Media Server Reviews

I examined the Astell&Kern AK100's electrical performance with Stereophile's loan sample of the top-of-the-line Audio Precision SYS2722 system (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. I made sure the AK100's battery was fully charged before performing the tests. For source signals, I both used WAV files stored in the player's flash memory and fed it S/PDIF datastreams via its TosLink input. The TosLink input successfully locked to data with sample rates of 44.1, 48, 88.2, and 96kHz, but not 176.4 or 192kHz. The maximum output level at 1kHz was very slightly higher than the specified 1.5V at 1.55V, and the player preserved absolute polarity for both internal and TosLink-sourced data. The output impedance was 22.5 ohms at all audio frequencies, including 6' of interconnect.

Fig.1 shows wideband spectra of the AK100's output while it decoded data sampled at 44.1kHz representing a full-scale 19.1kHz tone (cyan and red traces) and white noise at –4dBFS (blue and magenta). The white-noise spectrum rolls off rapidly above 22kHz, revealing that the AK100 uses a conventional reconstruction filter. An aliasing product of the 19.1kHz tone is visible at 25kHz, but this is suppressed by 84dB. These spectra were taken into the high 100k ohms load; the second and third harmonics of the 19.1kHz tone can be seen at –103dB (0.0007%) and –76dB (0.015%), respectively. The AK100's frequency responses with 44.1kHz data (fig.2, green and gray traces), 96kHz data (cyan and blue), and 192kHz data (blue and red) indicate that the output rolls offs rapidly just below half the sample rate and that the passband is free from ripples. Considering that this is a small, battery-powered device, the AK100's channel separation was superb, at >115dB in both directions below 1.5kHz (fig.3). The usual capacitive coupling between channels reduces the separation to 93dB at the top of the audioband, which is still excellent.

For consistency with my tests of digital products going back to 1989, I estimate a DAC's resolution by feeding it dithered data representing a 1kHz tone at –90dBFS with 16- and 24-bit word lengths and sweeping a 1?3-octave bandpass filter from 20kHz down to 20Hz. The result is shown in fig.4, with the 16-bit spectrum the top pair of traces, the 24-bit spectrum the middle pair. The traces are free from harmonic products and the increase in bit depth drops the noise floor by up to 15dB, suggesting resolution between 18 and 19 bits, which is excellent considering the low maximum output level. This is readily enough resolution to allow the AK100 to resolve a dithered 24-bit tone at –120dBFS (bottom traces). Fig.5 repeats this analysis using a modern FFT technique, which unmasks some very low-level 120Hz components with the 24-bit data (blue and red traces). As the AK100 is battery-powered, I have no idea where these arise, but they were also present with FFT analysis of a dithered tone at –120dBFS (fig.6).

The AK100's reproduction of an undithered 16-bit tone at exactly –90.31dBFS is shown in fig.7. The three DC voltage levels described by these data are clearly resolved, with a symmetrical waveform. Increasing the bit depth to 24 gives a recognizable, if noisy, sinewave (fig.8).

Although fig.1 indicates that the third was the most prominent distortion harmonic at high frequencies into high impedances, the second was the highest in level at low frequencies into 600 ohms (fig.9). Fig.10 compares the spectra of the AK100's output while it reproduced a 1kHz tone into 100k ohms (blue trace) and a low 100 ohms (red), the latter more typical of headphones. Again, into the high impedance the third and fifth harmonics are the highest in level, with both the second harmonic dominating into the low impedance and all the harmonics higher in level. But even the second harmonic into the low impedance lay at –79dB (0.01%), which is low in absolute terms. The picture was similar with intermodulation distortion (fig.11). While the high-order intermodulation products around the 19 and 20kHz fundamentals are unchanged by the drop in load impedance, the difference product at 1kHz rises from –120dB (0.0001%) to a still-low –86dB (0.005%).

I assessed the AK100's rejection of jitter using both internally stored WAV versions of the 16- and 24-bit Miller-Dunn J-Test signal and the same data fed to the AK100's TosLink input. The results were identical for both operational modes and are shown in fig.12, the 16-bit spectrum represented by the blue and magenta traces, the 24-bit spectrum by the cyan and red traces. With 16-bit data, the odd-order harmonics of the Fs/192 LSB-level squarewave all lie at the correct levels and are not being accentuated by the AK100. These disappear with the 24-bit data, of course, but a single tone of unknown origin can be seen at 8.2kHz. Perhaps of most significance, with both 16- and 24-bit data, the spike that represents the high-level Fs/4 tone has a considerable degree of spectral spreading at its base, suggesting the presence of random low-frequency jitter.

Overall, the Astell&Kern AK100 measures very well, not only as a standalone high-resolution player, but also as an affordable high-resolution D/A processor.—John Atkinson