John
Scary Monsters. And super creeps. David Bowie. And Robert Fripp. Listen intently to the album on a decent hi-fi system and you’ll not miss the King Crimson founder’s “Applied Frippertronics”: guitar sounds routed through a twin-tape-machine delay that tears at the very fabric of songs like “Fashion”, “It’s No Game Pt. 1“ and “Teenage Wildlife”. Fripp optimises Bowie.
I’ve been reminding myself of Scary Monsters’ high tensile combination of art and pop with a Kii Three loudspeaker system. Supplying the digital audio stream is an AURALiC Aries Mini (US$599, discontinued) that, when connected to the Kii Remote’s USB input, sees the ones and zeroes flow into each loudspeaker’s D/A converters over Ethernet wire. The sound of Fripp and Bowie through this hifi system is nothing short of arresting. Nervy. Wiry.
And yet inserting an Innuos Phoenix US re-clocker (US$3149) between the AURALiC and the Kii Remote tells us that Scary Monsters doesn’t have to sound quite so nervous and tense; that a better-defined illusion of front-to-back space between Bowie’s vocal and Fripp’s guitar can exist. The Innuos unit infuses the Bowie album with greater soundstage depth. The changes don’t stop there. Microdynamics lose some stiffness to sound more elastic in their snapback. This helps tenderise, just so, Fripp’s guitar sear.
Why does this matter to Music-First Audiophiles – those of us who get our kicks from music and its attendant sound but who don’t always dig gear for its own sake? The answer is surprisingly simple. With the Phoenix added to the Aries Mini, we can push SPLs higher on Scary Monsters without increasing the risk of a Frippertronics-induced headache — a delta that mirrors the qualitative differences between Scary Monsters’ harder-sounding 1992 Ryko remaster and Parlophone’s 2017 superior redo.
How is this possible?
A streaming-capable digital audio system is a lot like home-made espresso. A cursory glance at a handful of YouTube videos makes it look easy. To produce an espresso, a machine forces water through ground coffee to extract the flavour previously locked inside the bean. Theoretically, yes, but reality has other ideas. Experienced baristas will tell you that grind quality has a much greater impact on an espresso’s flavour and texture than the espresso machine.
Similarly, a DAC – the machine that converts digital to analogue – isn’t immune to upstream changes. Input quality influences output quality. Our ones and zeroes are like coffee beans in that they require more than cursory handling. Digital audio experts talk of how a USB audio connection’s jitter (data timing) and electrical (not acoustic) noise carriage can impact sound quality.
Just as a lesser-quality grinder’s inconsistent particle size can reduce the likelihood of a proper extraction in the espresso machine, a consumer-grade PC’s higher levels of jitter and electrical noise will reduce the audible capabilities of the DAC to which it is connected via USB. I’ve done this experiment too many times to count: USB tethered to a DAC, a Macbook sounds micro-dynamically more rigid and more tonally stonewashed than most dedicated network streamers. The latter is a new pair of jeans; the former, those same jeans after several washes when some of the dye has been washed away.
This doesn’t prevent one of the most common misconceptions about digital audio transmission from living it large in the mainstream: that USB audio connections come with in-built error correction. They don’t. Print jobs and hard-drives use Bulk Mode’s CRC correction. Digital audio’s time-sensitive data transmission uses Isochronous mode where CRC correction doesn’t get a run. We’ve already seen from DAC designers like Wavelength Audio’s Gordon Rankin, inventor of asynchronous USB DAC connections, that 1) a USB audio connection can drop packets and 2) that those packet drops aren’t trivial to formally measure.
That USB audio quality goes well beyond data transmission can be a tough adjustment for anyone fixating on the ones and zeros with a “bits are bits” mindset. For example, the second phase of Rankin’s DragonFly DAC development for AudioQuest focussed not only on the audio data travelling from laptop (or smartphone) to dongle DAC but the electrical noise profile of the DragonFly’s own USB receiver chip, specifically its power supplies.
The lower the noise, the better quality the USB signal. Electrical noise can disturb the accuracy of the clock that times the release of audio samples into the DAC chip at multiples of 44100kHz or 48000kHz. Mis-timed sample releases distort the analogue waveform drawn by the DAC chip.
Audio engineers like AudioQuest’s Garth Powell and Innuos CEO Nuno Vitorino have told us that this same electrical noise can additionally negatively influence the DAC’s analogue output stage — the section of circuitry that precedes the unit’s RCA or XLR connectors. Powell has published a white paper on electrical noise here.
Like AudioQuest, Innuos best substantiate their claims with A/B show demos. At Munich High-End 2017, I witnessed first hand the gasps of surprise as a Kii Remote – converting USB to Ethernet for a Kii Three/BXT system – translated the audible differences between Innuos’ ZENith MKII SE server/streamer and their (then new) two-piece Statement server/streamer.
Late last year, it happened again. At Berlin dealership Max Schlundt Kultur Technik, the audible delta between a ZENith MKIII and a Statement, again routed through a Kii Three/BXT system, was not difficult to hear. Also on the floor that day was a freshly minted Phoenix that extracted an even greater sense of ease from the already easeful ZENith MKIII — a result that I would later replicate at home with Bowie’s Scary Monsters. Interestingly, the Phoenix pushed the ZENith MKIII’s sound signature further from the Statement’s which, sans Phoenix, offered more sparkle and low-level jump factor.
For their USB reclocker, Innuos has relocated the Statement’s USB circuitry to its own enclosure. It can be applied to any USB audio source: USB in, USB out. In between, a signal tidy up takes place. Not by filtering the host streamer’s power supply but by slipstreaming in its own 5V feed.
Talking over the phone in the midst of 2020’s Corona lockdown, Vitorino first points out that this is not a third-party USB board. It is built at Innuos’ Algarve facility. The Phoenix’s other key ingredient is a pair of linear power supplies, both designed by one Dr. Sean Jacobs.
According to Vitorino, USB hubs require three separate voltages and ‘Amazon-type’ hubs often split a single incoming voltage into three using electrically noisy switching regulators. The Phoenix’s USB board gets its three voltages directly from the first of those two linear supplies.
The second linear power supply is reserved for the clock that times the release of data into the DAC’s USB receiver chip. Note: this is NOT the word clock that times the release of samples into the DAC’s decoder chip at multiples of 44.1kHz and 48kHz. This is the USB bus’ data clock that times ones and zeros into the DAC’s USB receiver chip at 480 Mbps (24 MHz x 20 bits).
Vitorino tells it like this: driven by a vanishingly low bill of materials, less-costly USB hubs feature a clock that gets the job done and nothing more; more advanced offerings will feature a superior ‘oven-controlled crystal oscillator’ (OCXO) but those often call for additional circuitry to convert the clock’s native 10MHz signal to the required 24MHz.
Into the Phoenix, Innuos has dropped an OCXO, housed in a metal enclosure, to 1) regulate the USB chip’s I/O from an inch away and 2) natively at 24MHz to require no additional conversion.
Vitorino sums up the Phoenix as having all the necessary ingredients for a very clean, very well-timed USB signal that’s low on the noise that would otherwise disturb the rest of the DAC’s circuitry.
We’ve seen USB pipe cleaners before – from the likes of Wyred 4 Sound, Schiit and UpTone audio – but never one as large, as heavy or as costly as the Innuos Phoenix.
Using the Kii Three loudspeaker system for consistency and talking scores out of ten to advance clarity, Wyred 4 Sound’s Recovery (US$149) lifts the Aries Mini from a 4 to a 5; it makes Kate Bush’s meisterwerk, The Hounds of Love, sound slightly less tense and better inks tonal colours. Each of this review’s musical mentions are cherry-picked from seven weeks of listening, swapping out gear and listening again; a few days with configuration X followed by another few days with configuration Y.
The Phoenix takes the Aries Mini in the same direction as the Recovery but further; to a 7. Tonal colours display less chalk and more crayon. Dynamics sound less restive, more reflexive. But does appending a $500 network streamer with a three grand USB reclocker make financial sense? What follows may sharpen your thinking.
Both the Recovery and the Phoenix put a stop to the Raspberry Pi 3’s (€49) intermittent audible pops and clicks, likely caused by USB buffer underrun. On Pis v1 – 3, network and USB share a single data bus. Technical glitches dispensed with, what about sound quality?
Playing The The’s cover of Hank Williams’ “Your Cheating Heart” (Hanky Panky), the Recovery elevates the Pi 3 from 1 (out of ten) to 4. Dramatic! But wait: the Phoenix adds a smidge more solidity to lower frequencies and better tempers the harmonica’s sibilant threat for a score of 6 — one point shy of the Aries Mini / Phoenix combo but two points short of Innuos’ ZENith MKII SE. Onto our crude source quality scoreboard, the ‘SE’ chalks an 8. It remains the finest sounding server/streamer heard by this commentator (at home) to date. It’s what I use for listening pleasure when I’m not evaluating other digital front ends. During this review, I was reminded of its ability to dial back the glassy piano ring of Kate Bush’s “And Dream of Sheep” and better specify Bush’s vocal in the “Jig of Life’s” soundstage potpourri, when compared to the Aries Mini.
In more esteemed company, we see the considerably less costly Wyred 4 Sound Recovery hit its glass ceiling. It subtracts two full points from the ZENith MKII SE’s standalone 8 for a 6. That plays in stark contrast to the Phoenix sitting between ZENITH MKII SE and Kii Remote where we say hello not to a 6 but a 9! That makes for a new king of the hill! From the ambient funk of Fila Brazillia’s Mess, I noted – again – a minor uptick in music’s overall ease but it would take the raucous guitar workouts of Neil Young & Crazy Horse’s Weld to clarify one specific facet of that ease: better dynamic elasticity in bass drum and snare hits.
Would I make the Phoenix a permanent addition to the ZENith MKII SE for a US$10K twofer? Not without first home auditioning the $13750 Statement. What I heard at Max Schlundt’s store in 2019 lacks the reliability of a home demo but it was enough to cause a present-day pause.
So far we’ve only looked at Roon streams pulled from a Nucleus+ server (US$2498) sitting in the kitchen. With the AURALiC, the Raspberry Pi and the ‘SE’ each operating as Roon Ready endpoints, all data travels through a Fritzbox router and Netgear network switch.
In relocating the Nucleus+ to the small table next to the Kii Three I could do what I’d not done before (but had long thought about): use the Roon piece as server AND streamer. My local music library lives on a 3TB ‘Rugged’ Lacie hard drive to occupy one of the Nucleus+’s USB ports. That left the other free for Kii Remote connection. The Nucleus+ would stream to itself.
It took me a full week to peg the Nucleus+’s standalone performance at a 5 (out of ten) and then another week to score the Phoenix’s addition at 7.5. That’s marginally better than the Aries Mini / Phoenix pairing and a shade short of the standalone ZENith MKII SE. Could the Nucleus+ pairing be this USB reclocker’s sweet spot?
Subtracting the Innuos box from this scene made the strongest case for its inclusion. With Elvis Costello and The Attractions’ visceral “Uncomplicated” (Blood & Chocolate), the solo-running Nucleus+’s stage depth falters, bass presence takes half a step backwards and drums are shaded with brittleness. Going back the other way, the Phoenix’s softer/gentler handling of upper-frequency transients makes it easier to enjoy Soulwax’s gnarly remix of MGMT’s “Kids”, especially at party levels.
In this final test scenario, we are reminded that the black box from Innuos doesn’t add to the incoming USB signal. It is inserted into the playback chain to minimise the damage done to the signal by the network streamer. The Phoenix’s intent is to pass the same data that it catches but clocked out more accurately and with lower electrical noise. Notice how USB data errors barely rate a mention? Listening tests tell us that the Phoenix can juice higher sound quality from the D/A conversion process. Just as Robert Fripp optimises David Bowie’s Scary Monsters, Innuos optimises the Kii Three’s D/A conversion.
The Phoenix isn’t an everyman hi-fi product. It’s more of a righteous demonstration of what’s possible from outboard USB correction. If you’re already committed to a network streamer and don’t wish to make the jump to one of Innuos’ ZEN devices, the Phoenix advances sound quality to put our digital front ends much closer to where we wish them to be than where they once were. Bravo.
Further information: Innuos
