Computer Audio Reviews

I examined the measured behavior of the ASUS Xonar Essence ST and Essence STX using 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, my Audio Precision System One Dual Domain. I performed a complete set of tests on the STX card, which was installed in an AMD Athlon-powered PC running Vista Home Premium, and repeated some of the tests on the ST card, which was installed in an older Pentium-powered PC running XP. Test-tone WAV files were played back using Adobe Audition 3.0 and Foobar 2000, with ASIO4 selected as the default audio device.

The card would optimally play files only with sample rates of 44.1, 48, 96, or 192kHz. When the card's S/PDIF output was selected, its sample rate followed that of the file being played (ASIO4) or that selected with the Xonar Audio Center. Both the ST and STX cards passed 24-bit data when set appropriately.

The maximum output level from the line outputs was 2.16V; from the headphone jack, it depended on the setting chosen: 885mV (0dB, for headphones with impedances below 64 ohms), 3.52V (+12dB for headphones with impedances of 64–300 ohms), and 7.03V (+18dB for headphones with impedances of >300 ohms). Peculiarly, this maximum level remained the same if the Playback Level was increased to its maximum (12dB) in the Mixer panel of the Xonar Control Center. All that happened was that signals below the selected gain were amplified, but signals at or above it were unaffected; ie, with 12dB of gain selected, signals at or below –12dB were boosted by 12dB to 0dB; signals above –12dB were hard-limited to the same 0dB. The line output impedance was a moderately low 99 ohms at all frequencies; the headphone output impedance was 10.7 ohms at all frequencies and settings. All the outputs preserved absolute polarity; ie, were non-inverting.

The STX's frequency responses at sample rates of 44.1, 96, and 192kHz are shown in fig.1. A small amount of rolloff (–0.25dB at 20kHz) with 44.1kHz data continues above the audioband with files of higher sample rate, the only difference between the 96 and 192kHz rates being that the former drops like a stone above 46kHz. Channel separation (not shown) was superb, at 120dB at all frequencies.

Fig.1 ASUS Xonar Essence STX, D/A frequency response at –12dBFS into 100k ohms with sample rate set to 192kHz (left channel cyan, right magenta), 96kHz (left blue, right red), 44.1kHz (left green, right gray). (0.25dB/vertical div.)

Fig.2 shows my usual 1/3-octave swept-bandpass analysis of the STX's line output while the card decoded dithered data representing a 1kHz tone at –90dBS with 16- and 24-bit word lengths, and a dithered 1kHz tone at –120dBFS with 24-bit data. Other than an almost negligible component in the left channel at 60Hz, the traces are free from any power-supply–related spuriae. The increase in bit depth drops the noise floor by 15dB, implying a resolution approaching 19 bits, which is excellent considering the noisy RF environment inside a computer. Fig.3 shows a similar spectral analysis, this time performed with an FFT technique and plotted on a linear rather than a logarithmic scale, which is why the noise floor is now flat with frequency rather than gently sloping up. The same 15dB drop in the noise floor is evident, and now some harmonics have been unmasked in the left channel. As in fig.2, the left channel is a little noisier than the right below 1kHz.

Fig.2 ASUS Xonar Essence STX, 1/3-octave spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with 16-bit data (top) and 24-bit data (middle at 2kHz), and dithered 1kHz tone at –120dBFS with 24-bit data (bottom at 1kHz). (Right channel dashed.)

Fig.3 ASUS Xonar Essence STX, FFT-derived spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with 16-bit data (left channel cyan, right magenta) and 24-bit data (left channel blue, right red).

With 16-bit data, the linearity-error test was dominated by the recorded dither noise, so I haven't shown the result. I couldn't replicate the manufacturer's claim of an A-weighted signal/noise ratio of 124dB for the Xonar Essence. However, ref. 1V output, I measured 114dB for the STX, which is still superb, but 102dB for the ST. It's possible that the actual ratio will depend on the computer in which the card is installed. Even so, the STX's level of noise was so low that the waveform of an undithered 16-bit sinewave at exactly –90.31dBFS was superbly well defined, with both the three DC voltage levels and the Gibbs' Phenomenon "ringing" readily apparent, and the waveform beautifully symmetric about the time axis (fig.4). Increasing the word length to 24 bits gave a superbly defined sinewave (fig.5).

Fig.4 ASUS Xonar Essence STX, D/A waveform of undithered 1kHz sinewave at –90.31dBFS, 16-bit data (left channel blue, right red).

Fig.5 ASUS Xonar Essence STX, D/A waveform of undithered 1kHz sinewave at –90.31dBFS, 24-bit data (left channel blue, right red).

The Essence STX's line output stage offered very low levels of harmonic distortion. Fig.6, for example, shows the spectrum of its output reproducing a full-scale 50Hz sinewave into the demanding 600 ohm load. The third harmonic is the highest in level in the left channel (blue trace), the second in the right (red), but all harmonics are at or below –110dB (0.0003%), which is superb.

Fig.6 ASUS Xonar Essence STX, line output, spectrum of 50Hz sinewave at 0dBFS into 600 ohms, 24-bit data (left channel blue, right red; linear frequency scale).

From the headphone jack, the THD depended on the setting. With the low-gain setting for low-impedance 'phones, all distortion harmonics other than the third were buried beneath the noise floor, even into 300 ohms (fig.7). And in the worst case, the +18dB setting into 30 ohms with a full-scale signal, all the harmonics were still below –90dB (0.003%, fig.8). With the high-level, high-frequency intermodulation test, actual intermodulation spuriae from the line outputs were all at or below –110dB, even into 600 ohms (fig.9). However, some odd scalloping of the noise floor is apparent, suggesting that something is not quite right.

Fig.7 ASUS Xonar Essence STX, headphone output set to 0dB, spectrum of 1kHz sinewave at 0dBFS into 300 ohms, 24-bit data (left channel blue, right red; linear frequency scale).

Fig.8 ASUS Xonar Essence STX, headphone output set to +18dB, spectrum of 1kHz sinewave at 0dBFS into 300 ohms, 24-bit data (left channel blue, right red; linear frequency scale).

Fig.9 ASUS Xonar Essence STX, line output, HF intermodulation spectrum, 19+20kHz at 0dBFS peak into 100k ohms, 24-bit data (left channel blue, right red; linear frequency scale).

Word-clock jitter in the STX's TosLink S/PDIF output, measured directly with the AP SYS2722 and a 44.1kHz datastream, was low at 1.48 nanoseconds, which is close to the system's resolution limit. Jitter in the analog output, assessed with the Miller Audio Research analyzer, was very low at 170 picoseconds left and 186ps right. However, this was mainly due to the Miller Analyzer's own noise floor obscuring the picture. (The National Instruments PCI card on which the Analyzer is based uses 16-bit A/D converters.) Graphed with the Audio Precision SYS2722, which has 24-bit converters, a number of sideband pairs can be seen (fig.10); only the highest in level, at ±1.63 and ±3.26kHz, were acknowledged by the Miller Analyzer.

Fig.10 ASUS Xonar Essence STX, line output, 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).

I then turned to the Essence STX's analog input, this operated at the same sample rates as the playback settings. The input level applied to the analog/digital converter can be adjusted with the Audio Center program's Record Gain control. With this control at its maximum, a signal level of 2.1V RMS resulted in clipping-free digital data with a level of 0dBFS; ie, the maximum possible. The input impedance was a fairly low 4300 ohms at all frequencies, which will stress some tubed preamps, and the input preserved absolute polarity; that is, positive-going pulses resulted in positive data words. The frequency response of the data depended on the chosen sample rate, of course. The low end was flat to 20Hz; the high end followed the same ultrasonic rolloff seen in the playback responses (fig.11).

Fig.11 ASUS Xonar Essence STX, line input, A/D frequency response at –12dBFS with sample rate set to192kHz (left channel solid, right dashed, 0.5dB/vertical div.)

Inexpensive A/D converters tend to start overloading as the input level reaches the maximum possible. The STX's ADC, however, was well behaved even at the very top of its dynamic-range window. Fig.12, for example, shows the spectrum (analyzed in the digital domain) of a 1kHz tone at a level equivalent to –1dBFS, with the word length set to 24 bits and recorded with Adobe Audition. The highest-level harmonics are the third, at –116dB (0.00015%), and the second, at –126dB (0.00006%). However, these are close to the residual level of the signal generator I used for these tests, Audio Precision's System One. Dropping the input level to –10dBFS (fig.13) gave a spectrum almost free from distortion, though some enharmonic spuriae are visible at a very low level (enharmonic means that their frequencies are unrelated to the signal frequency). But even with a signal level equivalent to –90dBFS, these enharmonic tones are still well below the signal level (fig.14).

Fig.12 ASUS Xonar Essence STX, line input, spectrum of 1kHz sinewave at –1dBFS, 24-bit data, 44.1kHz sample rate (left channel blue, right red; linear frequency scale).

Fig.13 ASUS Xonar Essence STX, line input, spectrum of 1kHz sinewave at –10dBFS, 24-bit data, 44.1kHz sample rate (left channel blue, right red; linear frequency scale).

Fig.14 ASUS Xonar Essence STX, line input, spectrum of 1kHz sinewave at –90dBFS, 24-bit data, 44.1kHz sample rate (left channel blue, right red; linear frequency scale).

Figs.12–14 were taken with a sample rate of 44.1kHz. Changing the sample rate to 48 and 96kHz gave very similar results (not shown). At 192kHz, however, you start to see the effect of the noiseshaping used in the A/D converter to get the necessary baseband resolution as the gentle rise in the noise floor above 50kHz (fig.15).

Fig.15 ASUS Xonar Essence STX, line input, spectrum of 44kHz sinewave at –12dBFS, 24-bit data, 192kHz sample rate (left channel blue, right red; linear frequency scale).

The ST (PCI) version of the Essence didn't measure quite as well as the STX (PCI Express) version, but both still offer very impressive sets of measurements for a product that is a) affordably priced and b) must be housed in the hostile RF environment of a personal computer.—John Atkinson