Musical Fidelity X-DACV3 D/A processor Measurements
I consistently get requests from readers that we publish measurements for the components reviewed in Stereophile's "Sam's Space," "Analog Corner," "Listening," "The Fifth Element," and "Music in the Round" columns. As much as I would like to do this, I don't have the resources to do anything more than scratch the surface of the gear that Sam, Mikey, Art, John, and Kal write about. But I do try to get samples when my interest has been piqued by what they've said. Sam Tellig's report on the Musical Fidelity X-DACV3 in December 2004 was a case in point, especially as Musical Fidelity claims that this $999 D/A processor exceeds the performance of its predecessor, the $2395 Tri-Vista 21.
Popping a sample of the X-DACV3 on my test bench, I measured a maximum output of 2.2V, this to specification and only 0.8dB higher than the CD standard's 2V RMS. Its output impedance was a low 48 ohms, and, unlike Musical Fidelity's X-24K, which I reviewed in February 1999, the X-DAC didn't invert signal polarity. When fed 24-bit data, this appeared on the digital output jack without being truncated, as it had with Musical Fidelity's CD-Pre from a couple of years ago.
The DAC successfully locked on to S/PDIF data with sample rates ranging from 32kHz to 96kHz, and its frequency responses with 44.1kHz and 96kHz data are shown in fig.1. The latter starts very gently to roll off above the audioband, reaching –1dB at 42kHz, and is overlaid by the 44.1k curve up to 20kHz. In common with, sadly, a growing number of modern digital components, the X-DACV3 doesn't apply the necessary equalization to pre-emphasized data, as shown by the massive treble boost applied to the lower pair of traces in fig.1.
Fig.1 Musical Fidelity X-DACV3, frequency response at –12dBFS into 100k ohms, with de-emphasis (bottom) and without (top). (Right channel dashed, 0.5dB/vertical div.)
Channel separation (not shown) was superb, at better than 115dB in both directions at 1kHz. Despite its affordable price, the X-DACV3 extracts superb resolution from its combination of the Crystal SRC4192 upsampling chip and Burr-Brown DSD1792 DAC chip. Fig.2 shows spectral analyses of the Musical Fidelity's analog output, performed with a swept 1/3-octave bandpass filter, while it decoded first 16-bit data (top traces), then 24-bit data (bottom traces), both sets of data representing a dithered 1kHz tone at –90dBFS. Other than a small bump at 120Hz, the X-DACV3's noise floor while it handles 16-bit data is entirely due to the dither used to encode the signal. Increasing the bit depth unmasks spuriae at the full-wave–rectified power-supply frequencies of 120Hz and 240Hz, but as these are both at almost 130dB below full scale, they are insignificant.
Fig.2 Musical Fidelity X-DACV3, 1/3-octave spectrum of dithered 1kHz tone at –90dBFS, with noise and spuriae, 16-bit data (top), 24-bit data (bottom). (Right channel dashed.)
What is much more significant in fig.2 is the fact that the noise floor has dropped by up to 24dB at high frequencies, suggesting that, even with its relatively low analog output voltage, the X-DACV3 offers true 20-bit resolution, which is one of the best I have measured. Perhaps of greater importance, with the exception of the similarly priced Benchmark DAC1, all other digital products that achieve this level of resolution cost very much more than the Musical Fidelity.
Confirming this result, the linearity error plotted with 16-bit data (fig.3, top traces below –100dBFS) really shows only the effect of the encoded dither noise. Extending the word length to 24 bits (fig.3, lower traces) indicates almost zero linearity error to below the –120dBFS limit of this graph. Similarly, the waveform of an undithered 1kHz tone at –90.31dBFS was essentially perfect (fig.4), the three voltage levels clearly defined and symmetrical. Increasing the undithered word length to 24 bits gave a pretty good sinewave (fig.5).
Fig.3 Musical Fidelity X-DACV3, departure from linearity, 16/24-bit data (right channel dashed, 2dB/vertical div.).
Fig.4 Musical Fidelity X-DACV3, waveform of undithered 1kHz sinewave at –90.31dBFS, 16-bit CD data.
Fig.5 Musical Fidelity X-DACV3, waveform of undithered 1kHz sinewave at –90.31dBFS, 24-bit CD data.
Despite the plethora of high-performance op-amp chips available for audio these days, Musical Fidelity's Antony Michaelson sticks with the quarter-century-old NE5534 family of devices for the output stages of his less expensive products. This is because, when properly used, this op-amp can drive low-impedance loads with very low static distortion. The X-DACV3, which uses the dual NE5532 version, is no exception, as revealed by the spectrum of its output while driving a full-scale 1kHz tone into 4k ohms (fig.6). The THD is just 0.0022% (average of the channels, true sum of the harmonics), the harmonics linearly decreasing in amplitude with increasing order. The second harmonic is the highest in level, though at –96dB, this won't bother anyone. Intermodulation distortion was also superbly low (fig.7).
Fig.6 Musical Fidelity X-DACV3, spectrum of 1kHz sinewave, DC–1kHz, at 0dBFS into 4k ohms (linear frequency scale).
Fig.7 Musical Fidelity X-DACV3, HF intermodulation spectrum, DC–25kHz, 19+20kHz at 0dBFS into 4k ohms (linear frequency scale).
Musical Fidelity claims very low jitter for the X-DACV3. Using an older version of the Miller Audio Research Jitter Analyzer that Musical Fidelity uses, I got slightly higher figures than MF's figures: 300 picoseconds peak–peak with 16-bit data fed from my PC with a TosLink connection, 244ps p–p using a coaxial connection from a PS Lambda transport—both figures are still low in absolute terms. Data-related jitter components were at the test signal's residual level with these 16-bit tests; repeating the TosLink measurement with 24-bit data reduced the jitter level to 220ps, due to the elimination of the data-related sidebands. Fig.8 shows the spectrum of the X-DACV3's output taken under these circumstances. The two highest-level sideband pairs lie at ±953Hz and ±1200Hz (purple "6" and "7" markers, respectively); I have no idea where these come from. No sidebands are present at the otherwise ubiquitous ±15.6Hz frequencies, but a sideband pair is evident at the power-supply–related frequencies of ±120Hz (blue "1").
Fig.8 Musical Fidelity X-DACV3, high-resolution jitter spectrum of analog output signal (11.025kHz at –6dBFS sampled at 44.1kHz with LSB toggled at 229Hz). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.
Considered overall, Musical Fidelity's X-DACV3 offers stunningly good measured performance that is even more commendable when you consider its very affordable price.—John Atkinson