Digital Processor Reviews

Note the convergence of the traces below 200Hz, which can also been seen in the spectra of the Musical Fidelity's output while it decodes "digital black" data (fig.8). I suspect that this is due to the unit's low-frequency noise performance being limited by the highish impedance of the wall-wart power supply. A bigger, stiffer transformer might work some magic in the bass.

Fig.8 Musical Fidelity X-24K, spectrum of digital silence with 16-bit data (top) and 24-bit data (bottom). (1/3-octave analysis, right channel dashed.)

As suspected from these noise-floor graphs, the X-24K's linearity (fig..9) is superb. The error is less than 2dB to below -115dBFS. The reproduction of an undithered 16-bit, 1kHz tone at -90.31dBFS (fig.10) is correspondingly excellent. Increasing the word length to 24 bits gives a reasonable facsimile of a sinewave (fig.11) overlaid with noise.

Fig.9 Musical Fidelity X-24K, departure from linearity (right channel dashed, 2dB/vertical div.).

Fig.10 Musical Fidelity X-24K, waveform of undithered 1kHz sinewave at -90.31dBFS (16-bit data).

Fig.11 Musical Fidelity X-24K, waveform of undithered 1kHz sinewave at -90.31dBFS (24-bit data).

It was hard to find any harmonic distortion lurking in the Musical Fidelity's output. Fig.12 shows the spectrum of the processor's output while it drove a 50Hz tone into 100k ohms. The only harmonics visible are the second and fourth—and these are at -100dB or better! Similarly, the X-24K passed the high-frequency intermodulation test (fig.13, taken with a 96kHz sample rate) with flying colors. Even the 1kHz difference component was 100dB down from peak level!

Fig.12 Musical Fidelity X-24K, DC-1kHz spectrum, 50Hz at 0dBFS, 100k ohms load (linear frequency scale, 20dB/vertical div.).

Fig.13 Musical Fidelity X-24K, HF intermodulation spectrum, DC-24kHz, 19+20kHz at 0dBFS, 96kHz sampling, 100k ohms load (linear frequency scale, 20dB/vertical div.).

Only when it came to jitter rejection did the Musical Fidelity stumble, and then only slightly. Testing its word-clock jitter with the Miller Audio Research analyzer gave the spectrum shown in fig.14. The grayed-out spectrum is that of the Meridian 508.24 one-box player, shown for reference—the Musical Fidelity is about 1dB noisier overall, which is still superb. But its much higher levels of data-related jitter (indicated with red numbers) lead to a peak jitter level at 240 picoseconds with a TosLink connection, almost twice that of the Meridian. This is good performance, however, and replacing the TosLink with a true 75 ohm electrical connection reduced the jitter to 231ps. Note the brown markers: these are jitter sidebands related to the power-line frequency of 60Hz. Substituting a better AC supply for the wall wart might reduce the level of these.

Fig.14 Musical Fidelity X-24K, high-resolution jitter spectrum of analog output signal (11.025kHz at -6dBFS with LSB toggled at 229Hz). Center frequency of trace, 11kHz; frequency range, ±3.5kHz. (Grayed-out spectrum is that of the Meridian 508.24).

All things considered, this is astonishing measured performance. As with the little Entechs reviewed by Brian Damkroger in the December '98 issue, even five years ago you would have had to pay very much more than the X-24K's $450 to get this level of digital performance. And the Musical Fidelity will run at 96kHz!—John Atkinson