MBL Noble Line N31 CD player-DAC

MBL's Corona C15 monoblock has been one of my amplification references since I reviewed it in 2014, and as I've been reviewing DACs the past year or so, it was high time I spent time with one of the German company's digital products.

"Black shiny products are tough to photograph at shows, so trust me when I say the new N31 is dripping with gorgeousness not reflected in this photo," wrote Jon Iverson in his report from the 2017 Consumer Electronics Show. And when I unpacked my review sample of the Noble Line N31 ($15,400), I was indeed taken with its looks. The N31's front panel is dominated by a large, 5", color TFT display, flanked on each side by three function buttons that MBL calls "soft keys," and below it the slot for the vibration-damped CD transport. (The N31 recognizes CD text and displays the title information.) On the top panel is a circular, indented touchpad that controls the brightness of the display in six steps and glows a soft white.

The N31's rear features AES/EBU, optical and electrical S/PDIF, and USB 1 and USB 2 digital inputs; AES/EBU, and optical and electrical S/PDIF digital outputs; and balanced and unbalanced analog outputs. There is also a slot for an SDcard, for firmware updates. The digital inputs and outputs are all galvanically isolated, and oscillators not required for a selected input are automatically shut down.

I interviewed MBL's chief designer, Jürgen Reis, at CES 2017 (footnote 1). When I asked him what was special about the N31, he told me that they'd concentrated on minimizing jitter. The USB inputs are powered from a separate transformer winding, and are optically isolated from the digital circuitry and the analog audio circuitry. There is a three-stage jitter-reduction stage: first, a digital PLL with a wide, 10kHz acceptance window to allow the N31 to lock to a source; then an analog smoothing PLL with a 1Hz corner frequency to reject higher-frequency jitter; and, finally, an asynchronous readout buffer. The output of this last stage feeds the N31's oversampling filter, which deals gracefully with "intersample overs."


Intersample overs occur when the digital data have consecutive samples at 0dBFS. When the digital reconstruction filter interpolates the analog waveform, the result is a waveform that could peak up to 3dB above 0dBFS. If the DSP engineer responsible for the digital filter has not allowed any mathematical headroom, the waveform will be clipped. Stereophile used to routinely examine digital-filter headroom, as you can see in our reviews originally published in the early 1990s. But that was before the Loudness Wars, when CDs were mastered so that there were never consecutive samples at 0dBFS. (Some digital-audio workstations calculate the waveform on the assumption that it would be processed by a typical digital filter, thus allowing the mastering engineer to avoid intersample overs.) Such overs are rare to nonexistent in classical recordings, I've found, but both Reis and Benchmark's John Siau have analyzed a lot of rock recordings and found many pathological examples. For example, in "Gaslighting Abbie," from Steely Dan's Two Against Nature, Siau found 1129 intersample overs: 3.7 per second (see "Measurements" sidebar).

Reis explained that the N31's digital filter has 3.5dB of headroom, so that when there are intersample overs, the DSP will never clip. As supplied, the DAC chip used in the N31, a balanced ESS Sabre 9018, doesn't have intersample overload protection and offers generic filters. However, the basic specifications are very good, and the chip's topology allows Reis to manipulate its behavior to eliminate the things he doesn't like about it. For example, while ESS's on-chip filter has 128 taps, Reis modified it to have just 32 taps, to obtain a subjectively desirable "short" filter. MBL also uses its own technology to differently weight the four DAC paths per channel, to smooth the transition between the delta-sigma and multi-bit sections that occurs, I understand, at around –36dBFS. The DAC is followed by an analog low-pass filter and a one-stage output buffer, resulting in a very wide dynamic range (again see "Measurements" sidebar).

The N31's remote handset offers CD playback controls, as well as the basic functions for MBL's N51 integrated amplifier. It's a substantial, circular beast, finished in black with chrome trim, and seems to have some degree of self-awareness—when you reach for the remote, even before you've touched it, its buttons' labels light up. The N31, too, senses the user's approach, the Menu and Device (input) icons on its display appearing before any of the buttons to left or right are touched. As I'm used to touchscreens, it took me a while to learn not to push the onscreen icons and instead press the associated buttons next to them.

As the N31 has no volume control, I performed my initial auditioning using an Ayre Acoustics KX-5 Twenty preamplifier. But I had shipped the Ayre to Art Dudley, for him to review for the March issue, I used a balanced NHT Passive Volume Control I'd bought a few years ago, using short, balanced Canare cables, with a ¼" TRS jack on the sending end and a male XLR on the other, to connect the NHT PVC to the amplifiers.


It's common to describe a product's greater resolution of recorded detail as always being an improvement. However, it's also common for this greater detail to be accompanied by too up-front a sound—"ruthlessly revealing," as the audiophile cliché has it. The MBL Noble Line N31 avoided this syndrome, opening a clean window on the recording's soundstage but without thrusting things forward at me. I chose for one of my "Records to Die For" in this issue Robert Silverman's performances of the two Rachmaninoff Piano Sonatas from 1980 and 1991, which we reissued in 2007 (CD, Stereophile STPH019-2). I hadn't played this disc for some years, but packing for an office move last November, I found a CD-R I'd burned from the DDP file set I'd prepared for the pressing plant. Playing it in the N31 reminded me why I hadn't wanted these recordings to disappear. Silverman's Hamburg Steinway in Sonata 1 was hanging there between and behind the GoldenEar Technology Triton Reference speakers, "palpably present" (as that other audiophile cliché puts it). Yes, I could hear an occasional touch of flutter, but I could ignore this in favor of the sheer tangibility of the N31's presentation of the sound, and the forceful nature of the piano's lowest register.

This was with the N31's Min filter. In my interview with him, Jürgen Reis had told me that he feels the minimum-phase Min to sound the most musically natural of the three reconstruction filters. With the MQA-encoded versions of Bob Silverman's recording of Beethoven's 32 piano sonatas (24-bit/88.2kHz FLAC files) sourced from the NAD M50.2 music server, which performed the first "unfolding" (see Jim Austin's article on MQA elsewhere in this issue), there was a slight ringing in the upper midrange with the Min filter that I hadn't heard with the Rachmaninoff CD. This was reduced with the Fast filter, but so was some of the palpability of the piano sound. With this Beethoven recording, I ended up using the Slow filter, though when I played the Portland State Chamber Choir's performance of Purcell's "Hear My Prayer," from their Into Unknown Worlds (16/44.1 ALAC file from CD, CDBaby), this filter was a touch too aggressive.

Footnote 1: Click here. Michael Lavorgna and I also talked to Jürgen Reis about more general aspects of digital audio here.
MBL Akustikgeräte GmbH & Co.
US distributor: MBL North America, Inc.
217 N. Seacrest Boulevard #276
Boynton Beach, FL 33425
(561) 735-9300

Axiom05's picture

Please correct me if I'm wrong, but isn't this whole issue a consequence of the mastering being done poorly? If the people doing the mastering didn't "push" the recorded levels so high, right up against 0 dB, then there wouldn't be any intersample overs? Isn't this another issue caused by the poor quality of the recordings that we are being offered? A higher quality of engineering would eliminate so many of the problems that we are facing in recorded music. These are not "digital" defects, these are symbolic of poor quality work. IMHO, of course...

CG's picture


But, this is being done deliberately by somebody in the recording chain. I can only speculate that this is what they feel they need to do in order for the recording to play well over the devices of the time. So, it's not "poor work", although we might think so.

It's funny... People wonder why young folks today don't show as much interest in "hifi" as kids did back in the 70's or thereabouts. Maybe part of the reason is that the contemporary music they like doesn't reproduce well over what we might call "hifi" systems.

supamark's picture

and mastering engineers (and consumers). Labels want it to sound "good" on earbuds and crappy bluetooth speakers because that's how most pop music is consumed so it's heavily compressed/limited, and mastering engineers do it because that's how you keep the lights on and the rent paid.

The upside, if you like vinyl, is that records are mastered with far less compression of the dynamic range (you simply can't cut a record as hot as modern CD's and expect it to be playable). Those young hipsters buying vinyl are the next crop of audiophiles. Unfortunately, a good vinyl playback chain is significantly more expensive (and fiddly) than a comparable DAC/transport.

supamark's picture

that white/gold finish looks TACKY in pictures. The black/silver is a bit cold, but at least it doesn't look like it has a comb-over and tiny fingers [rimshot].