Content and analysis made possible thanks to support from https://headphones.com, and https://patreon.com/goldensound supporters.
The Bartok was loaned to me for review by a friend who does not wish to be named.
ChitChat:
The name dCS is somewhat synonymous with ‘High end Audio’. They’re a very well known, very well regarded brand producing some eye-wateringly expensive digital to analog converters, CD transports, upsamplers and master clocks.
The Bartok is their lowest cost offering, and comes in at a hefty £12,500.
Firstly I have to mention, this is far and away the best build quality of any DAC I’ve tried. Seriously the chassis looks indescribably clean and precise in person, with a gorgeous finish. Even the knob for the volume control is just the perfect level of resistance and sooooo smooth.
This thing absolutely OOOZES quality.
I do wish the display was OLED though to eliminate the backlight glow.
dCS uses a proprietary ‘ring dac’ design, which they explain in detail in this post on head-fi:
https://www.head-fi.org/threads/dcs-ring-dac-a-technical-explanation.957387/page-6#post-16353403
The DAC is quite large, and it seems that it needs to be given how much space internally is used!
Unfortunately I cannot open this unit so am unable to take a close look at specifics, but we can see that a large portion of space is dedicated to the PSU, with the headphone amp board on the front plate of the chassis.
The ring dac board is on the right. There is mention on the dCS forums that the XLR outputs use a discrete output stage whilst the RCA outputs use opamps. However I could not find an ‘official’ source for this info.
Lets see what the objective performance of this DAC is like!
I should note: By default the Bartok is set to use internal DSD upsampling, but this can be disabled. For all measurements shown below I have disabled the DSD upsampling as doing so subjectively sounded better to me, and objectively performed better in many areas too. Personally I’m not sure why the DSD upsampling is enabled as it only seems to hinder the sound and performance of this dac.
Measurements:
Test Setup:
– Audio Precision APx555 B-Series Analyzer with 200kOhm input impedance
– Audioquest Mackenzie RCA/XLR cables
– USB Source: AMD PC via ifi iGalvanic 3.0 and iUSB 3.0
– All measurements shown in this post are taken with USB input via the dCS ASIO driver unless otherwise specified.
– All measurements shown are using the XLR outputs, with output voltage set to 6v. Filter set to ‘F1’ for both DSD and PCM
– Full reports containing additional data and test configurations are attached
– Exact analyzer/filter configurations for each measurement are detailed in the full reports
Full Measurement Reports:
Reports available here:
Dynamic Range (AES17): 118.2dB
SNR: 115.2dB
Noise Level RMS (20-20khz): 6.39uVrms
Noise Level RMS (20-90khz): 81.48uVrms
DC Offset: 140uV active, 205uV idle
1khz 0dBFS:
1khz -3dBfs:
1khz 0dBfs DSD Upsampling:
1khz -3dBfs DSD Upsampling:
Performance of the dac within 20hz-20khz does not seem to change seemingly at all when DSD upsampling is enabled, other than the output is 0.1v lower.
Headphone amp 4v output 300ohm:
Headphone amp 4v output 32ohm:
Headphone amp 700mv output (Headphone level) 300ohm:
Headphone amp 700mv output (Headphone level) 32ohm:
The headphone amp seems to measure very well overall.
Filters/Upsampling:
The bartok has several different upsampling options, and the 6 filters are shown below.
But, as mentioned before, the Bartok is set to upsample to DSD by default. Subjectively this mode sounded worse to me than PCM. In the 20hz-20khz band there was no objective benefit to upsampling to DSD, but if we look outside 20khz, DSD upsampling on the Bartok adds a quite frankly ridiculous amount of ultrasonic noise:
Given as to my ear the DSD upsampling does not sound as good as the PCM (DXD) processing, and that it adds a huge amount of noise, I’m not really sure why this is enabled by default.
Jitter:
The Bartok can be used via USB as normal, but also features two ethernet ports and can be used directly as a Roon endpoint.
This is a feature that I really wish more dacs had!
Jitter performance was seemingly identical on both USB and ethernet. In both cases it was ‘good’ but honestly not as good as you’d expect for this price. dCS recommends using their master clocks for best performance but it is a shame that jitter is not as good as other similar or lower cost dacs including competing products like the Rockna Wavedream (which keeps all jitter products below -160dB) or Holo May (which has no visible jitter products down to -170dB), or even many lower cost products (such as the Gustard X26 Pro which keeps all jitter products below -150dB). Unfortunately I can’t see from available internal pictures what specific clocks the Bartok is using.
THD+N / Frequency:
THD+N is flat across the frequency spectrum, good to see! Some DACs have a rise in distortion in upper freqs.
Linearity:
When DSD upsampling was enabled, linearity was drastically affected as shown below. Further measurements showing DSD vs PCM are in the next section.
Much poorer due to excessive noise
SINAD/THD+N vs output level:
IMD (SMPTE) vs Output Level:
Crosstalk:
-90.31dBfs 1khz Sine:
Idle Noise Spectrum upto 1.2Mhz:
Multitone:
One channel seemingly a fair bit better than the other, though both very good.
Overall the performance of the Bartok is generally excellent, I just wish that the ultrasonic noise wasn’t so high!
Kudos for exposing the biggest scam of dac industry. Now I can understand why Bartok sounds so bright. Odd order distortion dominating. Also imd SMPTE poor.
What “scam” are you referring to?
a $17500 dac having so high level of odd harmonics ? it does not measure even at par with much cheaper dac measured here ? so people will pay for chasis, looks and bright sound of odd order harmonic distortion ? and what about one channel behaving differently than other in this $17500 dac ?
A $17500 boutique dac with ridiculously low (less than 100db) stop band attenuation ? DCS whom you are fooling ? are you still living in 1980s ? :p
Erm, 100dB. Are you sure?
Hey…at least it looks good.
then better DCS should promote looks on their website than about sound quality.
Would you prefer this or the Wavedream Signature?
I’d take a wavedream, May, M1se, terminator plus over this without hesitation
“it is a shame that jitter is not as good as other similar or lower cost dacs including … Holo May (which has no visible jitter products down to -170dB)”
Stereophile has measured both the Bartok and the HoloAudio May L3. They observed the jitter as much, much lower for the Bartok. What gives?
My jitter measurement is near identical to stereophile’s: https://www.stereophile.com/content/dcs-bartok-da-processorheadphone-amplifier-measurements
With the two spurs about – 145dB,same as on stereophile.
I meant that’s much lower than the (Stereophile) measured jitter for HoloAudio May L3.
Look at GoldenSound’s DDC video on YouTube. With the phase lock loop PPL system enabled jitter on the May is below 150.
BTW, DSD upsampling is not “on by default”. What firmware version did you test?
Where are the impulse responses to which you refer as “shown lower down in this post”?
I agree w Darren, not sure where goldensound is sourcing his holo may jitter data? Stereophile measurements demonstrate even the old base bartok out performing the may. I am a may kte owner so would love for that not to be so. Was also a bit confused by his concern about noise above 20khz. He must really have an amazing room and sound system, not to mention better than human hearing. Really hope this skewed data is not confusing people in to believing there are issues w the bartok, but based on above comments, it seems to be the case. I generally really enjoy the subjective side of his reviews, but think when it comes to measurements, this one demonstrates you are probably better off referencing sources with a bit more expertise in this area.