Network Switches

Let me say first that I'm no expert at all in things digital.  But what I know as an indisputable fact is that there is RF noise associated with the interaction between toroidal transformers, amplifiers, other gear and routers/switches/cables.  This is easy to observe when I get LOUD obnoxious "motorboating" pulsing depending on placement of my router near my rack.  I get very audible noise carried on the physical ethernet cable as well if I'm not careful with placement relative to the rest of my gear.  I think none of this has to do with the 1's and 0's being transmitted as part part of the signal and has EVERYTHING to do with shielding and RF/EM interference.  Noise can get carried on the physical line (antenna essentially).  The materials that are used to shield the cable so it's not acting as a noise conduit is incredibly important.  I'm not convinced the "quality" of the bits transferred has anything to do with what my ears are capable of hearing.  I think separating the switch from the renderer in the router can reduce RF noise.  I used a $20 Netgear switch and observed this myself.  I will never pay $2000 for a network switch though!

Things that may or may not make a difference in what I will hear is that with my Roon endpoint, the Metrum Ambre;

"The board is completely and optically decoupled from the rest of the Ambre."

This is apparently done using;

"ultra-fast industrial optical decoupling"

and

"The network connection is also galvanically isolated."

My understanding is the main Pi board and the outputs board are optically decoupled.
They also apparently use two femto precision clocks by Tentlabs.

Looks like a gimmick.

Ethernet is always galvanically isolated. The connection is through a transformer.

@mitch2  Did the posts / responses answer your questions?