Steve, those are good questions. I wish I was more of an engineer to be able to answer them. Instead, I can only parrot what we have all read here, and what the Novus engineers have shared with me. Purely from a spec perspective, the Kronos is a more accurate clock in both the short and long term, with less phase noise/jitter than the Master. So, we start by not making anything worse. And we give the Vivaldi Clock something to “check itself against” over time (long term). As James has written, it’s clear that a proper reference clock can improve the Vivaldi’s long term stability, but whether any of us can hear that may be unknown. Andrew has also made reference to the possibility of a real laboratory-grade clock improving Vivaldi’s performance, as opposed to the “audiophile toys” in the market (his words, not mine). So, my assumption here has always been that the goal is to help Vivaldi Clock do its job even better. I believe dCS designers and engineers put that reference clock input in the box for a reason, and not merely as an ornament or a checkbox feature, to give Vivaldi something even more accurate than itself. So, my goal has been to find the best reference clock I could afford (actual cesium clocks being out of my price range) that met the different criteria identified by dCS. The SRS Perf10 meets those criteria (and even bills itself as targeted toward the dCS stack), and the Kronos ups the ante with even greater precision and less noise/jitter. The Kronos has three different time sources: an OCXO, a rubidium clock, and two GNSS receivers that effectively bring cesium clocking accuracy.
Both the SRS and the Kronos have, for my listening, been sonically rewarding. The unexpected surprise, that I now prefer my system without the passive pre, has been the icing on the cake.
Among the experiments I am still trying is listening with Dither turned off at the Master Clock. This was something Ben suggested during a phone call we had recently. As I read the various explanations of dither here and here, it seems to help address a very specific dead zone aspect of the PLL architecture, and so seems unlikely to be affected by the reference clock. Still, it seemed a worthwhile experiment. So far it appears to me (so far) that both the dither setting and the use of an external reference clock operate completely independently. I hesitate to state any opinion here, because changing and comparing these settings cannot be done remotely, let alone blindly. I am always aware of what setting I am listening to. So, I try to listen carefully, for extended times, take good notes, and be alert to noticeable changes, if any.