Nuno explained that like any USB hub, Phoenix works with any USB device. Think mouse, keyboard or printer. Reclocking USB packet transmission as he does isn't audio specific. Asynchronous DACs thus continue to operate in async mode. Unlike USB chips in hubs which generate their voltages with on-chip switching regulators, Nuno's carefully selected USB chip receives the three voltages he needs—1.2, 3.3 and 5V—from discrete high-end regulators. He called them the lowest-noise such parts currently made. With them he completely regenerates USB power and ground. Where typical USB clocks operate at 30ppm, his operates at 3ppb (that's parts per billion). Unlike common 10MHz clocks, his 24MHz clock needs no synthesizing by phase-lock loop. To assure ideal operation, his has its very own dedicated power spur. USB implemented correctly is his favored protocol because it is open. It supports all existing formats and sample rates and will support whatever comes next. This goes back to the company name. Innuos is a contraction of innovation + open system.
As to why/how optical/galvanic isolation can fail to benefit from upstream noise filtering, Nuno asks why Toslink hasn't become the gold standard for digital audio transmission. Conversion from electrical to optical and back to electrical is far from lossless. Likewise for magnetic isolation by miniature isolation transformers. Those act like their own low-pass filters when digital square waves require infinite bandwidth. Since infinite bandwidth is purely theoretical, squares waves "tremble". Minimizing their overshoots and wiggles which are noise i.e. not the signal helps a DAC work better. It's all about implementation. If a particular solution works better in a particular instance like a certain optical isolator for example, it is superior in that instance. But it's no universal panacea when in a different system, something else works better. Asked whether with USB there's a specifically critical bandwidth for 'digital' noise—a misnomer when a square-wave signal is analog even though it describes a binary signal with rapidly switching voltage states—Nuno explained that it's broadband. It can occur from in the kHz well into the GHz bands. Much depends on the actual hardware a listener combines. "If it were more specific, we'd have one universal fix. The difficulty for any engineer is that this noise varies from installation to installation. It can sit anywhere across a very broad bandwidth. That's specific to the components in use. Plus, there still are many things we as an industry don't yet understand about digital. We know a lot more than we did 10 years ago but I'd call anyone who claims to understand it all delusional." Nuno actually used a far spicier phrase. Finally he pointed out how unlike SOtM's modular building-block approach of reclocker with wall wart + outboard linear power supply upgrade + external superior clock upgrade, Phoenix puts all these functions into a single box. Then it executes them at the highest level without upgrade loopholes. "We don't like to play upgrade games with our customers." Hence Phoenix is a USB reclocker + regenerator + premium linear PSU all in one chassis and priced accordingly.
Popping its top, Phoenix revealed a symphony in mostly black best seen inverted and color enhanced. Its guts consist of five well-spaced assemblies – a shrink-wrapped AC filter probably from Schurter with a separate 115/230V selector; a 120VA toroidal power transformer; a first power module with 4 x 10'000uF Mundorf electrolytics bypassed with a small Epcos and eight voltage regulators fed from the transformer; a second power module fed from the first, this with six transistors coupled to the front panel as stylish heat sink; and the digital board with the clock and USB i/o fed by the second power module. From this hardware evidence, Phoenix isn't a stuffed turkey but leaner bird. Of course the job at hand is just to reclock a USB signal. These goods are already over-engineered and a far cry from the switching wall warts at the tail ends of cheap far smaller USB reclockers. The included USB cable was a step up from generic and came sleeved in white/black weave with big ferrite beads on either end. The inside cover of the case was damped with a thick bitumen mat. Time to put the lid back on and, like Nuno had asked, 24 hours of signal to get his temperature-controlled clock on song. After that he suggested not to turn it off again. That squared neatly with my habit of never turning off digital, just analog.