dCS Bartok D/A processor/headphone amplifier Measurements

Sidebar 3: Measurements

I measured the dCS Bartók using my Audio Precision SYS2722 system (see the January 2008 "As We See It"), using the Audio Precision's AES/EBU and S/PDIF digital outputs and USB data sourced from my MacBook Pro running on battery power, with Pure Music 3.0 playing WAV and AIFF test-tone files.

Apple's USB Prober utility identified the dCS processor as "dCS Bart\363k USB class 2" (footnote 1) from "Data Conversion Systems Ltd," with the serial number "2." The Bartók's USB port operated in the optimal isochronous asynchronous mode. Apple's AudioMIDI utility revealed that, via USB, the Bartók accepted 16- and 24-bit integer data sampled at all rates from 44.1 to 384kHz. The AES/EBU and coaxial and TosLink S/PDIF inputs all locked to datastreams with sample rates of up to 192kHz.

With the Bartók's volume control set to its maximum, a 1kHz digital signal at 0dBFS resulted in a balanced output level of 6.03V into 100k ohms with the output level set to "6V," 2.04V with it set to "2V," and 603mV with it set to "0.6V." The maximum levels from the unbalanced outputs were all very slightly lower and the maximum level from the headphone output was 6.82V. (Except where indicated, all measurements were taken with the "6V" setting.) All the outputs preserved absolute polarity (ie, were noninverting). The balanced output impedance was extremely low, at 1.5 ohms from 20Hz to 20kHz—half the specified 3 ohms. The headphone output impedance was even lower, at 0.7 ohms, 20Hz–20kHz, while the unbalanced output impedance was as specified, at a still-low 51 ohms.

The Bartók's impulse response with 44.1kHz data depended on which reconstruction filter had been selected. F1 and F6 had relatively long, time-symmetrical, linear-phase impulse responses (fig.1), while F2, F3, and F4 were shorter linear-phase filters (fig.2). The odd man out is F5, which is a minimum-phase type, with all the ringing following the single high sample (fig.3). With 44.1kHz-sampled white noise (footnote 2), the Bartók's responses with the F5 and F6 filters (fig.4, red and magenta traces) reached full stop-band suppression at exactly half the sample rate, indicated by the vertical green line in fig.4, and are therefore apodizing types. F1 rolls off a little less quickly (not shown) but reaches full stop-band suppression a little lower in frequency than F2 (fig.5). The latter still suppresses the aliased image at 25kHz of a full-scale tone at 19.1kHz (blue and cyan traces) by almost 70dB. F3 and F4 offer increasingly slow ultrasonic rolloffs—with F4 (fig.6) I had to reduce the level of the 19.1kHz tone by 6dB to avoid the audioband noise floor being contaminated with aliasing products. The harmonics of the 19.1kHz tone all lie at or below –107dB, however.

1019dCSBartfig01

Fig.1 dCS Bartók, F1, impulse response (one sample at 0dBFS, 44.1kHz sampling, 4ms time window).

1019dCSBartfig02

Fig.2 dCS Bartók, F4, impulse response (one sample at 0dBFS, 44.1kHz sampling, 4ms time window).

1019dCSBartfig03

Fig.3 dCS Bartók, F5, impulse response (one sample at 0dBFS, 44.1kHz sampling, 4ms time window).

1019dCSBartfig04

Fig.4 dCS Bartók, F5, wideband spectrum of white noise at –4dBFS (left channel red, right magenta) and 19.1kHz tone at 0dBFS (left blue, right cyan), with data sampled at 44.1kHz (20dB/vertical div.).

1019dCSBartfig05

Fig.5 dCS Bartók, F2, wideband spectrum of white noise at –4dBFS (left channel red, right magenta) and 19.1kHz tone at 0dBFS (left blue, right cyan), with data sampled at 44.1kHz (20dB/vertical div.).

1019dCSBartfig06

Fig.6 dCS Bartók, F4, wideband spectrum of white noise at –4dBFS (left channel red, right magenta) and 19.1kHz tone at –3dBFS (left blue, right cyan), with data sampled at 44.1kHz (20dB/vertical div.).

The Bartók's frequency response with F1 and data at 44.1, 96, and 192kHz rolled off sharply above half of each sample rate (fig.7). The rolloffs started earlier with the slower-rolloff filters, but even with F4, the output was flat to 17kHz or so (fig.8). Unlike the reconstruction filters found in most D/A processors, where the responses at the higher sample rates follow the same shape, flat to 20kHz, with then a slow rolloff disturbed by a steeper drop at each Nyquist frequency (footnote 3), fig.7 and fig.8 indicate that each filter is optimized for each sample rate. Peculiarly, while F5 and F6 were functional at the 44.1kHz and 192kHz sample rates, they weren't available at 96kHz. Both these filters offer steep rolloffs above 20kHz, but F5 was down by 1.5dB at 20kHz with 192kHz data (fig.9). F6's output with 192kHz data was flat to 20kHz and down by 3dB at 48kHz (fig.10).

1019dCSBartfig07

Fig.7 dCS Bartók, F1, frequency response at –12dBFS into 100k ohms with data sampled at: 44.1kHz (left channel green, right gray), 96kHz (left cyan, right magenta), 192kHz (left blue, right red) (1dB/vertical div.).

1019dCSBartfig08

Fig.8 dCS Bartók, F4, frequency response at –12dBFS into 100k ohms with data sampled at: 44.1kHz (left channel green, right gray), 96kHz (left cyan, right magenta), 192kHz (left blue, right red) (1dB/vertical div.).

1019dCSBartfig09

Fig.9 dCS Bartók, F5, frequency response at –12dBFS into 100k ohms with data sampled at: 44.1kHz (left channel cyan, right magenta), 192kHz (left blue, right red) (1dB/vertical div.).

1019dCSBartfig10

Fig.10 dCS Bartók, F6, frequency response at –12dBFS into 100k ohms with data sampled at: 44.1kHz (left channel cyan, right magenta), 192kHz (left blue, right red) (1dB/vertical div.).

The headphone output offers a crossfeed function intended to optimize the listening experience with headphones. I measured the effect of this by feeding the Bartók stereo data with the channels both in-phase and 180° out of phase. The result (fig.11) is similar to that with the original HeadRoom Supreme headphone amplifier (footnote 4). When a signal is the same in both channels (cyan and magenta traces), it is shelved up by around 1.5dB below 1kHz, with slightly suppressed highs. For a signal that is identical but out of phase in the two channels, this is equalized in the complementary manner (blue and red traces). Both the dCS and HeadRoom algorithms appear to be based on the work of the late Ben Bauer when he was at CBS Labs in the 1950s (footnote 5), but less extreme.

1019dCSBartfig11-2

Fig.11 dCS Bartók, F1, Headphone output, Crossfeed response at –12dBFS into 100k ohms with data sampled at 44.1kHz with channels in-phase (left channel cyan, right magenta) and 180° out of phase (left blue, right red) (1dB/vertical div.).

Channel separation was superb, at >123dB in both directions below 1kHz, and still 100dB, L–R, and 118dB, R–L, at 20kHz. The low-frequency noise floor was both low in level and almost completely free from any power-supply–related artifacts. When I increased the bit depth from 16 to 24 with a dithered 1kHz tone at –90dBFS (fig.12), the noise floor dropped by 25dB, meaning that the Bartók offers more than 20 bits' worth of resolution, which is close to the state of the art. With undithered data representing a tone at exactly –90.31dBFS, the three DC voltage levels described by the data were well resolved and the waveform was perfectly symmetrical (fig.13). With undithered 24-bit data, the result was a superbly clean sinewave (fig.14).

1019dCSBartfig12

Fig.12 dCS Bartók, spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with: 16-bit data (left channel cyan, right magenta), 24-bit data (left blue, right red) (20dB/vertical div.).

1019dCSBartfig13

Fig.13 dCS Bartók, waveform of undithered 1kHz sinewave at –90.31dBFS, 16-bit data (left channel blue, right red).

1019dCSBartfig14

Fig.14 dCS Bartók, waveform of undithered 1kHz sinewave at –90.31dBFS, 16-bit data (left channel blue, right red).

Harmonic distortion was extremely low in level, even into 600 ohms (fig.15), the lowest impedance dCS recommends using with the Bartók's line-level outputs. The third harmonic was the highest in level in both channels, but at –122dB in the left channel (0.00008%, blue trace) and at –114dB in the right channel (0.0002%, blue trace) this is negligible. Intermodulation distortion was also extraordinarily low, with the second-order difference product produced by equal-level tones at 19 and 20kHz with the combined waveform peaking at 0dBFS, lying at –134dB (0.00002%) (fig.16). Higher-order products are a little higher in level but not so much that it matters. This graph was taken with F1, which has a fast ultrasonic rolloff; repeating the test with F5 and F6 gave identical results. However, with F2, F3, and F4, the aliased images at 24.1kHz and 25.1kHz became increasingly apparent; with F3 and F4, the 0dBFS test signal contaminated the audioband noise floor with aliasing products (fig.17). With F4, I had to reduce the peak signal level by 6dB to obtain a clean noise floor (fig.18). Fortunately, music with top-octave levels approaching 0dBFS is rare.

1019dCSBartfig15

Fig.15 dCS Bartók, spectrum of 50Hz sinewave, DC–1kHz, at 0dBFS into 600 ohms (left channel blue, right red; linear frequency scale).

1019dCSBartfig16

Fig.16 dCS Bartók, F1, HF intermodulation spectrum, DC–30kHz, 19+20kHz at 0dBFS into 100k ohms, 44.1kHz data (left channel blue, right red; linear frequency scale).

1019dCSBartfig17

Fig.17 dCS Bartók, F4, HF intermodulation spectrum, DC–30kHz, 19+20kHz at 0dBFS into 100k ohms, 44.1kHz data (left channel blue, right red; linear frequency scale).

1019dCSBartfig18

Fig.18 dCS Bartók, F4, HF intermodulation spectrum, DC–30kHz, 19+20kHz at –6dBFS into 100k ohms, 44.1kHz data (left channel blue, right red; linear frequency scale).

Tested for its rejection of word-clock jitter with 16-bit TosLink data, the Bartók gave superb results: All the odd-order harmonics of the LSB-level, low-frequency squarewave were at the correct levels (fig.19, sloping green line), and there was very little broadening of the peak that represents the high-level tone at one-quarter the sample rate. With 24-bit J-Test data, the result was a clean noise floor (fig.20).

1019dCSBartfig19

Fig.19 dCS Bartók, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 16-bit TosLink data (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

1019dCSBartfig20

Fig.20 dCS Bartók, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 24-bit TosLink data (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

As with the more-expensive dCS digital processors I have tested, the dCS Bartók offers state-of-the-art measured performance. In this crusty old engineer's view, "dCS" means "Digital Done Right!"—John Atkinson


Footnote 1: Apple's USB Prober utility doesn't recognize Unicode text, just ASCII coded text. The "\363" is the code for ó.

Footnote 2: My thanks to Jürgen Reis of MBL for suggesting this test to me.

Footnote 3: See, for example, fig.6 here.

Footnote 4: See fig.8 here.

Footnote 5: See "Stereophonic Earphones and Binaural Loudspeakers," B.B. Bauer, Journal of the Audio Engineering Society, Vol.9 No.2, April 1961, reprinted in Stereophonic Techniques, published in 1986 by the AES.

COMPANY INFO
Data Conversion Systems, Ltd.
US distributor: Data Conversion Systems Americas, Inc.
Waltham, MA 02454-1443
(617) 314-9296
ARTICLE CONTENTS

COMMENTS
Ortofan's picture

... match the levels by measuring at the speaker terminals with a precision voltmeter.

georgehifi's picture

Am I missing something? or did not the reviewer try going direct to poweramp and set the gain level so the Bartok's volume control was used near full?

Cheers George

Bogolu Haranath's picture

May be Jim Austin could also do a comparison follow-up review with the PS Audio DirectStream DAC with the newest 'Windom' software update? :-) ..........

tonykaz's picture

I applaud folks that try to find something special about any modern DAC.

I tried and failed.

but...

in my systematic hunt I learned about Tube Rolling and how economical it is to make significant changes to the performance of a Pre-Amp by Tube Shopping. Hmm. Art Ferris of Audible Illusions spoke of his Modulus Preamps having great sounding tubes but I wasn't paying attention properly, I had to relearn that basic truth from another Schiit Valhalla 2 Audiophile at an Ann Arbor Michigan Headphone Meet.

I also felt my hearing was defective, so I hired the University of Michigan Audiologists do a rather complete work-up; my hearing response curve tapers off over 8k.

The good news could be that the psychiatrists reveal how a person's brain adjusts and extrapolates sounds to complete the missing bits. ( if allowed and encouraged )

Bob Katz the Mastering Engineer has useful opinions about DACs, as does Jason Stoddard.

For all I know, now, DACs are dam good and won't "Move the Needle" like a Power Cord, PS Audio Power Plant, a good 12ax7, new CAPs in an Vintage AMP., Good Cabling and Class A's sweet sound.

Tony in Venice

ps I'd be surprised if this review read any different.

JBLMVBC's picture

Bartok? Vivaldi? Haut Brion?
I guess the low end of the line will use a Ducon DAC, a National Lampoon turntable, a Burger King amp feeding some American Standard speakers through Pepsi Zero cables...

barrows's picture

from dCS: "He acknowledged that chip DACs Big IC companies, he pointed out, tend to go where the money is. There's a risk that in their efforts to make DAC chips more mobile-friendly—more compact or energy efficient—the unique needs of audiophiles will be sacrificed, or at least neglected"

While the above may be true in some cases, or in the future, it is clearly not the case now for all DAC ICs. If we hold this comment up to the light for a minute, we see that ESS indeed makes some "compromised" DAC chips, designed specifically for mobile uses (the Q2M series I believe), and therefore to run on less power, and perhaps not at the top possible level of sound quality. But none of those things are true of ESS' top of the line DAC chip, the 9038 PRO, which requires stout and very sophisticated power supplies, and is a fairly expensive and non-compromised part. By no means do I intend to suggest that the ESS 9038 Pro is the "best" way to make a DAC, or that discrete implementations have no possible advantages, I am just pointing out that this part is not compromised in any of the ways which dCS' John Quick suggest "could" generally be the case with DAC chips. In any case, I suspect the Bartok, like most of the current offerings from dCS which I have heard, is an excellent sounding DAC.

Jim Austin's picture

While the above may be true in some cases, or in the future, it is clearly not the case now for all DAC ICs.... By no means do I intend to suggest that the ESS 9038 Pro is the "best" way to make a DAC, or that discrete implementations have no possible advantages, I am just pointing out that this part is not compromised in any of the ways which dCS' John Quick suggests "could" generally be the case with DAC chips. In any case, I suspect the Bartok, like most of the current offerings from dCS which I have heard, is an excellent sounding DAC.

I couldn't have said it better myself. John Quick and David Steven's forward-looking point is well-taken, and the Bartok sounds superb. But the fact remains that, for the moment, based on this audition at least, the best chip-DACs appear to be sonically competitive with more expensive-to-produce technologies.

None of this--neither the competitiveness of cheaper technologies nor dCS's response to my review, lessens my respect for the company or the excellent products they produce.

Jim Austin, Editor
Stereophile

Bogolu Haranath's picture

May be JCA could also review the Auralic VEGA G2 streaming DAC ($6,599)? :-) ..........

Bogolu Haranath's picture

AKM also makes some excellent DACs (chips) which are used in several hi-end audio DACs :-) .........

barrows's picture

"None of this--neither the competitiveness of cheaper technologies nor dCS's response to my review, lessens my respect for the company or the excellent products they produce."

Totally agreed Jim! Thanks or the review.

joemariano's picture

Way out of my league for now lol but great review Jim! What you said about A/B comparisons really resonated with me and my way of evaluating my upgrades.

JRT's picture

Based on the review, I am confident that this DAC delivers performance fully adequate to the intended applications.

It is also an expensive DAC, and I think it integrates too much functionality within one box. I don't mind high levels of integration at price levels suitable to short-lived disposable electronics, but for me the price of this well exceeds what I would spend on short-lived disposability.

I would not buy one for the simple reason that it does not separate out the IP streaming functionality to a separate box that can be separately placed in the trash when it soon becomes obsolete, when orphaned out of active technical support and security vulnerabilities are unpatched.

This DAC is priced high enough to afford a separated multi-box solution.

JRT's picture

The known cybersecurity vulnerabilities "Urgent/11" related to orphaned IPnet code mentioned in the links below extend well beyond medical devices to many industrial, defense, security, and consumer electronics devices. I am not implying that this DCS device includes these vulnerabilities, because I do not know that it does or does not. And I have no way of knowing that if it does, that it has been patched or will soon be patched, or if it will have long continued support in this to provide future patches for any future problems, if and when any are discovered. That is why I do not want that functionality integrated in what for me would be an expensive device.

https://www.fda.gov/medical-devices/safety-communications/urgent11-cybersecurity-vulnerabilities-widely-used-third-party-software-component-may-introduce

https://www.armis.com/urgent11/

https://www.us-cert.gov/ncas/current-activity/2019/07/30/cisa-releases-advisory-wind-river-vxworks-platform

https://www.windriver.com/security/announcements/tcp-ip-network-stack-ipnet-urgent11/

Bogolu Haranath's picture

So, if it is hacked, it will start playing Rap and Hip-Hop, when you want to play Classical music ........ It will stream Howard Stern show when you want to listen to Rush Limbaugh ............ Just kidding :-) ........

Robin Landseadel's picture

Is it gonna hook you up to the broadcasts of Madame Psychosis?

Bogolu Haranath's picture

"The man was so cross-eyed, he could stand in the middle of the week and see both Sundays" ......

That would be 'Infinite Jest' :-) ..........

Jose luis's picture

How would you compare it to the Mytek Manhattan 2?

X