Hi-Rez Disc Player/Transport Reviews

I measured the Luxman D-06u with my Audio Precision SYS2722 system (see the January 2008 "As We See It"). As well as playing test CDs and the "provisional" test SACD produced by Sony, I examined the behavior of the digital inputs with either S/PDIF signals fed from the SYS2722, or WAV and AIFF test-tone files played with Pure Music 3.0 by my MacBook Pro running on battery power and fed via USB. Apple's USB Prober utility identified the processor as the "Super Audio CD Player D-06u" from "Luxman," and confirmed that the D-06u's USB input operated in the optimal isochronous asynchronous mode. The coaxial S/PDIF input accepted data sampled at up to 192kHz, the TosLink input up to 96kHz. Apple's AudioMIDI app indicated that the Luxman operated at all PCM sample rates from 32 to 384kHz via USB, with bit depths of up to 32 integer. The D-06u successfully decoded DSD64 and DSD128 data fed it via USB from the MacBook.

Tested with the Pierre Verany Digital Test CDs, the D-06u played without glitches until track 35, which has 2.4mm gaps in the data spiral. This is excellent error correction/concealment. The player's maximum output level at 1kHz with PCM data was 2.36V from both the balanced and single-ended outputs, and slightly higher with SACD, at 2.49V. With the front-panel Phase LED illuminated, while the balanced inputs preserved absolute polarity, the single-ended outputs inverted polarity. This suggests that the XLRs are wired with pin 3 hot, the opposite of the modern convention. The unbalanced output impedance was a relatively low 297 ohms; the balanced value was twice that amount, as expected.

The three reconstruction filters for PCM data—P-1 is labeled Normal, P-2 Low Latency, and P-3 High Attenuation—appear to be identical to those used in Luxman's DA06 D/A processor, which Art Dudley reviewed in July 2014. The Normal filter, which AD preferred, is described by Luxman as a "normal FIR [Finite Impulse Response] filter," but its impulse response with 44.1kHz data reveals it be a time-asymmetrical type, with almost all the ringing following the single sample at 0dBFS (fig.1). P-2's impulse response was similar, while that of P-3 (fig.2) was typical of a time-symmetrical FIR filter.

With 44.1kHz-sampled white noise (footnote 1), the D-06u's ultrasonic output rolled off rapidly, but with a complex series of nulls apparent between 24 and 100kHz (fig.3, red and magenta traces). P-2 was a little better behaved (fig.4), and P-3 offered a conventional, very sharp rolloff above the audioband (fig.5). The blue and cyan traces in fig.5 show the spectrum with a full-scale 19.1kHz tone; note the very low levels of the distortion harmonics produced by this tone. Fig.6 was taken with P-1: the outputs at 44.1, 96, and 192kHz follow the same basic shape, with a sharp rolloff just below each sample rate's Nyquist frequency (half the sampling frequency). P-2 offered a less steep rolloff, however.

The D-06u's channel separation was superb, at >120dB in both directions below 3kHz, and still 113dB at 20kHz. The player's noise floor was free from any power-supply–related artifacts (fig.7), and increasing the bit depth from 16 to 24 with a dithered 1kHz tone at –90dBFS dropped the noise floor by almost 30dB (fig.8), suggesting resolution of better than 20 bits, which is superb. This graph was taken with S/PDIF data; repeating the test with 24-bit USB data raised the noise floor by 10dB (fig.9), reducing the resolution by 1.5 bits' worth, which surprised me. There should be no difference. But with its low level of noise and high intrinsic resolution, the D-06u's reproduction of an undithered 16-bit tone at exactly –90.31dBFS was essentially perfect (fig.10). This graph was taken with P-2, the Low Latency filter, and the asymmetric impulse response is readily apparent. With undithered 24-bit data at the same level, the result was a perfect sinewave with all three filters (fig.11).

As mentioned earlier, the D-06u offers very low levels of harmonic distortion, even into the punishing 600-ohm load (fig.12), where the second and third harmonics both lie just above –110dB (0.0003%). Tested with SACD data, the third harmonic was slightly higher in level (fig.13), and the rise in the ultrasonic noise floor due to the DSD encoding can be seen. (This graph was taken with the steeper DSD low-pass filter; the spectrum looked identical with the gentler DSD filter.)

When I tested for intermodulation distortion with an equal mix of 19 and 20kHz tones, I got anomalous results. While the actual intermodulation products were extremely low in level, the noise floor was 30dB higher than I was expecting from the test used to produce fig.8. I was using S/PDIF data fed to the coaxial input, so I repeated the test with a Toslink connection. This produced a graph with similarly low levels of intermodulation distortion but a much lower noisefloor (fig.14). All I can think is that the D-06u's electrical S/PDIF input has some issues that lead to the rise in the level of the high-frequency noise that I noted. Fig.13 was taken with the P-1 filter; P-3 (fig.15) gave greater suppression of the aliasing products, as expected.

When I tested the D-06u's rejection of word-clock jitter, I again got anomalous results. Fig.16, for example, shows the spectrum with 16-bit CD data: Although the green line shows what should be the correct levels of the odd-order harmonics of the low-frequency, LSB-level squarewave, the harmonics are obscured by many higher-level noise spikes. Repeating the test with 16-bit TosLink J-Test data gave a much cleaner-looking spectrum (fig.17), though the harmonics closest to the 11.025kHz tone are variously accentuated or attenuated a little. Paradoxically, S/PDIF data fed to the D-06u via the coaxial input, which in theory should result in identical behavior, gave a worse result (fig.18). And when I repeated the test with 16-bit USB data (fig.19), the result was a scalloping of the noise floor that looks similar to that seen surrounding the base of the spectral spike that represents the 19.1kHz tone in fig.3 (blue and cyan traces); and what is probably an idle tone can be seen just below 10kHz. This behavior was unchanged with 24-bit data (fig.20).

In many ways, the Luxman D-06u offers excellent measured performance. I have no idea why it measures so much worse than Luxman's DA06 processor in terms of jitter rejection and the rise in its high-frequency noise floor with the coaxial S/PDIF input. A mystery.—John Atkinson