Media Server Reviews

I measured the Acoustic Research AR-M2 with my Audio Precision SYS2722 system (see the January 2008 "As We See It"). Source materials were WAV and AIFF test-tone files, as well as DSD64 and DSD128 files of test tones prepared for me by Malcolm Hawksford. The AR-M2's battery was fully charged at the start of the testing.

The maximum output level was 1.89V from the line output jack and 3.71V from the headphone jack, both with the volume set to its maximum. The output impedance was 100 ohms from the line-out jack and 10 ohms from the headphone jack, both figures across the audioband. Though 10 ohms is a little higher than optimal, with headphones like the AudioQuest NightHawk and Audeze LCD-X, both of which have a constant impedance of 22 ohms across the audio band, this will not be an issue other than a reduction in volume. And the AR-M2 played still louder with these headphones than anyone would want. Both outputs inverted absolute polarity, as can be seen from the impulse response with 44.1kHz data (fig.1). This graph also reveals the M2's reconstruction filter to be a conventional FIR type, with symmetrical ringing before and after the single full-scale sample.

Wideband spectral analysis of the player's output while it decoded 44.1kHz-sampled white noise at –4dBFS (fig.2, red and magenta traces) revealed that this filter rolled off the output very rapidly above the Nyquist frequency (half the sample rate, indicated by the green vertical line). As a result, the aliased product at 25kHz of a full-scale tone at 19.1kHz (blue and cyan traces) is suppressed by 105dB. The distortion harmonics associated with this tone are also low in level, all lying below –90dB.

Fig.3 shows the AR-M2's frequency response with data sampled at 44.1, 96, and 192kHz. The responses at the lower two rates follow the same basic shape, with a sharp rolloff just below each Nyquist frequency. With 192kHz data, the output continues to roll off smoothly above 48kHz, reaching –12dB at 95kHz. The responses were identical from both line and headphone outputs. Channel separation (not shown) was >100dB in both directions below 400Hz, but decreased with increasing frequency, reaching 66dB at 20kHz, due to the usual capacitive coupling between the channels somewhere in the output circuitry.

The AR-M2 offered low levels of noise with 24-bit PCM data, though with DSD64 data the audioband noise was about 3dB higher. Increasing the PCM bit depth from 16 to 24 with a dithered 1kHz tone at –90dBFS dropped the noise floor by 10dB (fig.4), suggesting that the player offers close to 18-bit resolution. With an undithered tone at exactly –90.31dBFS (fig.5), the three DC voltage levels described by the data can be seen, along with very slight degrees of DC offset, but the waveform is overlaid with some HF noise.

With 24-bit PCM data and a 600 ohm load, the AR-M2 offered low levels of harmonic distortion (fig.6), with the second harmonic lying at –97dB (0.0014%) and the third at –90dB (0.003%). With DSD64 data, the harmonics were all a little higher in level (fig.7)—I have extended the analysis window to 50kHz to show that while the usual rise in ultrasonic noise with DSD data is present, it remains below –83dB. Intermodulation distortion with an equal mix of high-level tones at 19 and 20kHz was very low (fig.8). However, the left channel (blue trace) has a spurious tone just below 19kHz, and pairs of tones at 12.4/13.4kHz and 25.6/26.6kHz can be seen. Yes, all of these tones are at very low levels, but they are nevertheless unusual; while the 1kHz spacing of each pair is obviously related to the difference between the fundamental tones, the absolute frequencies suggest that these are indeed high-order products.

With 16-bit J-Test data (fig.9), the odd-order harmonics of the low-frequency tones are all very close to their correct levels (green line), and no jitter-related sidebands are present. While a small amount of spectral spreading can be seen at the base of the 11.025kHz tone, this is minimal. With 24-bit J-Test data (fig.10), the noise floor was absolutely clean.

The Acoustic Research AR-M2 offers generally excellent measured performance.—John Atkinson