To take the ethers, I had upstairs and downstairs. Upstairs uses a 6m industrial CAT8a Ethernet link for I²S transmission between Soundaware's D100Pro SD card files and Denafrips Terminator Plus (no PC/server in this system). That length allows me to place the source next to the listening chair and read its display without squints. Unlike S/PDIF, I²S doesn't embed its clock in a serial stream. It runs parallel data legs so the clock gains its own conductor. It needn't be extricated by the DAC. Downstairs a 27" iMac serves as similar next-to-chair source with keyboard and mouse not remote. That supplies a Soundaware D300Ref with USB data through Audirvana's buffer spinning down Apple's FusionDrive and caching music in RAM. Acting as pure USB bridge, the super-cap'd D300Ref dispatches reclocked digital data to a Sonnet Pasithea R2R DAC via AES/EBU.

In both instances Aether would 'remotely' tiktok the Soundaware clocks to present the DACs with a more precisely clocked signal. The converters obviously still run their own clocks. Why serialized reclocking would have audible gains I don't grasp. I've just heard it before also with a Mutec reclocker. As so often, understanding isn't needed to benefit. I leave that to the engineers. iFi's Pro iDSD Signature DAC on my desktop also reclocks the preceding Singxer SU-2 USB bridge. That too makes a difference into DMAX SC5 active speakers.

Taking the air(s)? It's an old way of saying that one goes for a walk. Putting on airs even without graces still means acting superior. How about external masterclocks for home hifi? On one hand, it's another example for when separates create a higher box count with extra cables. On the other hand, there is a difference between home cooking and fine dining. The Aether concept with its Terminator Plus oven-controlled crystal oscillators struck me as fine dining without the pretentiousness. What if a DAC already cooks at two Michelin stars? At what star count does the potential for sonic improvement by outboard clock stop? Where l play, a reclocker packaged as a USB bridge between PC/Mac and DAC thus far always made a difference. Taking a page from Mutec's book, here we see parallel/star and serial/chain ways of distributing a masterclock signal in a recording studio where more than just three devices may share a clock to want syncing. With Aether's eight outputs, even complex home-based systems can adopt the star configuration at left. I'd clock just one slave device at a time. From this graphic we appreciate why external masterclocks are more typical¹ in complex audio/video production facilities and far less common or even necessary at home.

As Cees Ruijtenberg of Sonnet/Metrum finds, "there's a somewhat exaggerated notion on reclocking. The fact is that you never get perfect sync with a studio clock. Jitter will shorten or lengthen samples. In an ideal world, the DAC would follow the longer or shorter samples which isn't the case. So it's quite senseless to apply an extremely high-quality clock except for commercial reasons. I regularly see that if a central clock distributes over adjacent studios with different cable lengths, synchronization between studios no longer applies. Now comes a problem. Suppose a multitrack has several musicians play their parts. The next studio's DACs play a sample that was originally long but in the mix becomes short. For the DAC it's very important that the edges of both clock and 0/1 transitions on the data line and left/right frame jump simultaneously to another state. An external clock can cause a time offset whereby the edges of the data lines no longer equal the clock so shift in time. The clock ignores this until it switches itself to another state. Based on those undesirable effects, a high-quality PLL is the better solution but still won't solve the studio's jitter. The problem remains that in some cases, a long sample translates as short. I believe that's why making jitter ever lower must end somewhere. Beyond that, measurable improvements stop being audible. I understand from researchers in this field that in some cases a central clock even leads to more jitter than a PLL would have corrected locally."

¹ Video production mandates a masterclock since each picture frame must sync to an integer number of audio samples.