Cambridge Audio Azur DacMagic D/A converter Measurements
Sam Tellig was so impressed by this $449 D/A processor that he sent me the review sample (which he subsequently bought), so I could examine its measured performance. I used the Audio Precision SYS2722 system (see www.ap.com and "As We See It" in the January 2008 issue), as well as, for some tests, our Audio Precision System One.
The DacMagic's maximum output was 2.1V from its unbalanced RCA jacks, 0.43dB higher than the CD standard's 2V, which will be audible in direct comparisons. The balanced XLR output was exactly twice this voltage, as expected, and both sets of outputs were non-inverting with the blue front-panel polarity LED unlit. (The XLRs are wired with pin 2 hot.) The source impedance was very low, at 47 ohms at all frequencies from the RCAs and, again, twice that figure from the XLRs.
The processor's frequency response was flat within the audioband with all three reconstruction filter settingsSteep, Linear Phase, Minimum Phaseall of which featured the same 0.18dB droop at 20kHz. This rolloff gently continued with 96kHz-sampled data, reaching 0.7dB at 40kHz, above which the output dropped like a stone (fig.1). Although it is hard to see in this graph, there is a very slight amount of passband ripple with the Linear setting. Channel separation (not shown) was superb, at better than 110dB in both directions below 16kHz from both the balanced and unbalanced outputs.
Fig.1 Cambridge DacMagic, frequency response at 12dBFS into 100k ohms with 96kHz-sampled PCM data and Linear Phase filter (left channel blue, right red), Minimum Phase filter (left green, right magenta), and Steep filter (left cyan, right gray). (2dB/vertical div.)
For reasons of consistency, my primary examination of a digital component's dynamic range is to sweep a 1/3-octave-wide filter down from 20kHz to 20Hz while the component decodes a low-level dithered tone. Fig.2 shows the spectra produced in this manner while the DacMagic decoded 16- and 24-bit data representing a 1kHz tone at 90dBFS. The increase in bit depth drops the noise floor by up to 15dB, which can also be seen in FFT-derived plots of the same signals (figs. 3 and 4). This would suggest that the DacMagic offers almost 19-bit resolutionexcept that the 24-bit representation (fig.4) has some harmonic components visible, and there is actually frequency doubling evident with 24-bit signals lower than 90dBFS (fig.5). It looks as if the Wolfson WMB8740 DACs used by the DacMagic have some code errors at the lowest signal levels, though this may well not be an issue with CD program. Linearity error with 16-bit data (not shown) was minimal down to below 105dBFS, which is excellent.
Fig.2 Cambridge DacMagic, 1/3-octave spectrum with noise and spuriae of dithered 1kHz tone at 90dBFS with 16-bit data (top) and 24-bit data (bottom). (Right channel dashed.)
Fig.3 Cambridge DacMagic, FFT-derived spectrum with noise and spuriae of dithered 1kHz tone at 90dBFS with 16-bit data (left channel blue, right red).
Fig.4 Cambridge DacMagic, FFT-derived spectrum with noise and spuriae of dithered 1kHz tone at 90dBFS with 24-bit data (left channel blue, right red).
Fig.5 Cambridge DacMagic, FFT-derived spectrum with noise and spuriae of dithered 1kHz tone at 120dBFS with 24-bit data (left channel blue, right red).
Probing the DacMagic's USB data input with a utility program, the Cambridge identified itself to the host computer as "C-Media USB Headphone Set," and confirmed that the only sample rates supported were 44.1 and 48kHz, and that the maximum bit depth it could handle was 16. If I sent 24-bit data to the DacMagic from the Mac mini I use as my music server, these data were truncated to 16 bits.
With its low audioband noise and excellent linearity, the DacMagic produced an essentially perfect representation of an undithered sinewave at exactly 90.31dBS, with the three DC voltages clearly visible and the waveform symmetrical about the time line (fig.6). This graph was taken with the Linear filter (the Steep filter was identical); with the Minimum Phase filter, the fact that all the ringing now occurs after the transient can be seen (fig.7).
Fig.6 Cambridge DacMagic, Linear Phase filter, waveform of undithered 1kHz sinewave at 90.31dBFS, 16-bit data (left channel blue, right red).
Fig.7 Cambridge DacMagic, Minimum Phase filter, waveform of undithered 1kHz sinewave at 90.31dBFS, 16-bit data (left channel blue, right red).
The DacMagic's analog output stage produced very low levels of distortion, with only the low-order second, third, fourth, and fifth harmonics visible above the 24-bit noise floor with a full-scale tone into 100k ohms (fig.8). Dropping the load to a punishing 600 ohms didn't change the picture by much, all the harmonics remaining below 106dB (0.0005%, not shown). This was with PCM data; feeding 16-bit data to the DacMagic's USB input gave a dirtier-looking noise floor, with jitter-related sidebands visible around the fundamental tone (fig.9). Intermodulation distortion was also very low (fig.10), though a very faint trace of noise modulation is evident.
Fig.8 Cambridge DacMagic, balanced output, spectrum of 1kHz sinewave at 0dBFS into 100k ohms, 24-bit data (left channel blue, right red; linear frequency scale).
Fig.9 Cambridge DacMagic, balanced output, spectrum of 1kHz sinewave at 0dBFS into 100k ohms, 16-bit USB data (left channel blue, right red; linear frequency scale).
Fig.10 Cambridge DacMagic, Steep filter, balanced output, HF intermodulation spectrum, 19+20kHz at 0dBFS peak into 100k ohms, 24-bit data (left channel blue, right red; linear frequency scale).
Driving the DacMagic with PCM data representing the Miller-Dunn diagnostic signal gave an output spectrum that really shows only the residual harmonics of the low-frequency squarewave (fig.11). The DacMagic obviously features superb jitter rejection via its conventional data inputs. The USB input, however, performed significantly worse on this test, with both a raised noise floor and significant sidebands apparent (fig.12).
Fig.11 Cambridge DacMagic, high-resolution jitter spectrum of analog output signal, 11.025kHz at 6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz, 16-bit data. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red).
Fig.12 Cambridge DacMagic, high-resolution jitter spectrum of analog output signal, 11.025kHz at 6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz, 16-bit USB data. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red).
Although its USB input is really of only utility quality and shouldn't be used for serious listening, the Cambridge Azur DacMagic otherwise offers superb measured performance. In fact, considering its street price of $400, this level of performance is astonishing. I am not surprised that Sam Tellig liked the DacMagic's sound. However, I am surprised that he preferred the Linear Phase filter setting; from my experience of the Meridian 808i.2 CD player, which also offers a minimum-phase reconstruction filter, I would have thought he might prefer the DacMagic's Minimum Phase setting. Perhaps there are other differences between the filters in the Cambridge and Meridian products.John Atkinson