CD Player/Transport Reviews

I measured the MBL N31 with my Audio Precision SYS2722 system (see the January 2008 As We See It"), using both the Audio Precision's optical and electrical digital outputs and USB data sourced from my MacBook Pro running on battery power with Pure Music 3.0 playing WAV and AIFF test-tone files. Apple's USB Prober utility identified the N31 as "MBL USB Audio Class 2" from "MBL Akustikgerate, Berlin," and its serial number as "Streamlength(tm)." The MBL's USB port operated in the optimal isochronous asynchronous mode. Apple's AudioMIDI utility revealed that, via USB, the N31 accepted 24-bit integer data. The optical input accepted datastreams with sample rates up to 96kHz, and the USB 2.0, AES/EBU, and S/PDIF inputs accepted streams of up to 192kHz.

The CD drive's error correction was good rather than great, glitches appearing in the player's output when the gaps in the data spiral on the Pierre Verany Digital Test CD reached 1.25mm in length at standard track pitch or 1mm with minimum track pitch. This was still much better than the CD standard, the so-called "Red Book," which requires only that a player cope with gaps of up to 0.2mm. The maximum output level at 1kHz was 4.15V from the balanced outputs, 2.04V from the unbalanced, and both outputs preserved absolute polarity. (The XLR jacks are wired with pin 2 hot.) The output impedance conformed to the specification at 198 ohms balanced and 93 ohms unbalanced.

The Fast filter's impulse response (fig.1) indicated that it is a conventional linear-phase finite impulse response (FIR) type, with symmetrical ringing either side of the single sample at 0dBFS. The Slow filter (fig.2) is also an FIR type but is much shorter, while the Min filter (fig.3) is also short but is a minimum-phase type, with all the ringing following the single high sample. The Fast filter rolled off rapidly above half the sampling frequency when the N31 decoded white noise sampled at 44.1kHz (fig.4, footnote 1), and reached the full stop-band suppression at 25kHz. By contrast, the Slow filter did indeed roll off slowly above the audioband (fig.5, magenta and red traces). As a result, there was very little suppression of the aliased image of a full-scale 19.1kHz tone (cyan and blue traces). The Min filter (fig.6) rolled off a little faster than the Slow, but the ultrasonic noise floor has an odd-looking, scalloped appearance.

With the Min filter and 44.1kHz data, the response was flat up to 18kHz but was then –3dB at 20kHz (fig.7, gray and green traces). With the Slow filter (fig.8), the rolloff began lower in the audioband, above 12kHz. At higher sample rates with all three filters, the rolloff continued smoothly until half of each sample rate, at which point the response dropped rapidly. Channel separation was >125dB in both directions below 1kHz, decreasing slightly to a still-superb 113dB at 20kHz. The N31's analog noise floor was both free from power-supply–related spuriae and extraordinarily low in level. When I increased the bit depth from 16 to 24 with a dithered 1kHz tone at –90dBFS (fig.9), the noise floor dropped by almost 30dB, meaning that the MBL offers 21 bits' worth of resolution—the current state of the DAC art. With undithered data representing a tone at exactly –90.31dBFS (fig.10), the three DC voltage levels described by the data were well resolved, as was the Min filter's minimum-phase ringing at the level transitions. With undithered 24-bit data, the result was a clean sinewave (fig.11).

Not only did the N31 offer very low levels of analog noise, its distortion was also superbly low. With a full-scale 50Hz tone, the distortion harmonics all lay below –120B (0.0001%), even into the punishing 600 ohm load (fig.12). Intermodulation distortion was also vanishingly low in level (fig.13), though with the Slow and Min filters, the aliased images of the 19 and 20kHz tones with which I test IMD were not well suppressed (fig.14). The MBL's rejection of word-clock jitter, with both CD (fig.15) and 24-bit AES/EBU data (fig.16), was superb. Digital audio engineering doesn't get any better.

Finally, as Jürgen Reis emphasizes the N31's ability to cope gracefully with intersample overs, I looked at the CD track of Steely Dan's "Gaslighting Abbie" using the no-longer-available BIAS Peak program, which doesn't interpolate between samples for its waveform display. There were indeed multiple instances of consecutive samples at 0dBFS that would result in intersample overs. The red arrows in fig.17, for example, show two such instances in the non-interpolated waveform at 3:27, the one in the left channel (top) having three consecutive samples at 0dBFS, and that in the right (bottom) two consecutive samples.—John Atkinson