dCS Rossini Apex D/A processor Measurements

Sidebar 3: Measurements

I measured the dCS Rossini Apex with my Audio Precision SYS2722 system, repeating some measurements with the higher-performance APx500. I performed the testing with the serial data and USB inputs then repeated some of them with JVS's preferred Network input, using Roon.

The AES3 and coaxial S/PDIF inputs accepted data sampled at all rates up to 192kHz. TosLink was restricted to 96kHz data. I didn't test the Rossini's dual-AES3 input, but as with earlier dCS processors, it will accept data sampled at rates up to 384kHz. Apple's AudioMIDI utility revealed that the dCS Rossini Apex accepted 16- and 24-bit integer data via USB sampled at all rates from 44.1kHz to 384kHz. Apple's USB Prober app identified the Rossini Apex as "dCS Rossini DAC USB class 2" from "Data Conversion Systems Ltd" and confirmed that the USB port operated in the optimal isochronous asynchronous mode.

The dCS Rossini Apex's analog outputs preserved absolute polarity (ie, were noninverting) from all the digital inputs. The maximum output can be set to "6V," "2V," "0.6V," and "0.2V." With full-scale 1kHz data and the volume control set to its maximum, I measured 5.95V, 2.014V, 594.7mV, and 201.3mV from the balanced outputs and very slightly lower voltages from the single-ended outputs. The balanced output impedance was an extraordinarily low 2 ohms from 20Hz to 20kHz. The single-ended output impedance was 51 ohms, again at all audio frequencies.

The Rossini Apex offers a choice of six reconstruction filters. All six filters are functional with data having a sample rate of 44.1, 48, 176.4, and 192kHz, but only the first four (F1–F4) operate with data sampled at 88.2 and 96kHz. (In this respect, the Rossini Apex is identical to the dCS Vivaldi D/A processor that Michael Fremer reviewed in January 2014.) The behavior of these filters was the same as that of the Vivaldi's filters. Fig.1 shows the F1 filter's impulse response with 44.1kHz data. (F6's impulse response was identical.) It is typical of a linear-phase reconstruction filter, with equal amounts of ringing before and after the single sample at 0dBFS. Filters F2, F3, and F4 also had linear-phase impulse responses but with progressively smaller amounts of ringing. F5 was different. Its impulse response was a minimum-phase type, with all the ringing following the single full-scale sample (fig.2).

922dCSRossfig01

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

922dCSRossfig02

Fig.2 dCS Rossini Apex, F5, impulse response (one sample at 0dBFS, 44.1kHz sampling, 4ms time window).

With 44.1kHz white-noise data (footnote 1), F1, F5, and F6 rolled off rapidly above the audioband (fig.3, magenta and red traces), reaching full stop-band attenuation at 22.05kHz. They are therefore apodizing types. F2, F3, and F4 offered progressively slower ultrasonic rolloffs with 44.1kHz data, with F4 not reaching full stop-band attenuation until 30kHz (fig.3). With a 19.1kHz tone at –3dBFS (cyan, blue; with the tone at 0dBFS, many aliasing products were present in the audioband), the slow rolloff means that the aliased image at 25kHz is only suppressed by 12dB. The harmonics associated with the 19.1kHz tone were all extremely low in level, however. Figs.3 and 4 were taken with DXD upsampling. Changing to DSD upsampling gave a sharp upward slope in the ultrasonic noisefloor (fig.5). DSD2 upsampling also gave a rise in ultrasonic noise (fig.6), but with a less extreme upward slope.

922dCSRossfig03

Fig.3 dCS Rossini Apex, F1 & DXD upsampling, wideband spectrum of white noise at –4dBFS (left channel red, right magenta) and 19.1kHz tone at 0dBFS (left blue, right cyan) into 100k ohms with data sampled at 44.1kHz (20dB/vertical div.).

922dCSRossfig04

Fig.4 dCS Rossini Apex, F4 & DXD upsampling, wideband spectrum of white noise at –4dBFS (left channel red, right magenta) and 19.1kHz tone at –3dBFS (left blue, right cyan) into 100k ohms with data sampled at 44.1kHz (20dB/vertical div.).

922dCSRossfig05

Fig.5 dCS Rossini Apex, F6 & DSD upsampling, wideband spectrum of white noise at –4dBFS (left channel red, right magenta) and 19.1kHz tone at 0dBFS (left blue, right cyan) into 100k ohms with data sampled at 44.1kHz (20dB/vertical div.).

922dCSRossfig06

Fig.6 dCS Rossini Apex, F6 & DSD2 upsampling, wideband spectrum of white noise at –4dBFS (left channel red, right magenta) and 19.1kHz tone at 0dBFS (left blue, right cyan) into 100k ohms with data sampled at 44.1kHz (20dB/vertical div.).

The F1 filter's frequency response with data sampled at 44.1, 96, and 192kHz (fig.7) was flat to just below half of each sample rate, with then a fast rolloff. The rolloffs were slower and started progressively earlier with F2, F3, and F4 (fig.8). F5, which I understood from JVS was his preferred filter for 44.1kHz data, gave a sharp rolloff at that sample rate (fig.9, green and gray traces) but a slower rolloff with 192kHz data (blue, red) that reached –6dB at 40kHz. F6 behaved identically to F5 in this respect.

922dCSRossfig07

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

922dCSRossfig08

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

922dCSRossfig09

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

Channel separation (not shown) was superb, at >125dB in both directions below 1kHz decreasing to a still excellent 113dB at the top of the audioband. The low-frequency noisefloor (fig.10) was free from any power supply–related spuriae, with a very low level of random noise.

922dCSRossfig10

Fig.10 dCS Rossini Apex, balanced output, spectrum with noise and spuriae of dithered 1kHz tone at 0dBFS with 24-bit TosLink data (left blue, right red) (20dB/vertical div.).

The red trace in fig.11 plots the error in the analog output level as a 24-bit, 1kHz digital tone steps down from 0dBFS to –140dBFS. (The Rossini Apex was set to Map 1, its output level to 6V, and the volume control set to its maximum for this and the next three measurements.) The amplitude error is negligible until the signal lies below –135dBFS, which implies very high resolution. Repeating the test with Map 2 and Map 3 gave virtually identical results. An increase from 16 bits to 24 bits with dithered data representing a 1kHz tone at –90dBFS (fig.12) dropped the dCS Rossini Apex's noisefloor by 26dB, which implies a very high resolution of between 20 and 21 bits. When I played undithered data representing a tone at exactly –90.31dBFS, the waveform was symmetrical, with the three DC voltage levels described by the data cleanly resolved (fig.13). Repeating the measurement with undithered 24-bit data gave a well-formed sinewave (fig.14).

922dCSRossfig11

Fig.11 dCS Rossini Apex, Map 1, left channel, 1kHz output level vs 24-bit data level in dBFS (blue, 20dB/vertical div.); linearity error (red, 1dB/small vertical div.).

922dCSRossfig12

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

922dCSRossfig13

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

922dCSRossfig14

Fig.14 dCS Rossini Apex, waveform of undithered 1kHz sinewave at –90.31dBFS, 24-bit data (left channel blue, right red).

JVS told me that he preferred the Ring DAC's Map 1 with network data. I repeated the low-level resolution tests with data sourced over my network from Roon and with Maps 2 and 3, but I didn't find any significant differences.

The dCS Rossini Apex produced very low levels of distortion—so low, in fact, that it challenged the SYS2722's resolution (fig.15). I therefore used the higher-resolution APx500 analyzer to examine the harmonic distortion. With a full-scale 1kHz tone and the Rossini set to output its highest level of 6V, the THD+noise measured just 0.00026%! The second harmonic was the highest in level, at a vanishingly low –124dB (fig.16), and though some higher-order harmonics were visible in this graph, they all lie at or below –130dB. This spectrum was taken with TosLink data and a load of 100k ohms; the levels of the harmonics were the same with network data and didn't increase by any significant amount when I reduced the load impedance to 600 ohms.

922dCSRossfig15

Fig.15 dCS Rossini Apex, balanced output, 24-bit TosLink data, spectrum of 50Hz sinewave, DC–1kHz, at 0dBFS into 100k ohms (left channel blue, right red; linear frequency scale).

922dCSRossfig16

Fig.16 dCS Rossini Apex, balanced output, 24-bit TosLink data, spectrum of 1kHz sinewave, DC–1kHz, at 0dBFS into 100k ohms (left channel blue, right red; linear frequency scale).

Intermodulation distortion with an equal mix of 19 and 20kHz tones with a peak level of 0dBFS was vanishingly low in level (fig.17). This graph was taken with DXD upsampling and the fast-rolloff F6 reconstruction filter. Aliased images of the primary tones appeared above the audio-band with the slowest-rolloff F4 filter, but actual intermodulation products were still as low as they had been with F6.

922dCSRossfig17

Fig.17 dCS Rossini Apex, balanced output, F6, 24-bit TosLink data, HF intermodulation spectrum, DC–30kHz, 19+20kHz at 0dBFS into 100k ohms, 44.1kHz data (left channel blue, right red; linear frequency scale).

Fig.18 shows the spectrum of the Rossini Apex's output when it was fed high-level, optical, undithered, 16-bit J-Test data. The odd-order harmonics of the undithered low-frequency, LSB-level squarewave lie at the correct levels, indicated by the sloping green line, and the noisefloor between the sidebands is extremely low in level. Repeating the test with 24-bit J-Test data (not shown) gave a similarly superb result.

922dCSRossfig18

Fig.18 dCS Rossini Apex, 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.

Overall, the dCS Rossini Apex's measured performance was beyond reproach.—John Atkinson


Footnote 1: My thanks to Jürgen Reis of MBL for suggesting this means of displaying the performance of a DAC's reconstruction filter.
COMPANY INFO
dCS (Data Conversion Systems), Ltd.
US distributor: Data Conversion Systems Americas, LLC
PNC Bank Bldg.
300 Delaware Ave., Suite 210, Wilmington, DE 19801
(302) 473-9050
ARTICLE CONTENTS

COMMENTS
georgehifi's picture

If someone showed that pcb to me
https://www.stereophile.com/images/922dcs.ins.jpg
And said it was an R2R ladder dac, I'd believe them. Please enlighten me, is a RingDac dac, R2R based??? I never saw R2R dac chip or discrete in Arcam's DCS ring dac players when replacing laser.

Cheers George

Archimago's picture

Based on this (and also the article PDF):
https://dcs.community/t/dcs-ring-dac-a-technical-explanation/2724/20

The dcs RingDAC technology appears to be a multibit (5-bit) SDM system upsampled to rates like 2.8-6.1MHz depending on the model. There's a first step oversampling to 768kHz which is presumably also where the digital filtering options are applied.

With faster technology over time, these numbers can certainly increase with each generation. Not sure if necessarily needed or will imply better sound quality. Objective results look excellent already!

miguelito's picture

https://dcs.community/t/dcs-ring-dac-a-technical-explanation/2724

Long-time listener's picture

Given that DACs, such as the $700 Topping D90SE, routinely reach 21 bits of resolution (with one or two reviewed in Stereophile having 22), what is it about this particular DAC that justifies it being priced at more than ten times that amount?

georgehifi's picture

Would be real interesting to do a blind A/B with those two.
However I do with Redbook PCM CD replay believe in R2R dacs, they just sound more "fleshed out" to me.

Cheers George

ok's picture

the analog stage - which is the sole object of the apex update - is the most important part of a dac; thankfully dcs realized it at last.

miguelito's picture

I am told the upgrade involved three things:
1- Better current sources to the Ring DAC
2- Improved trace locations in the Ring DAC board
3- Improved analog output stage

miguelito's picture

Best I can say is just go for a listen. If you cannot tell the difference, then you can save a lot of money. I just upgraded my Rossini to Apex and can very much tell the difference. YMMV.

David Harper's picture

Or at least you imagine that you can. Placebo is a powerful thing.

ChrisS's picture

...skills, a highly resolving stereo system, a nice room, and great music!

X