Sonic differences between servers

However, the Auralic server easily bested the MiND server built into the Sim unit.

It seems the term "server" is being misused here.

What @mgrif104  compared was two different software player/renderers and how they handle a buffered audio stream, from a server somewhere else. The Sim Audio does not have a server, nor do the Auralic products. The above statement is simply not true.

In an attempt to clarify for the OP - the term "server" refers to the computing device that the file you are streaming originates from, as well as the software component necessary for the hardware to provide the file to an endpoint using some such or another protocol.

The reasons servers can vary in terms of sound quality can be due to many different things, however in a nutshell the less processing/manipulation of the data happening in the server's software-based component typically has the largest impact.

In other words, your question is about the affect the "server" has on the SQ, not the player or endpoint, and so the differences between the two should be clarified.

A server’s quality depends on noise elimination
- from the ethernet in (RFI/EMI galvanic and filteror through optical)
- from USB out (ground noise and RFI/EFI from power supply, vibration)

Again, the confusion between the player/endpoint and the server here needs to be cleared up. A device with a USB output to a DAC is a renderer/player/endpoint, not a music server. It may be a component shared within the same chassis as a server, and is typically a simple piece of software on a relatively simple architecture.

Also the idea that the quality of a server is only based on elimination of noise is a highly flawed viewpoint. The software component of the server is likely just as (if not more) important than the hardware in terms of SQ.

If the job of the server is to send the digital signal to the DAC

That is not the job of the server, that is the job of the player/renderer (an application layer software) and it's associated hardware interface capabilities (i.e. USB, SPDIF, AES/EBU, Ethernet stream).

A server does what it implies. It serves files to any number of potential endpoints, each of which can have their own unique player/rendering software. 

With respect to Roon, a Roon Core contains both the server and the player. So your music "server" in the case of Roon, is also your music "player". The Roon Core sends real-time audio playback out over your LAN to a Roon certified endpoint. The server/player/endpoint scenario with Roon is therefore a bit more complicated, because essentially the server and the player are the same device; the Roon endpoint has the ability to "capture" the real-time audio playback happening on the Core and sync the file playback to the DAC, over the network. (What they call RAAT, which is basically a glorified version of AirPlay given the above analysis).

In this case the difference between endpoints is going to be related to how they interface with the DAC, since the player is identical between Roon endpoints.

I suppose what I am trying to point out is that because of the confusion surrounding all of these terms and products, speaking about differences in sound quality pretty much ends up becoming more or less a moot point when it comes to "servers", especially if you are using Roon. Personally I would avoid Roon altogether if SQ is actually your ultimate goal.  

Files streamed via Auralic’s lighting DS application sounded significantly better than the same via Roon.

I'm just curious if the files were local (i.e. on a NAS or computer) or did you use a streaming service like Tidal or Qobuz?

In either case you heard the difference between the Roon player software (on whatever you were using to act as the Roon Core) and the player software embedded in the Auralic streamer. The Lightning DS app is just a control point software, meaning you made your song selections using the app, but the software doing the playback itself is in the Auralic (the App just provides control and feedback, not playback in other words).

I've had similar experiences so it's not surprising to see them confirmed by others!

I haven’t an explanation, but have theorized that LDS manages the substantial buffer in the Auralic better than does Roon.

When using Roon Qobuz content is requested by the Roon player on the Core, and then multicast to Roon endpoints (like the Auralic) using RAAT as a protocol on the network.

When queuing Qobuz content using the Auralic Lightning app the Auralic Aries' built-in player software requests the content directly, so it is a much simpler process, with fewer hops, and no concern for the overall quality of the upstream hardware (since the Qobuz content is spread across servers no one knows exactly where it is coming from).

OP, it sounds like what you are looking for is an improvement in sound quality by replacing the Roon Core you are currently using (Nucleus) with a better one. As far as I am aware the best products for this job are the Antipodes line of music servers. You can purchase just the server to run Roon Core for Roon endpoints on the network, and they also offer both separate endpoint hardware (players) as well as music servers with integrated players.

Also what would help to guide you in the right direction is whether or not you are using the Roon Nucleus as a Core (server and player) connected directly to your DAC via USB, or if you are using the Nucleus as a Core with endpoint (i.e. Roon Ready) hardware built into your DAC, or with some other endpoint hardware other than the Nucleus.