The m9XX’s AKM4490 DAC is up to the job. It handles every sampling and bit rate you can shake at. It handles hi-resolution files with ease, and measurably keeps up with the rigours of 24-bit audio, including DSD.
Irrespective the adapter or battery pack, the m9XX’s noise signature is stable and non-granular. It is broken neither by active peaks nor swimmy undulations. Volume-wise, however, that noise is roughly on par with the Lynx Hilo, or a bit higher than an Astell&Kern AK100. Sensitive earphones will pull low levels of background noise from the m9XX. Even when music is playing, I can hear it, but just so. Again, I listen to music at mature levels. And, as Headfi’s Hiss King, I’m a special case. Most people will either not be bothered by it, or not notice it. And, to be honest, apart from it being there (in this case there is a deprecatory term) it doesn’t put me off. Which means that the m9XX less suited to Shure’s SE846 or Ultrasone’s IQ than is Mojo.
In all other metrics, the m9XX is brilliant. Channel balance from volume settings of 0 to 99., is impeccable and initial gain is low, and perfectly suitable to accurately control the most sensitive earphones with loads of headroom. For instance, I listen to the IQ at volumes of 12 to 20; at least 24 steps exist between zero output to what I consider comfortable. The 600Ω T1 takes the scale up to between 52 and 65. The less sensitive 600Ω DT880 takes it up to a maximum of which tops out around 80. Those are my numbers. Your mileage will vary.
What won’t is the m9XX’s power. If you can’t get enough power for your headphones from the m9XX you’re either deaf, or you’ve already broken your headphones. It provides overkill voltage for the DT880/600, the T1, and the HD880 at current-stable levels. You’ll bleed long before high-performance mode runs into IMD.
Even under load, stereo crosstalk never dips below ~88. For reference, Mojo falls to -82,1 when pushing the Audio Technical ES7. In its defence, Mojo’s stock crosstalk of -115,4dB is roughly 10dB better than the m9XX’s best effort ()detected by the Lynx HILO @+6dBv). Still, the gulf between loaded and unloaded tests suggests that the m9XX (load effect of ~15%) carries its signal better than Mojo (load effect of ~28%). Impressive. The results are similar across all metrics, but again, the base unloaded scores that Mojo kicks out sometimes quite a way ahead of the m9XX.
At the same voltage levels and through my setup (through which I’ve tested gear that tops out at +120dB), m9XX’s dynamic range tops out 114,1dB. Mojo squeaks by it at 116dB. Not that you’ll hear the difference at normal listening levels.
When loaded with an Earsonics SM2, the m9XX is stabler than Mojo by a small factor, outputting roughly 42% of Mojo’s IMD and about 26% of its THD. It’s a foregone conclusion that the m9XX is stabler than the single-ended output of any high-end modern DAP.
Maxing voltages between the two reveals about 2dB of extra volume from Mojo, but at the expense of massive upticks in IMD. Reduced by two or three clicks to the m9XX’s maximum output, Mojo stabilises. The m9XX’s top volume is rock solid and more usable loaded, and marginally so unloaded.
It is important to remember that while m9XX’s amp section may be marginally stabler than Mojo’s, Mojo’s unloaded performance is better in almost every metric; sometimes (as seen in stereo crosstalk), by a large amount. If all you’re after is a perfect signal to an outboard amp, it gets tricky. Mojo measures better. But it’s way easier to hook up RCA outs to RCA inputs; and m9XX is a way better set up for concurrent use of both headphones and line outs. If you’ve got the space, and the right power adapter, the Grace is a more elegant solution.
If you’re primarily an on-the-go listener, Mojo’s strengths are impossible to overcome. It outperforms almost every high-end portable amp in almost every metric. Its output is nearly as silent as to boot (certainly it is more silent than the m9XX). Because it’s got a battery inside, there is no need to fuss with cables. Finally, it is much smaller. On the road, there’s not a moment in which I would choose the m9XX. At home, or firmly grounded for the long term at a desk, I will reach for the m9XX every time.
The m9XX is slightly smoother and darker, while Mojo is more detail-oriented. The differences are slight, but are most noticeable in the high frequencies. The Grace is detailed and kicks out a nice, 3D stage that pushes both wide and tall, with decent y-axis space. But, Mojo has more detail and space on the y-axis. It also presents a more neutral emphasis on stereo imaging, where all frequencies receive roughly the same sound pressure (unloaded). Loaded, m9XX’s greater stability equalises much of that (small) disparity.
I prefer listening the flighty Ultrasone IQ through the m9XX. Conversely, I prefer the warmer Beyerdynamic T1 through Mojo, whose more detail-oriented signature helps cool things down.
And then there are the filters.
Grace’s crossfeed filter is brilliant, flattening stereo separation to around -22dB. No, it won’t comfort up your headphones into a two-channel setup, but it is nice for a number of older recordings. Unless listening for unbroken hours at a time, I feel no need for crossfeed for modern recordings. Your mileage will vary. Better, the crossfeed is so well implemented negatively impacts almost nothing else. THD bumps up just barely, but everything else stays roughly the same. Basically, the m9XX nets the benefits of crossfeed with little to no performance penalty.
The DAC filters are basic roll offs, with associated analogue-like minimum phase effects, or digital-domain linear phase equalisation. While I’m not qualified to talk about the accuracy of either method, I’m qualified to say that I’m impressed. I’ve not got a favourite, but F2 and F4 are often on when my DT880/600 is on. Page 12 of the m9XX manual gives a good run-down of when, why, and where to use each. F2 and F4 filters begin slow roll-offs from about 10K. Both are lovely when running sometimes-harsh sounding headphones. And while it is just my subjective option, I feel that they are less suited to warmer, mid-heavy headphones such as the T1 and HD650.
Naturally, your mileage will vary. What won’t is the thought that went into each filter, that went into each DAC option, that went into the output circuit, that went into the selection of the m9XX’s DAC. Brilliant.
I wish there were space for another alphanumeric. And, m9XX’s feet are crap. Nevertheless, the m9XX is brilliant. As a desktop DAC, it is phenomenal. I think I should add this: it would be nice if the power, performance, and beautiful DAC filters sat behind a nice balanced headphone output. There’s little about which to complain.
If you don’t need analogue input, and you’re primarily listen whilst sitting at a desk, the m9XX is about as good as it gets – if you’re not super overly sensitive to hiss. Measurably it is better than JDS Labs’s The Element, but also costs 200$ more. That extra may go to raise the performance bar. It certainly has gone into making the m9XX one of the most elegant desktop headphone amps around.
And if the numbers mean anything, loaded, it even beats out Mojo. Of course, Mojo hisses quite a bit less, and boasts better unloaded performance-by-numbers. The twain represent the best in their respective classes: m9XX for the desk, Mojo for the pocket or purse.
Looks-wise, the m9XX is a Leica among Nikons and Fujifilms. And like Leica, it focuses on the essential. It is a deliberate, elegant machine.