Improving SQ of Streaming

Moving from another thread.

I am not quite satisfied with the current SQ of music streaming in my system (a full Vivaldi stack), which is far from comparable to the quality of vinyl, SACD or even redbook CDs.

My system is externally re-clocked with a Mutec Ref 100 SE and it’s streaming music mostly from Qobuz through Roon (with a Roon Nucleus acting as the Roon Core). I am also using a SotM sNH-10G ethernet switch with a SoTM PSU.

Any idea of possible actions to substantially improve SQ (other than changing cables and the like)?


Numbered for ease of reference

  1. Did you initially use the system without the Mutec and without the switch? Did the addition of each make an audible and consistently positive contribution to sound quality?

  2. Have you tried using Qobuz direct rather than via Roon? I know of others who despite their love of Roon’s convenience settled on not using it because it did something to the sound which ultimately they didn’t like as much.

  3. How long and what sort are your router-switch cable and switch-streamer cables? Are they optical or copper, and if the latter are they shielded or unshielded?


Hi Franco,
Before you tinker too much have you tried removing roon from the equation?
I used to run roon but found it couldn’t replicate the quality from my Nagra cdp
When I moved to dCS I used mosaic to stream Qobuz directly to the dac with a £15 unshielded Ethernet cable……
Huge difference in resolution, depth and soundstage for me, to the point I no longer have the CDP


I guess I will start by saying where I am at. So, like you, Vivaldi stack ( as least as far as streaming is concerned. ( silver disc replay is Rossini Transport). For me with Qobuz UK the quality is excellent. Not quite as good as replay of some original CDs via the Rossini but the difference is only marginal and needs concentration n to spot. BTW paid for downloads from Qobuz remove that small negative.

My streaming of Qobuz is via a Melco N17hd used as NAS for UPnP but Qobuz is AFAIK a pass through. However the Melco is connected to internet ( router) with a Melco S100 switch.

The first thought is that Qobuz is different in each territory that it serves so one piece of information that may be of use would be the Qobuz service that you subscribe to.

Having a full Vivaldi stack means that you must have a Vivaldi system clock and I wonder about that Mutec re-clocking and its value. As you seem to be experiencing a worse outcome than I am then this might cast some doubt on the efficacy of the Mutec.

I would start to find out what may provide a solution by stripping your system back to basics. Use Mosaic rather than Roon ( a fair number of subjective postings that despite a “bits is bits” argument that Roon and Mosaic should sound identical the majority of postings suggest not) and remove all the Mutec stuff. Then reconsider what you are hearing. One can move forward from there if necessary.

Of course how it is connected can make a difference e.g USB v. network .


The Mutec had a positive effect on precision and focalization. Not sure about the switch.

Need to A/B again. Not always Roon sounds better.

Cable is unshielded, standard copper cat 5 or 6 (can be improved), except for a slightly better one shielded cat 6 silver cable between the Ethernet switch and the upsampler. 1m long. There is also a fiber converter switch (which I believe worsened the situation).

Both the Roon Core and the NAS sit in closets away from the system and are connected with standard cable. Both have cheap PSUs.

Thanks. How long and what type is cable from router to switch? I think I missed that.

Ignore “can be improved” for now as often “improvements” in ethernet cables are not improvements if used with a switch.


1 Like

The Vodafone router is connected to a multiple switch inside a closet in another room at least 20 meters from the audio system. The Vodafone router is connected to this multiple switch through a cat 6 cable, 1m long.

The dCS is connected through an Ethernet socket in the wall in a separate room.

I can tell you for sure that the SQ of streaming is far away from the SQ of my other sources. Miles away.

1 Like

Out of interest and for experiments sake, have do you the ability to run a long enough unshielded Ethernet cable directly from the router to the dCS (in mosaic) bypassing all other elements for now?


Not sure I can. In any case if I do this test I should also connect the NAS and the Nucleus to the same router. I’d need a 20m shielded cable and two other smaller ones.

Do you think this would improve SQ?

1 Like

Thanks Franco, that helps a lot.

So is this correct? Apologies for interview but it will help if we have a full picture:

  • 1m Cat 6 cable router to “multiple switch” - to be clear, do you mean a single switch with multiple ports? Does this switch feed direct into the wall and the next cable in the chain which is…

  • 20m cable inside the wall to ethernet socket. What cable is this? Cat 5? Shielded or unshielded? Not convinced it makes a difference but for completeness.

  • 1m Cat 6 cable from ethernet wall socket to SotM switch (shielded? unshielded?)

  • 1m shielded Cat 6 silver cable from SotM switch to upsampler

Here’s an exciting theory: let’s assume the above is correct and the 1m Cat 6 cable from wall to switch is U/UTP = unshielded, and let me take a punt at a possible solution.

20m in-wall ethernet cable can pick up one helluva lot of noise.
I suspect the problem here is your shielded cable from SoTM to upsampler. This cable should be UNSHIELDED. a shielded cable negates the positive impact of galvanic isolation offered by a switch. That means 20m of noise is getting straight through to your upsampler.

Experiment: swap the two 1m cables attached to your SoTM. Or if you have another stock unshielded (U/UTP) Cat 6 cable, use that instead. That should let the switch do its job and bring a significant improvement in sound quality. Report back.

Experiment 2 take Roon out of the equation as @Klint and I have mentioned. Report back!

Only when you’ve tried this should you consider power supplies and other stuff.

I have my fingers crossed (not sure this is a global thing…)


I agree with others, stripping back your system to the essentials is a good place to start - a single run of cat5e from your Vodophone router to the Vivaldi up sampler and play back via mosaic will give you a reliable baseline to evaluate from.


Also be sure the connection status LEDs on your SOtM switch are turned off by pushing the power switch down instead of up. Not likely to be your root cause, but in my experience those LEDs are not good for SQ.



Sorry I am a little confused now.

Since I have various ethernet switches in the apartment, are you suggesting to try and by-pass this issue by connecting directly the dCS Upsampler to the Vodafone router with a single cable? If that is the case, as I said in my previous post, I would need a very long cable (20m?) since the router is sitting inside a closet in another room.

If you do not recommend the single cable route, which ethernet cables would you recommend to use and where? What do you think about the ethernet switch?

Why do you say that unshielded is better than shielded. I am not an expert but I’ve noticed that TOTL cables are generally shielded (?).

Just to be clear, here is a recap of my current situation (apologies for lack of accuracy but I am not sure what type of ethernet cables I am currently using, except for their respective length. I just believe that, except for one they’re stock cheap cables):

  • the Vodafone router is connected to a device called Fritz 7590, which in turn is connected to a fast gigabit switch with multiple ports. I suspect cables are cat. 5 unshielded;

  • the switch ports correspond to different ethernet socket in the various rooms of the apartment;

  • there are three Fritz repeaters in other rooms creating a meshed Wi-Fi network;

  • the Ethernet SoTM switch is connected directly to the wall and to the dCS upsampler. 1m long shielded cat 6 silver cable (not sure which connection);

  • there is a fiber converter switch in the line between the wall and the SoTM;

  • both the Roon Core (Roon Nucleus) and the NAS sit in closets away from the system and are connected with standard cable.

Thanks again.

Apologies for staccato responses as I’m on work calls for much of the day.

Just focus on the cable from SoTM switch to upsampler first. Ensure this is Unshielded. Don’t change anything else yet. Report back!

This last stage is the most critical. It should clean up most of the noise accumulated to that point IF you let the galvanic isolation do its job. A shielded switch-upsampler cable won’t do that.

1 Like

Thanks Nigel. Will give it a try.

1 Like

Hi Franco,

I’m confused by these two statements:

Are you saying you have an Ethernet cable connected directly from the wall to your SOtM switch, PLUS you have another Ethernet cable going from the wall to a fiber converter, then a fiber cable from the fiber converter to a fiber port on the SOtM switch?

The reason is not down to personal preferences but electrical engineering. The shield on a cable acts similarly to a Faraday cage. Noise picked up by the shield has to go somewhere or it may be induced into the conductor. The “somewhere” is to earth ( ground). dCS equipment t was designed to be used with unshielded ethernet cables and hence does not have lugs to connect a cable shield to earth. Unshielded ethernet cable is in any case inherently insensitive to noise used in a normal domestic setting.

So unshielded it is and all that it needs is to be properly certified, TIA or ISO standards ,as Cat 5e or 6 ( many audiophile cables are not, there is no official Cat 7 or 8). If you wanted to try that 20m length excellent cable is pretty cheap even in such a long run. As an example I just checked the cable I use ( I will not recommend specific cables in a forum) is 69 g.b.p custom built in that length.


There is an ethernet cable connected from the wall to the SOtM. Then there is a fiber optic converter switch sitting between the SOtM and the dCS Upsampler. Ethernet cables

Thanks Peter.

Ahhh, I understand now. Thank you.

PS: After you try the unshielded cables per Nigel’s and Pete’s recommendations, you might try placing the fiber converter on the wall side of the SOtM, with only the unshielded ethernet cable from SOtM to Upsampler. Maybe the power supply for the SOtM is quieter than the fiber converter’s.

1 Like