Every once and a while, a product will turn up at the house for review that is so special, it changes what I understand to be possible for an audio component. And let’s face it; most of us don’t have anything resembling “unlimited budgets”, so finding something magical at a price you might (barely/one day/maybe-possibly) afford can make for a great day. For me, I found the dCS Bartok DAC to be just such an animal.
I started reviewing DACs in the headphone space many years ago and I’ve slowly worked my way up the food chain. My current reference is the PS Audio DirectStream DAC (reviewed here) and it’s companion Memory Player (reviewed here). I was curious what a more-than-twice-as-expensive (compared to my DirectStream DAC) component would sound like, and admittedly, I was a bit skeptical that I would get at-least-twice-the-sound of the superb-sounding PS Audio product. At $13.5k for the DAC and another $1.5k for the headphone amp version, dCS had its work cut out for it.
I find that reviewing digital to analog converters is difficult because the very best ones create a feeling that is hard to describe, but I guess that could be said for other audio hardware categories as well. How does one adequately describe the way music makes you feel? How does one adequately describe how a fine DAC presents the music? What makes a dCS product sound better than other competitor’s products?
Digging in: the dCS Bartok DAC
At the Munich Hi-End show, I ran into John Quick of dCS who was running a fantastic system with Wilson Audio Sasha DAW speakers and the latest Rossini on the (sublime!) v2 firmware version. I mentioned that I’d been trying to get a Bartok in for review and John offered a unit. Soon, I was back home in Atlanta unpacking a fresh new, silver Bartok after a brief showing at our excellent local high end shop, HiFibuys for the “Guys Night Out” event. I hooked it up to my office system for a few weeks to break it in. It wasn’t long before I fell in love with the sound. Unfortunately, I now had two strikes against me. I had to describe the sound AND I had to make a case for spending $15K on a DAC.
Around this time, I was able to finally get a sports car, a Porsche Cayman. I say “finally” as I was driving the same car for many years, in part to save up. I mention this as the new car taught me a thing or two about what quality is and why we should pay for it. This new car is superb. It has all the controls in exactly the right place. The feel of steering, the engine responsiveness, and the sense of being fully in touch with the road is just perfect. Every time I search for a change in the settings, the layout is intuitive. Everything just works. It’s a very satisfying feeling.
The Bartok DAC is a lot like my new Porsche. It just works. The display is super high quality. The inputs handle anything from USB to AES/EBU to SPDIF to Ethernet. The build quality is excellent. The paint on the new sports car is a metallic silver that sparkles but is not as flashy as some silvers. The finish on the Bartok reminds me of it; the silver as it is a finished aluminum that sparkles and changes in the light from white to silver to gray. It’s just a beautiful effect. Being their “entry-level” converter, the dCS Bartok is missing the swoopy curves of the Rossini and Vivaldi but the finish is still gorgeous. It’s really spectacular to know there are companies out there finishing even entry-level products like this.
The display is a high-resolution screen. Very clear to read, and if you adjust the volume knob, the numbers get large. And that brings us to the volume control. A light touch applied to the round, silver knob moves you up and down in 0.5 dB increments. It’s so refined. It feels like silk. It’s a small thing but it just speaks to the “quality mindset” of the dCS team in Cambridge.
The case itself is a big rectangular block and the fit and finish look just as good to my eye as the Rossini. It could have been just a big silver flaked box. But dCS left some stylistic thin gaps in the panel joinery to give it a more nuanced look. On the back of the unit, the connections are of the highest quality. Even the power on/off switch has a solid feel and goes “thunk” when rocked to the “on” position. ‘Solid’ is a good word for the Bartok DAC. Not in the sense of “bulky casework” but in terms of “really high quality”; as I said, it really looks like a team of people thoughtfully considered every detail.
At the house here, my favorite DAC-in-residence has been the outstanding PS Audio DirectStream DAC, running the latest “Snowmass” software. I run it also with the superb Memory Player from PS Audio because I have been collecting those little silver discs which seem to have fallen out of favor lately since 1983. It is my considered opinion that the DirecStream is a superb DAC. I offers up oodles of resolution and sounds at times like an analog source. For $6K retail, it took my digital to another level. I love the DirectStream and find it to be one of the best deals in digital audio.
Now one thing that makes the comparison to the dCS Bartok interesting is the firmware upgradability of both units. Both are designed around full software upgradability. But you’re thinking, “sure Lee but what a totally unfair comparison, given the vast price difference.” You are not wrong, but keep in mind the DirectStream punches way above its weight, particularly after the RedCloud and Snowmass updates. And I am sure Ted Smith is writing even better code for the next update. The DirectStream is an amazing product and set a foundation or baseline for my review of the Bartok DAC.
For the first, few weeks I listened to the dCS Bartok in isolation in the office system which has Wilson Audio TuneTot speakers amplified by either my own highly modified Audio Research VT-100 amp with KT-120s or a Schiit Audio Vidar and Aegir amp. I find the best gear takes a while to settle in so I did not take any listening notes for a couple of weeks. I ran a USB cable from a new MacBook Pro running Roon and largely playing my own hirez recordings or Tidal and Qobuz. This sounded amazing right out of the box but after a few days, a veil of fuzziness lifted from the Bartok DAC’s sound. It’s hard to describe but it just became more musical. it always had the resolution but everything became more round in a way that great audio offers. Frankly, I kept the unit in the “upstairs system” longer than anticipated because I enjoyed listening to it so much. Things were off to a really good start.
As I moved the dCS Bartok down to the main system, I heard even more magic from it. The resolution and detail had me spinning those antique silver discs again. I ran an SPDIF out of the PS Audio Memory Player and into the Bartok DAC. I put in Phil O’Hanlon’s demo CD from a year ago and teed up track 19. Leonard Cohen’s “Nevermind” sprung to life. A few things stood out right away. Cohen’s voice was superbly fleshed out in his guttural rolling, slightly foggy vocals. The decay of the notes from his voice trailed off into infinity. The backing singers were well to the far edges of the soundstage. Instrument and vocal separation was just spectacular. The vocals were deeper in the soundstage than I heard with the DirectStream.
Another good listen is “Children Go Where” by the Fairfield Four, featuring Lee Ann Womack. The voices on this recordings just sound damn realistic. They hang in space, separated by air, just like a live performance. Lee Ann’s amazing voice anchors the performance. Whoa. This is quite realistic on the Alexias. My wife loves the sound, saying “that’s just like real life!” Yep. The Ring DAC technology is the real deal. Listen to the room reverb. Listen to the trailing off of the notes. Listen to the bass notes from the Fairfield Four. Perfect.
So here’s what we hear: superb resolution of the recording details, a wide and deep soundstage, a gorgeous midrange.
I switched over to my ATT router to listen through a Starlight 8 ethernet cable from WireWorld going into the Network connection on the back straight from my townhome’s router. Let’s give the Tidal and Qobuz accounts a workout, shall we?
I pick up my trusty iPad which is running Mosaic, a nifty app that is downloaded from apple’s app store and serves as dCS’ music app. This music controller interacts with the Bartok DAC much like my Roon app on the MacBook does. You log into a variety of streaming service options (Tidal, Qobuz, Deezer, Spotify, etc.) and it simply acts as a controller where you can play individual albums or build “queues”.
The track “Celestial Echo” from Malia’s album Convergence has deep foundational bass notes at the start of the track. These are deep and well defined. Surprisingly so, in fact. This is another character trait of the Bartok…well-articulated bass and strongly deep bass. It is pure ear candy that really lays a foundation to the song that contrasts perfectly with Malia’s voice.
“I Feel It Like You” is the third track on this album. I love this song as it’s very catchy and romantic. A celebration of a great relationship between Malia and her man. Now we hear another trait of the Bartok DAC, the coherence from the bass to the highs. Clarity and openness are words that come to mind. I dare you to not move your feet on this gem! This song swings in an elevated pop music way. Boris Blank, of Yello fame, collaborated on this album and the usual atmospheric layers are present and add to the effect as does the refined percussion. A genuinely musical and fun presentation.
The next case study is dynamics, the MQA version of Sinatra and Swingin’ Brass. Listen to “They Can’t Take That Away From Me”. The dynamic brass is full and steep in attack. Frank’s perfect phrasing provides a soothing narrative to another big band assembly by the great Neal Hefti. “I Get A Kick Out of You” is flawless. The real star of this MQA show is “Tangerine”. The opening brass is visceral on the opening notes. Maybe due to tape quality and/or mastering, this track is a real standout sound quality. There is a compelling illusion of a big band laid out wide in front of me between and beyond the speakers. Absolutely marvelous.
Bartok USB Input
Now we move to a third input, that of USB1. I attach my MacBook pro to the device with a USB cable and fire up Roon. Cue up Yello’s Toy album and select “Limbo”. This MQA version really swings. The bass is deep and present. Midrange is crystal clear. The female backing vocals, “I will be cruising through that night…”, are natural. Dieter Meier’s vocals are lifelike and realistically slightly raspy. “Cold Flame” is another diamond performance. It starts slow and builds momentum. Malia guests on this song and, as usual, is perfect. Ethereal sounding vocals that fade off into space. A driving bass and percussion track with synth notes overlaid keeps my toes tapping. Listening to the Bartok DAC is a bit like a ride in my sports car: exhilarating!
This is the beauty of dCS build quality: every input tried was dead neutral and musical. Fully transparent sound across the board. Music as clear as I have heard. Everything I threw it as a source of ones and zeros was handled with aplomb.
Comparison with the PS Audio DIrectStream DAC
I compared the Bartok DAC running its current software with the DirectStream with the latest “Snowmass” software. Both sounded great but the Bartok had two distinguishing advantages: more resolution and a much deeper and wider soundstage. I felt it had the most transparent and neutral sound, whereas the DirectStream had a more smoothed-out “analog” sound. Both scored high on musicality but I believe I heard another level of detail and air between instruments on a perfectly laid out soundstage when I had the Bartok playing.
I marvel at what PS Audio can do for the money, but I long for the much more expensive Bartok as it really was a new level of sound. Everything was just more clear to my ears. I listened to the Wailin’ Jennys’ “Beautiful Dawn” from their 40 Days album. The guitar has a more natural sound and presence on the Bartok DAC. The vocal harmonies are magic on this track. The Bartok gets them more right than the DirectStream.
The Great MQA Test
One of the great features of the Bartok, and all dCS gear for that matter, is the implementation of the MQA filters. My understanding is that dCS and MQA worked very closely together. When I mentioned I was getting a Bartok in for review to Bob Stuart, a smile came over his face and he said I would be impressed. He was right. MQA files played through Tidal are spectacular on the Bartok. And I have an engineer’s test files where there is music in both unencoded format and MQA-encoded format.
The MQA files, level adjusted, have more presence and clarity on the instruments. The Bartok does a really good job of highlighting these differences, more so than three other MQA DACs I have tested with these files. From Khatchaturian to Yello, the Tidal Masters files never let me down with the exception of a couple of poor masterings. It may be the best $20 a month I spend on music. The Bartok is so resolving that with high-quality MQA files, you can tell instantly (sometimes even by the “room tone” before the music(!)) whether you are listening to the non-MQA or MQA version. The Bartok takes the advantages of MQA and highlights them even more than a good MQA-capable DAC. I believe that is due to a more linear and robust implementation of the MQA filters.
Product Features of Note
Another aspect of the device is the ability of software from dCS to upgrade the sound performance. There are firmware upgrades for the DAC every one-to-two-years, not unlike the DirectStream’s. I believe this is important as a $15K investment should be future-proofed to a large degree. It’s not a DAC where, every year, a new Sabre chip comes out and all of a sudden one has a nice “doorstop DAC” that is now dated. That gives me comfort.
As well, a device at this price point deserves a solid build. It’s heavy — almost 37 lbs! As I mentioned, the casework is very well constructed. Look at my accompanying photos to see the casework joinery and thickness. It’s also somewhat inconveniently big with a very wide and deep chassis (17.5” x 17.0” x 4.6“), but it’s a beast in the best possible way. Even the feet are well designed and solid. It feels like no detail has been ignored. Even the owner’s manual is readable and well designed. Geez.
What don’t I like about the Bartok? Uh, give me a minute…okay…the print on the tiny buttons up-front is hard to read unless you are very close. But it looks kind of British modern cool anyway in its sleekness.
“That’s all you got, Scoggins?”
Well sadly, yes. This is one thoroughly executed product.
Wrap It Up (For Me)
This type of product is one that causes a reviewer to pause and consider the price of purchase. And then reconsider. Unfortunately for me, the timing was bad, having just purchased the car. However, Christmas and bonus season are not too far away.
I’ll be honest, I want a dCS product. To be even more honest, if the Bartok is this good, I’m now daydreaming about what a Rossini might do.
The Bartok sounded amazing in Munich paired with those Sasha DAWs. My time with the Bartok has confirmed that this is “reference-quality sound”. Combined with it’s Swiss Army knife-like flexibility, incredibly solid build quality, and firmware future-proofness, it starts to feel like a reasonable deal. I have to admire dCS offering this much sound quality in their entry-level product. Indeed, the Wilson Alexias have spoken. The Bartok will one day soon be mine. Bank on it.
So there you have it…my over the moon, rave unto the sound fantastic, highest possible recommendation. If you are anywhere near this price category, get yer butt over to the local dCS dealer with some discs and files.
I know you’ll enjoy the ride.
For more information on the dCS Bartok DAC, see https://www.dcsltd.co.uk
For more from the Part-Time Audiophile team on the dCS Bartok DAC, check out:
[Editor’s Note: Here follows “Part One” of a two-part series featuring the new dCS Bartok DAC. In this installment, Lee Scoggins will be evaluating the new DAC’s two-channel performance. In the companion piece, coming this Fall, Brian Hunter of The Occasional Podcast and Audio-Head will be diving into the product’s headphone/personal-audio superlatives. Stay tuned. — Scot Hull]
An Interview with John Quick, dCS North America
I wanted to explore some specifics of the technology and product so I sent along some questions to John for his take. As always, John was on his game and insightful in his replies.
Ring DAC – What are the primary advantages of the ring DAC and how does that add to the a. resolution and b. wide/deep soundstage I am hearing.
Real resolution (i.e. extreme linearity, not just good THD+N measurements) allows for the special cues that can provide not just focus, but extremely wide and deep soundstages if the recording calls for it. Here are a few datapoints:
- The RingDAC is a discrete DAC of our own design that combines proprietary hardware (Control Board “motherboard” + Analog “DAC” Board) and software
- The RingDAC’s name is actually attributed to the “mapping” algorithm (i.e. mapper) that alternates/changes the pattern of which of the various current sources on the Analog Board are called to contribute to the reconstruction of the signal on a sample-to-sample basis
- The RingDAC mapper has been called a “randomizing” algorithm by many, when in fact it is not; rather it is a carefully calculated set of patterns used to minimize noise, distortion, and crosstalk while primarily keeping the highest degree of linearity by averaging out the contribution of parts that fall out of spec over time (re-read that last part, will revisit below)
- Since its inception (circa 1989) the central processing architecture of the RingDAC has used FPGAs to translate all incoming data to RingDAC format, which for years has been 5-bits @ 2.822 or 3.07 mS/sec (or MHz, if you will), synchronous with the incoming data rate. Recent advances in our hardware, and research into bettering the RingDAC mapper, now allow the system to run at twice that speed, or 5.644/6.14 mS/sec (Vivaldi 2.0 and Rossini 2.0)
- Either/both FPGAs and DSPs are used for the digital filters, and all dCS filters are written specifically for each sample frequency by our engineers
- The RingDAC Analog Board is a completely balanced, differential dual-mono design that (currently) consists of 48 equally(unitary)-weighted current sources per channel that are switched at a rate of between 2.822MHz and 6.14MHz, depending on the model and the incoming data. The outputs of these current sources are summed and filtered by discrete analog componentry, and this signal is then amplified by an electronically floating, discrete, fully-differential Class-A transistor output stage
- Although the RingDAC analog board hardware features an array of resistors and corresponding current sources, it is NOT a ladder DAC
As for the main components that make a dCS RIngDAC go: The main purpose of the Control Board (aka dCS motherboard) is to create a 5-bit audio stream at 64x the base rate, so 2.8824MHz for 44.1kHz based sample rates, 3.072MHz for 48k based ones (coincidentally, this audio stream is 128x the base rate in the new mappers contained in the Vivaldi and Rossini 2.0 software). It does this by performing quite involved digital filtering first (which creates very wide, very high-rate samples) before passing the stream through a highly-optimized noise shaper, whose function is to reduce the number of bits by shifting quantization noise out of the audio band.
What’s frequently described as the ‘DAC board’ we call the Analog Board (as its discrete design has a section that translates the digital code into a low-level analog signal and a pure, Class-A analog section that filters, mixes, and amplifies what comes off the DAC section). The DAC section of the Analog Board uses a mapping algorithm of our design to take in the 5-bit samples from the Control Board and map them onto an array of unitary-weighted current sources, using a set of rather convoluted rules. It is important to note that this is *not* the same as 5 one-bit DACs – the quantization noise is entirely different (and lower).
The key point to all of this is that the RingDAC’s mapper decorrelates any one/particular current source from the signal, thus reducing distortion significantly. This decorrelation is also important and significant because parts values change over time. Decorrelating minimizes any given component’s ultimate contribution to the end-result, meaning, WHEN a component’s specified value begins to drift, distortion and linearity remain at the same exacting levels. This allows for low-level detail to be heard that just goes missing in other designs- chip, ladder, or discrete with a limited-number of output elements- and, as the DAC ages, it will measure and sound virtually identical for the life of the product.
Casework – I love. the finish and solidity of the Bartok’s case. How did dCS get the sparkle like luster on the case itself and how does such a heavy chassis contribute to sound quality.
Thanks for the kudos! First, the finish of your Bartok sample is done with a special “bright anodizing” process that is only available in ONE place in the UK (near Cambridge) due to environmental factors. The combination of the anodizing process and the machining prep work (all panels are machined from billet) allow for the silky, sparkly finish. Second, we’ve always worked extremely hard to build beautiful components whose designs stand the test of time, but we also pay special attention to things “under the hood”. Weight does have some effect in potentially contributing to a more quiet chassis through raising the chassis resonant frequency and absorbing (or blocking) certain vibrations from outside; but we also go to great lengths to isolate boards and PSU transformers internally on sub-assemblies that are also treated to other proprietary isolation methods.
Headphone amplifier. how did dCS create such a strong amplifier product?
To be honest, the Class-A headphone amplifier bears significant resemblance to our traditional analog output stage, just better optimized for driving headphones. I certainly don’t mean this response to down play the sophistication of the headphone amp design, rather it speaks to the importance of the discrete analog output section of the RingDAC, which has been largely unchanged in its overall circuit topography for many years.
Mosaic – why did dcs feel compelled to develop this control app? What was the strategy and does it have an impact on sound?
We think that any manufacturer who designs a network audio product needs their own control App. We began a partnership with a network hardware provider in 2011 to write code and develop a GUI for the first Vivaldi App, and over the years we continued developing with them on Apps for Rossini, Network Bridge, and Vivaldi One. In 2017 (around the time we developed our unique MQA decode/rendering solution) we decided to bring all of this in-house. After spending considerable time cleaning up the old Apps to make Roon, MQA, and other general App operations better it became clear that we could do a much better job starting from scratch, so came Mosaic.
As for impact on sound- Mosaic’s code is far more efficient than our old code, meaning it better utilizes the available processing power, and it no longer requires lines of “compatibility” code to ensure the RoonReady endpoint code cohabitates well with our UPnP and native streaming services. Furthermore, it sees all dCS devices on a network at once (versus requiring different Apps for different dCS products before), and it allowed us to add more streaming services (Qobuz, Deezer, Internet Radio) that were difficult to add to the old firmware platform.
Firmware updates – it sounded to my ears like the Rossini recent update was magical and a big leap forward. Would a Bartok owner be able to expect updates periodically over the next ten years? I am trying to make a case for ownership with myself and readers. What can you say to give comfort that a $15K investment will remain robust well beyond the usual DAC upgrade cycle?
The goal of any dCS product introduction is to ship the very best product we know how to build on release and ensure there is ample headroom in the design (through available processing power, expandability or exchangeability of some parts such as network hardware) to continue to improve and enhance it over its lifecycle- which for us is a 6-10 year commitment. Knowing how much effort has been put into everything the Bartok can do today- basically, everything we’ve developed and improved since 2013 on Vivaldi and since 2015 on Rossini- I have no idea what’s coming next… but Bartok is equipped for it, and we’re ready for the challenge when it presents itself.