartók is the most expensive piece of gear to grace my audio rack. Late in the review period, I got to wondering: the Bartók must surely be far into the diminishing-returns region of the price-performance curve. How much better could its bigger brother, the Rossini, really be? And - do I even have a resolving enough system to be able to tell? There’s only one way to find out! My local dCS dealer, Casey McKee came through and graciously lent me his demo unit for a couple of weeks. His unit is actually the Rossini Player + DAC (MSRP $28,499), but since I only used it as a DAC, I’ll pretend the comparison was with the Rossini DAC (MSRP $23,999).
Since the Rossini doesn’t have a headphone amp, I did this comparison via the Cavalli headphone amp.
I’ve been living under a rock, and had not heard of Tool’s latest album, Fear Inoculum (RCA Records, 24/96), until my listening buddy Eric rectified that. What an amazing album! How had I missed this? Any hopes I had that the Rossini’s advantage would be subtle and small were dashed within the first few seconds of the opening title track. Instead, my jaw dropped. Remember how I said comparisons were cruel? Well, the same Bartók I was praising for its dynamics and physicality earlier now sounder positively skeletal compared to the Rossini! The sheer realism of the instruments was eye-opening. Nowhere was this more evident than on the percussion. The level to which the Rossini conveyed the weight, the volume, and the texture of the djembe drums was remarkable. Later in the track, when the guitars kick in to create a veritable wall of sound, that wall was darn near impenetrable with the Rossini!
On the first movement of Elgar’s Cello Concerto, from the album by the BBC Symphony Orchestra (Harmonia Mundi, 24/44.1), the realism with which the Rossini reproduced Jean-Guihen Queyras’ cello was quite overwhelming. If you’ve ever heard a live cello, you know the weight and physicality the instrument conveys, the incredible detail and texture as the bow strokes the strings, and the satisfying bass of pizzicato notes. The Rossini came close to that live experience.
Suffice it to say that the next two weeks passed in audio bliss as I listened every moment I could to the Rossini. If I lived in a universe where I could afford this beast, I would get it.
But let’s come back to the original questions I posed at the start of this section. I was flabbergasted to hear the magnitude of improvement that still exists between the $15k Bartók and the $24k Rossini, and the fact that it was clearly audible in my system. It’s actually rather depressing, as us plebeians would love to believe that diminishing returns for DACs kick in at the - pick a number, $500 or $2000 - mark. While it is true there are many excellent DACs in the sub-$2k price range, the improvements do continue on up the price spectrum, and they are substantial.
The dCS Bartók is one of the most impressive pieces of audio gear I’ve had in my system: ever. Just as a DAC alone, its sound quality is stunning. As a DAC/headphone amp, it competes head to head with my reference Chord Hugo M Scaler + Hugo TT 2 combo. But wait - there’s more. It contains a network streaming processor that delivers the most comprehensive feature set and the best sound quality I have heard from a streaming DAC to date. Add it all up, and for some audiophiles, the dCS Bartók is an absolute bargain.
Before you worry that I’ve lost my mind, calling a $15k device a “bargain,” let me explain. If you’re a high-end headphone enthusiast, considering building a streaming music system to match the quality of your top-of-the-line headphones, you’d be considering a bespoke headphone amp, a high-end DAC, and some combination of music computer and streaming endpoint device. Once you start pricing reference-class pieces in each of these categories and add up the cost, then yes: the Bartók really looks like a bargain. The Bartók can give you all of that functionality in one box. It tastes great, and is less filling. You can get started with just a network connection, and perhaps a simple NAS UPnP server for your local music.
Now, that is not to say that the Bartók’s performance in a system cannot be enhanced by paying attention to the source. Look who’s talking, after all! If you do optimize, the Bartók will reward you with better sound quality. But due to its excellent input isolation, these improvements are more modest than I’ve experienced with most other gear.
Perhaps the biggest compliment I can pay the Bartók is that even an inveterate tinkerer like me was strongly tempted to give that up and settle down with the Bartók. There is an elegance to a single box like the Bartók that does it all, and very well. It makes a compelling case to stop tinkering, and just enjoy the music. After all, isn’t that what it’s all about?!
- Product: dCS Bartók DAC with Headphone Amplifier
- Price: US MSRP $15,000, or $13,500 without headphone amp
Documentation: dCS Bartók Documentation
Music Computer: Custom computer: H370M-ITX/ac, i7-8700T, 8GB RAM, HDPlex H3 case,
32GB M10 Optane SSD for OS, JCAT Net Card Femto,
running Euphony OS with Stylus or Roon+StylusEP music software
Music Storage: Synology NAS DS916+ 4-bay, attached to router via Ethernet
Headphone Amplifier: Cavalli Liquid Gold
Headphones: Sennheiser HD800 (SD Mod), Meze Empyrean, Abyss AB-1266 CC
USB Regenerator: SOtM tX-USBultra SE
Ethernet Switch: SOtM sNH-10G SE switch
Reference Clock: Mutec Ref 10 10MHz clock driving the tX-USBultra and sNH-10G
Power supplies: Paul Hynes SR-7 SR (single regulator) for switch & tX-USBultra,
Paul Hynes SR-4 for JCAT Net Card Femto,
HDPlex 400W ATX LPS for music computer motherboard
Power Details: Dedicated 30A 6 AWG AC circuit, PS Audio P5 PerfectWave Regenerator
Power Cables: PS Audio AC-12 (wall to P5), Cardas Clear Beyond (Cavalli Amp),
Cardas Clear for all other components
USB cables: Phasure Lush^2 USB
AES/EBU cables: Cardas Clear
Clock cables: Habst 5N Cryo Pure Silver 50Ω and SOtM dCBL-BNC 75Ω
Ethernet cables: SOtM dCBL-Cat7, Supra Cat 8
DC cables: Ghent Audio custom OCC JSSG360 ATX and EPS cables
Paul Hynes fine silver (SR-4, SR-7)
Interconnects: Cardas Clear XLR balanced
Headphone cables: Transparent Ultra cable system, Cardas Clear balanced and SE cables
Accessories: Synergistic Research Tranquility Base XL UEF with Galileo MPC
Alternate system for speaker-based evaluation
Roon Player: Innuos ZENith MkII SE
Preamp: Audio Research Reference 6
Power Amp: Hegel H30
Speakers: Magnepan 20.7
Subwoofers: 2x Rhythmik F12g
USB Regenerator: SOtM tX-USBultra
DDC: SOtM dX-USB Ultra HD (for USB to AES)
Ethernet Switch: SOtM sNH-10G SE switch
Reference Clock: Mutec Ref 10 10MHz clock driving the tX-USBultra
Power supplies: Paul Hynes SR-7 DR (dual regulation) for tX-USBultra, switch, and DDC
USB cables: Phasure Lush & Lush^2 USB
BNC cables: Amphenol and Digi-Key
Ethernet cables: SOtM dCBL-Cat7
DC cables: Paul Hynes fine silver (SR-4, SR-7)
Interconnects: Cardas Clear, Audience AU24 SX XLR
Accessories: Synergistic Research Tranquility Base UEF
Many thanks to the following companies for supplying cables to aid in this evaluation:
- Cardas Audio for a full loom of Cardas Clear cables, and
- Transparent Audio for the Transparent Ultra headphone cable with a full complement of headphones leads and source terminators.
About the Author
Rajiv Arora - a.k.a. @austinpop - is both a computer geek and a lifelong audiophile. He doesn’t work much, but when he does, it’s as a consultant in the computer industry. Having retired from a corporate career as a researcher, technologist and executive, he now combines his passion for music and audio gear with his computer skills and his love of writing to author reviews and articles about high-end audio.
He has "a special set of skills" that help him bring technical perspective to the audio hobby. No, they do not involve kicking criminal ass in exotic foreign locales! Starting with his Ph.D. research on computer networks, and extending over his professional career, his area of expertise is the performance and scalability of distributed computing systems. Tuning and optimization are in his blood. He is guided by the scientific method and robust experimental design. That said, he trusts his ears, and how a system or component sounds is always the final determinant in his findings. He does not need every audio effect to be measurable, as long as it is consistently audible.
Finally, he believes in integrity, honesty, civility and community, and this is what he strives to bring to every interaction, both as an author and as a forum contributor.