Hi Lee,
Although pretty simple on the face of it your question got me thinking. So although I am not sure this provides an answer (in fact I’m fairly sure it doesn’t) here are my thoughts on some of the considerations here.
Firstly, what is a music server? I came up with three pieces of core functionality:
- Storage for your music files.
- A management layer allowing you to select and play them. Most that I am aware of also integrate streaming services behind a “single pane” and provide a nice UI.
- Inherently a server also forms the “bridge” between the “asynchronous domain” (hard disk or IP network) where music is encapsulated in files like DFF or FLAC and the synchronous domain where a clock manages feeding samples/a bitstream to the DAC. So servers tend to have asynchronous inputs (ethernet) and synchronous/isochronous outputs (AES, SPDIF, USB) and in some cases clock inputs.
And I interpret “benefit from” in three ways:
- Functional. Does adding a server add some new functionality or capability?
- Non-functional. Does adding a server improve convenience or usability?
- Non-functional. Does adding a server improve sound quality? (I suspect this is where you were going with your question).
From a functional standpoint all current dCS DACs (or systems in the case of Vivaldi/Varèse) offer all the above functionality except local storage. The dCS architecture accesses local storage, typically a NAS, via UPnP. From a storage perspective these servers may offer usability/convenience benefits compared to a NAS, but I think that depends on what you compare with what.
When it comes to connectivity there are basically three (I’ll ignore Toslink here) choices of protocol for connecting an external server to a dCS DAC: AES (single or dual), SPDIF or USB. For PCM, whereas dual AES and USB support up to 24/384 (the same as the ethernet input) single AES and SPDIF are limited to 24/192. For DSD, all inputs support DSD/64 and DSD/128, but only the ethernet input supports native DSD (DSF/DFF), the others only support DoP as mandated by Sony’s licensing restrictions. So using an external server restricts the source material you are able to play in some way regardless of which input you use, i.e. a drawback rather than a benefit. How much this affects you depends of course on what you listen to.
In terms of convenience/usability most of the servers I have seen (Aurender, Melco, Antipodes, Lumin, Taiko etc) come with their own Apps providing a user experience that is richer and many seem to prefer to Mosaic which is rather “minimalist”. So that would seem to be a benefit in many cases. However for those who prefer a third party UI (i.e. neither dCS’s nor the server manufacturer’s) such as Roon, Audirvana or Tidal (via Connect) this is moot.
Sound quality is where I think it starts to get really interesting. I should say straight off the bat that I have never compared any servers with my Vivaldi Upsampler in my own system so I have no empirical observations or opinions to offer (sorry about that!). But looking at the way dCS products are architected and how well external servers fit into that architecture a couple of things become clear:
- A core part of dCS’s philosophy is the importance of clocking, and their architecture (whether integrated like Rossini or discrete like Vivaldi) allows the source to be slaved to either the DAC’s clock or an external master clock. To provide maximum clock integrity dCS separates the clocks for the 44.1kHz and 48kHz families so for maximum compatibility with dCS’s clock architecture an external server should be able to accept two separate word clocks.
- dCS’s own preference for interconnecting their units in the “synchronous” domain in the L/B/R/V generation is dual AES. Although USB provides equivalent sample rate compatibility dCS makes it clear that they have provided this interface for compatibility with USB sources (such as PCs) and it is not their preferred interface for ultimate sound quality. So if you want to follow dCS’s philosophy in this regard you will be looking for a server with dual AES outputs.
The only product I am aware of that meet these criteria (dual AES inputs and ability to integrate with the dCS clocking architecture) is the Aurender W20SE, where clock compatibility is provided via an external dongle. All others provide only single AES (if at all) and many offer either only reference clock inputs or no external clock inputs at all. Note however that connecting a W20SE to the dual AES inputs of a dCS DAC will entail cable swapping if a dCS Transport is also in use and will bypass the Upsampling functionality in the case of the Vivaldi system, so it is not exactly a “hand in glove” fit into the dCS architecture. Most server products I have seen appear to be optimized around their USB outputs and targeted at DACs optimized around their USB inputs.
So, while none of this really helps answer whether adding an external server will improve sound quality, my observation is that while it will bring local storage and possibly some usability benefits it will in most cases also entail some sample rate restrictions and/or stepping away from dCS’s preferred clocking and system interconnection schemes. Looking forward to hearing others’ takes on this.