CEntrance DACmini CX D/A converter Measurements
Sidebar 3: Measurements
I used Stereophile's loan sample of the top-of-the-line Audio Precision SYS2722 system to measure the CEntrance DACmini CX (see www.ap.com and the January 2008 "As We See It"); for some tests, I also used my vintage Audio Precision System One Dual Domain and the Miller Audio Research Jitter Analyzer. As well as driving the DACmini with S/PDIF data from the Audio Precision analyzers, I used a MacBook running Mac OS10.6.8 and Pure Music 1.8 to play test-signal files. I repeated some of the USB input testing using a PC running Windows XP and CEntrance's ASIO driver/control panel program, but there were no differences to be found.
The DACmini successfully locked to data with sample rates ranging from 32 to 192kHz via its coaxial and optical S/PDIF inputs. Tested via the USB input, the DACmini correctly handled data with sample rates of 44.1, 48, 88.2, and 96kHz, and with depths of 16 and 24 bits. The Mac's USB Prober utility identified the product as the "CEntrance DACmini" from "CEntrance Inc.," and indicated that the USB interface operated in the usual isochronous adaptive mode.
Our review sample of the DACmini CX was fitted with the variable line-level output and analog line-level input options, as well as the headphone output. The maximum output level from both line and headphone outputs with S/PDIF data was 5.22V, and both outputs preserved absolute polarity (ie, were non-inverting). The output impedance from the line output jacks was a little higher than specified, though still usefully low, at 40 ohms across the audioband, but was the specified 10 ohms from the headphone jack, again at all frequencies.
Fig.1 shows the DACmini's frequency response tested via S/PDIF with data sampled at 44.1kHz (green and gray traces), 96kHz (cyan and magenta traces), and 192kHz (blue and red traces). As you can see, the expected increase in bandwidth from the 192kHz sampling is absent, the response being the same for 192kHz data as for 96kHz data. The DACmini uses an asynchronous sample-rate converter chip (an Analog Devices AD1896), so I suspect that, like the Benchmark DAC1, the DAC chip is fed data at a fixed rate that, although sufficient for 96kHz data, is less than 192kHz. Channel separation via the digital inputs (not shown) was superb, at 103dB at low frequencies from both the line and headphone outputs, and still 84dB (RL) and 90dB (LR) at 20kHz.
For consistency with my measurements of digital products going back more than two decades, I assess a product's resolution by sweeping a 1/3-octave bandpass filter from 20kHz to 20Hz while it decodes dithered data representing a 1kHz tone at 90dBFS. The top pair of traces in fig.2 were taken with 16-bit data; no power-supply or distortion products can be seen, and the spectrum is dominated by the dither noise. Increasing the word length to 24 bits (middle traces) drops the noise floor in the midrange and treble by 13dB or more, implying resolution of at least 18 bits: enough to allow the DACmini to readily resolve a tone at 120dBFS (bottom traces). FFT analysis (fig.3) confirms the DACmini's excellent resolution and freedom from spuriae.
Linearity error with 16-bit data (not shown) was negligible to below 105dBFS; that and the low level of noise in the analog output allowed the DACmini to readily resolve the three DC voltage levels that are described by data representing an undithered tone at 90.31dBFS (fig.4). With 24-bit data, an undithered tone at the same level gave a good sinewave (not shown).
With some products that use an asynchronous sample-rate converter, mathematical limitations in the chip give rise to noise modulation; ie, the level of the noise floor varies with signal level. However, while the DACmini's noise floor with 0dBFS data is higher than it is with data at 40dBFS and 60dBFS (fig.5), the variation is not as great as it has been with some other products.
The variation in noise can also be seen in the spectra of the DACmini reproducing a 50Hz tone at 0dBFS (fig.6) and 10dBFS (fig.7), but more important, these graphs show that the spectral composition of the DACmini's harmonic distortion is both primarily low-order and extremely low in level. Intermodulation spuriae were also very low in level, the difference products from an equal mix of 19 and 20kHz tones lying at 100dB (0.001%, fig.8). These graphs were taken from the line outputs into 100k ohms; the picture didn't change into 600 ohms, or from the headphone output.
Even though it operates in the theoretically more jitter-prone isochronous adaptive mode, the DACmini offered superb rejection of jitter for S/PDIF data. With the DACmini tested with a 16-bit version of the 44.1kHz Miller-Dunn J-Test signal via a 15' length of plastic TosLink, the only sidebands visible to either side of the 11.025kHz tone lie at the level of the residual odd harmonics of the low-frequency tone (fig.9, cyan and magenta traces). With 24-bit J-Test data (fig.9, blue and red traces), no spuriae can be seen above the noise floor. However, this graph was taken with the DACmini driven by the AP SYS2722's digital output, which has very low jitter. Repeating the test using the same 15' length of plastic TosLink, but from the RME soundcard fitted to the PC in which the Miller Analyzer resides, increased the level of the sidebands at ±229Hz to give an estimated jitter of 440 picoseconds. Though this is still low in absolute terms, the difference in measured jitter between the Audio Precision and RME sources suggests that the DACmini does have some sensitivity to the source in use via S/PDIF. Fed data via USB, the jitter spectrum was commendably clean (fig.10).
The DACmini's analog inputs don't appear to be digitized, and offer a maximum gain of 8.15dB. The unity-gain setting of the volume control was 2:00. The input impedance was 10k ohms at all audio frequencies, and the analog input preserved absolute polarity. The response was flat from 10Hz to 200kHz, and was not affected by the volume-control setting. However, a channel imbalance of 0.25dB crept in at volume-control settings of 1:00 and below. In all other respectschannel separation, harmonic distortion, and intermodulation distortionthe performance of the DACmini from its analog input was not appreciably different from that via its digital inputs.
Although its measured performance includes some idiosyncrasies, the CEntrance DACmini CX is, overall, a well-engineered component.John Atkinson