Crap. Birds do it. Ask me about the two swallow's nests above our entry. So do lesser power supplies, physical vibrations, proximity of the wrong parts from sub-optimal PCB layout. It's why black USB boxes that do better are colloquially known as USB decrapifiers. Going in, we can assume that Innuos paid special attention to their linear power supplies, to stabilize and filter sundry voltage rails, to select their temperature-controlled 24MHz clock for best stability and long-term behavior. We assume that they heavily regulate USB's 5V to remove noise from the source PC then regenerate the line.
We might assume that the Phoenix also reclocks the music signal even though the majority of modern DACs use their own clock in asynchronous mode whereby the receive and not send clock becomes the master.
But Innuos director Nuno Vitorino was on record that "we don't reclock the audio signal at all, just the USB communication protocol between source and DAC."
The complete USB protocol is well beyond my pay grade. If we care to read, Wikipedia has the basics. There's more to it than just one type of data package. Perhaps Nuno could point out the essentials of transferring music data from host to DAC and where superior clocking makes the difference of better USB reclockers?
Before my traveling review loaner landed, our German colleagues at fairaudio.de had published their findings. For his Kii III active speakers, Fritz Schwertfeger had a Mutec MC-3+ USB comparator. "It couldn't keep up though in fairness, it costs less than half and offers far more expansive socketry." He found that "to expect tonal balance changes is wrong. It's not about more bass weight, higher treble energy. The qualitative differences are soft skills like gains in atmosphere, in rhythm, timing and dimensionality. Such apparent intangibles determine whether a musical piece convinces us or not. Comparisons showed that without the Phoenix, things grew more studio stark, with it more nuanced and lively. Intensity and expressivity scaled up, image localization and depth layering became more effortless to comprehend." Fritz even used a headphone setup with MacBook, Chord Hugo 2 and Meze Empyrean. Much to his surprise, here too he found the Phoenix improvements "far from homeopathic. Without it, what went amiss was a certain plasticity, a microdynamic tension and the more holographic layering."
Via factory contact Mandy de Castro, I sent Innuos my four key questions.
• Today’s DACs use asynchronous mode whereby the receive not send clock is the master. If one uses such a DAC whose USB transceiver establishes the async protocol, does the Phoenix override it so that its own superior clock becomes the master? If not, why/how would its better clock make a difference?
• Nuno is on record with "we don’t reclock the audio signal at all, just the USB communication protocol between source and DAC". How does that differ from what other firms are doing? What in the USB comm protocol is critical to audio and how/where does reclocking that (rather than the audio signal) improve sonics?
• How does galvanic isolation by transformer, optocoupler or silicone-oxide isolation barrier fail so that USB converters which already use it still remain susceptible to noise to benefit from preceding noise filtering by so-called USB decrapifiers, reclockers and D/D converters?
• Most D/D converters I'm familiar with eliminate a USB output to instead convert USB to I²S, Ethernet, AES/EBU, S/PDIF or coax. Why do you prefer to output USB and eliminate all other formats? Is it purely to avoid sample-rate limits?
To illustrate a real-world not theoretical digital signal, the graph showed what on the next page Nuno calls "trembling" square waves. That's because electronics don't have infinite bandwidth to track high-frequency square wave signal with its theoretically endlessly stacked harmonics without introducing distortion.