When you transmit a music file over the network, it can be in a variety of different forms. It can be a voltage in a wire, a radio wave in air, a light pulse in a glass fibre, a file stored in RAM as a series of charged and uncharged capacitors etc. etc. In order to exist in these different forms, the stream typically has to be converted from one form to another….from voltage to light, from voltage to radio waves, from radio waves to ethernet parcels, from ethernet parcels to a series of charged and uncharged capacitors on a chip. Etc. All those processes have certain physical characteristics….wavelength, intensity, power, time, voltage etc. All work within tightly defined specifications and all suffer from irregularities caused by noise of a multitude of different types and sources, depending on the physical state. For example, a crystal oscillator may be affected by HFI or vibration which interfere with its resonant frequency and cause minute variations in the timing of its output, resulting in jitter and possibly phase noise. A power supply may respond very quickly or very slowly to demands, depending on things like impedance, reserve power, regulation, quality of magnetics etc, resulting in clean or ragged edges to square waves. The supply may generate its target DC voltage but may also output noise coming from its own componentry, from its mains supply or from nearby sources of EMI. The noise spectrum from one PS may interact with that of another PS to give complex harmonics.
Every process on the network will provide a cleaner, more accurate, purer and more perfect output when its own input and components are optimised. The cleaner, purer and more carefully each step is performed, the cleaner and purer the final input into the DAC……each process on the network has a particular sound signature that impacts the final sound. The effect may be small or large but it’s always there.
A shortcoming early on in the network transmission will ripple through the entire network to impact the output. For example, routers based on the Puma chipset will almost always sound inferior to routers based on the Broadcom chipset. While that shortcoming can be compensated for in the balance of the network, it can never be removed and will always impact the final sound. Improvements later in the network may offset the effects of the router, but you’ll always hear an improvement when the router is upgraded, regardless of how much better the rest of the network makes the sound. Taking out anything, anywhere in the network that causes degradation to the purity and accuracy of the physical structure of the stream will always bring improvements. Whether its better cables, better power supplies, less EMI, less vibration, less mains contamination, better power supplies, less network traffic, less error correction, less latency, fewer CPU interrupts, whatever. Obviously some improvements have more impact than others and not all changes are improvements. Some improvements early on may be offset by major degradations later in the network, so changes that are clearly heard on system A may be almost completely masked on system B.
On to the question of Wi-fi. Essentially we are talking about converting a voltage to radio waves, then back to a voltage. This is essentially a noisy conversion and while it will certainly provide the system with galvanic isolation the degree of improvement will very much depend on the quality of the components employed and how well the Server and DAC are isolated from the RFI. I used Wi-fi in my system, but subject to a lot of improvements, The transmitting router was isolated from the rest of the network with its own dedicated 5GHz band, it was placed on a vibration isolated platform, it had a high-grade LPS
and a custom DC cable. On the receiving end, the Wi-Fi - Ethernet Bridge was provided with vibration isolation, its power supply was massively improved and its Ethernet output was passed through a cascade of switches, before being fed into my Server. Untreated, Wi-FI provided a small improvement over a directly wired Ethernet cable, likely due to the inherent galvanic isolation. The treated version was transformational.
The above photo shows the TPLink RE650 Wi-Fi-Ethernet bridge on wall mounted isolation. The second photo shows the literally cheap-as-chips RE650 power supply that was replaced by a Sean Jacobs Mini ARC6 DC4