Hello @martin.cassidy
Thanks, and one detail in your reply matters more than the rest.
The Nucleus One has no Wi-Fi. It only has an Ethernet port. So please tell us exactly what its network cable plugs into: the main TP-Link router itself, or one of the mesh satellites. If it is a satellite, then everything the Nucleus downloads crosses the wireless link between that satellite and the main router before it reaches the internet, and that link is the first thing we would suspect.
What the numbers say. Roon needed roughly 1,883 kbps sustained to keep the buffer fed and was getting about 400. That is not a marginal shortfall, it is a fifth of what is needed, and it explains why the low-res track failed too. Even 16/44 needs more than 400 kbps.
On the laptop comparison, which is the part that feels most confusing. It is a fair thing to try, but the two are not doing the same job, for two separate reasons.
First, the Qobuz app can buffer as far ahead as it likes. It is playing to one place and it does not have to deliver a bit-perfect stream, in sync, to a network audio device. Roon does. Roon has to keep a real-time stream fed to your AirLens continuously, so a connection that is perfectly adequate for the app can still be too slow for Roon. The app can absorb a slow patch by getting ahead; Roon cannot.
Second, Qobuz delivers audio to third-party applications through a different content network than it uses for its own app. So when you compare the two you are not even fetching from the same servers.
That is why your result and ours can both be true at once, and why we want to eliminate everything local before looking further afield. If the local side is clean, then a delivery problem on the Qobuz side becomes a real possibility and we will pursue it, but we cannot tell the two apart while the wireless hop is in play.
The test that settles the local side. Please take the Nucleus One temporarily to wherever your main TP-Link router is and plug it straight into the router with an Ethernet cable. It does not need to stay there, an hour is enough. Then play the same tracks. If they play cleanly, we know where to work. If they still fail with the Nucleus wired directly to the router, then the local network is not the cause and we move on to the delivery side.
Also, please open the TP-Link app and tell us what it reports for the satellite the Nucleus is connected to: its connection quality or signal strength, and whether its backhaul is wireless or wired.
One more thing worth mentioning, since it also points at “nothing has changed” being consistent with this. Mesh backhaul quality drifts on its own as neighbouring networks appear and channels get busier. A link that was just good enough can stop being good enough without you touching anything. Your own history fits that: it failed when the Nucleus was farther away, worked for months once you moved it closer, and has now started failing again.