Roon Core Alternatives What Actually Worked

General
1

In a previous post I asked the question “Alternate Roon Core. What are my options?”.

As promised, I finally completed the comparisons I mentioned earlier and wanted to report back with real listening impressions rather than theory.

For context, everything terminated at the MSB Cascade with Cascade Clock where applicable. Same room, same power, same cabling, extended listening sessions over multiple days.

Here is the full set of paths I evaluated.

  • Roon Nucleus Plus B to MSB Renderer over Ethernet
  • Roon Nucleus Plus B to Cascade ProISL via USB
  • Roon Nucleus Plus B to Aurender N20 to Cascade via ProISL USB
  • Roon Nucleus Plus B to Aurender N20 to Cascade via AES EBU with Cascade Clock
  • Tidal and Qobuz direct on Aurender N20 to Cascade via ProISL USB
  • Tidal and Qobuz direct on Aurender N20 to Cascade via AES EBU with Cascade Clock

I also added two more variables during the process.

  • Xact Audio S1 Evo Server to Cascade via ProISL
  • Antipodes K41 serving as Roon Core to MSB Renderer via direct Ethernet using its secondary network port

Starting with the Nucleus. Running Roon from the Nucleus directly into the MSB Renderer over Ethernet versus USB into ProISL was largely a wash. Differences were subtle at best. If someone told me they preferred one over the other I would believe them, but neither path fundamentally elevated the presentation. Clean, competent, but emotionally a bit matter of fact.

Things changed immediately when the Aurender entered the chain. Using the Nucleus as the Roon Core and feeding the Aurender N20, both ProISL and AES EBU sounded better than the Nucleus alone. More density, better timing, more authority. Between those two, AES EBU with the Cascade Clock had the edge. Better image solidity and low frequency grip without sounding etched.

Dropping the Nucleus entirely and running Tidal and Qobuz directly on the Aurender was another clear step forward. Less digital haze, more flow, and better spatial organization. Again, AES EBU with Clock was the standout. ProISL was excellent, but the clocked AES path brought more coherence and drive.

Then came the Xact Audio S1 Evo. This server is really good. The soundstage depth and front to back layering were exceptional. Presence was the word that kept coming to mind. Voices and instruments had real dimensionality. That said, compared directly to the Aurender, it gave up a bit of slam and macrodynamic punch. Depending on system balance and musical priorities, I could see someone going either way here. Pure musical realism enjoyment the Xact wins this one.

Finally, the Antipodes K41.

This one surprised me the most.

Used purely as a server and Roon Core, replacing the Nucleus, and taking advantage of the second Ethernet port to stream directly into the MSB Renderer with nothing in between, this was the best sound+features of the entire comparison. Not incrementally better. Clearly better than all but the Xact where there were tradeoffs but I think you could ask 10 people and 50% of the time you’d get one or the other as the answer. One standout on really poorly recorded digital files, the Xact got a bit more lipstick on the pig, not by a mile, but enough that it’s worth noting, question becomes is it editorializing in some way - I don’t know but it sounded good.

It was miles ahead of the Nucleus in every aspect of musicality. It combined the depth and dimensionality I liked from the Xact with the authority and rhythmic confidence of the Aurender at its best. Tone density, microdynamics, bass articulation, image stability, all improved. And as a bonus, I retained full play, pause, forward, and back control from the MSB remote, which matters more than I expected.

I have not yet heard what Pink Faun or Taiko are doing, so I am not making any claims beyond my own system and experience. But based on everything I have tried to date, the K41 into the MSB Renderer over Ethernet is the most complete and musically convincing solution I have had here.

It will be staying for the foreseeable future.

Hopefully this helps anyone else navigating similar questions. Happy to answer specifics if useful.

Cheers,

Matt

8 Likes
2

Great comparison! Thanks Matt and welcome to Antipodes family! In my system Antipodes really brings music to life! Great combo with MSB gear. Please do try the different server mode (eg. squeeze/minimserver), you might be even more surprised! :smiley:

Wade

3 Likes
3

Matt, thank you so much for your comprehensive comparisons.

I am wondering whether you are using Roon/Roon for server/player or has tried also Roon server with Squeezebox player?

I am using the Antipodes K21 via USB output to a ProUSB module to a ProISL input. I used Roon with Squeezebox player softwares for the unbeatable user interface!

1 Like
4

I am always puzzled by posts like this. The Cascade DAC, like most good DACs these days, galvanically isolates its inputs and uses its own internal clock for conversion. As long as the music file that arrives in its input buffer is an exact copy of the file that left the recording studio, nothing that happened in between should mater at all. I can not come up with a technical theory to explain what you experienced. :thinking:

2 Likes
5

Intellectually I am in agreement with you. As a technologist I have been trained to know an 802.11 packet that arrives successfully it s perfect packet and if it’s corrupt then the protocol asks for it to be resent. So there’s no way there could be a difference. I can logically explain the differences between the non-Ethernet connections.

That said hearing is believing. I’ve never been a believer that unmeasurable nuance can exist, but I am starting to think there are sensitivities in the human ear/brain that cannot be measured. The differences were very apparent. Maybe it’s all just bias, but I didn’t have my heart set on one device or another to be the winner.

One of life’s many mysteries I guess.

2 Likes
6

This is just my own opinion.

The Digital Myth

As always, use your own ears to decide if a piece of equipment is useful in your system. That’s really the only measurement that counts.

It’s about the noise, whether self-generated, conducted, or absorbed, or whatever. Anything you can do to quiet down a “digital” source, like a streamer, will cut down the noise it injects into components it is attached to… No matter what means is used to attach it, no matter how that component filters or processes that signal, some noise will get through.

It is a myth that there is such a thing as a “digital” signal, all signals are analog and are subject to all the grief that noise causes in signals. If you shove noise into one end of a glass fiber, or an air gap, or an isolation transformer or whatever, some of it will come out the other end. All signals, including digital, include noise that just hops onto it for the ride.

So, in general, things like Aurender, Pink Faun, etc. that are designed from the start to be electronically quiet, will inject less noise into Cascade (or any DAC) than purely digital designs will. How much that affects the overall system they are in (that includes the listening room too) depends on how precise the system is. Lots of systems won’t notice any difference at all because they are not that precise to start with or include other noise generators that mask the difference between a quiet and a noisy digital source.

Dan

4 Likes
7

Although digital signals are represented as analog voltages in connections like Ethernet and USB, that does not make them analog signals. For Ethernet, any positive voltage differential between TX+ and TX- above a specified threshold is a one and any negative voltage differential below a specified threshold is a zero. USB is similar except the wires are called D+ and D-, and the thresholds are different. The beauty of digital is you can send a perfect signal through an imperfect world. It doesn’t mater if there is some noise as long as it is not so much that it causes bits to be misinterpreted. Noise is not propagated through a digital signal path like it is with analog. Each device in the chain receives the data and then retransmits it fresh. You can stream music from a source half way around the world and still get a perfect signal no mater how many routers and switches the signal passes through. The final device, which is the DAC in our case, ignores any noise in the signal it receives in the same way.

2 Likes
8

We need to keep in mind that the noise in the system can easily piggyback along the retransmitted data. Not that it is copying and republishing the noise, but that environmental noise will persist on a “local” system. I.e. from transport to Dac and amp and then speaker.

Absolutely, the local system level noise is the area to focus on.

Indeed. Optical barriers are some of the best to negate this. That is why we have multiple fiber barriers with the Pro USB and Cascade Link.

What we do at MSB is give every digital signal as much filtering and reclocking as possible so that the source is less critical. We want our DAC’s to sound as good as possible in any system. That being said, a better source is indeed a better source. On some level, you can always push for improvements.

We have bit perfect testing with sources, and still hear differences. Jitter an noise have to be accounted for.

1 Like
9

Claims of digital noise propagating through a DAC into its analog outputs are usually difficult to argue against, primarily because they lack specificity. First, it is usually not clear why this noise is believed to exist. Did someone measure it? Was it so bad someone actually heard it? Is it purely theoretical? Second, there is usually no description of the path it took through the DAC from digital input to analog output. Did it pass through the ground plane or power supply rails? Was it EMI through the air? Can you elaborate more on the environmental noise you mentioned? How does it get into the DAC? Where does it persist? How does it make the jump to the analog outputs?

My previous example of streaming from half way around the world is relevant because it is evidence that complete digital input isolation is possible. If noise were to piggyback along through routers and switches in a long digital signal chain, it would eventually accumulate to enough noise to cause errors. This does not happen and is not an issue that receives any thought in the world of network design. If common routers and switches can isolate their inputs and not propagate noise, why can’t a DAC?

Thank you for the discussion Daniel. It is cool you guys are so engaged.

3 Likes
10

Hehe, all of the above! @Dustin_Symanski would be the best to discuss this topic as I am the mechanical guy.

I think I saw on your other thread discussing noise in the “audible” range as well. The thing is, it does not need to be in the audible range to be detrimental to the audio. A lot of the digital filtering and dithering happens outside of the “audible” range, but has an immense impact of the sound. High frequency noise can adversely excite and affect electronics which will show up in the performance. So much of the work we do is power conditioning, isolation, EMI protections and it all matters.

In the Sentinel Analog converter, we have many EMI barriers in the product to protect from the environmental noise, but than additional barriers around the clock to protect it from the internal DAC electronics. The rabbit hole goes deep.

2 Likes
11

As Antipodes K50 and Rf+DD user, after trying Roon+Roon: k50+MSB renderer V2/K50+MSB USBv2/K50+MSB AES; Squeeze+Squeeze: K50+MSB USBV2/K50+MSB AES; Roon+Squeeze: K50+MSB USB V2/k50+MSB AES; there are some difference between different ways. Normally it is keep with K50+MSB Renderer V2, it is a little bit better than other ways for me. Also it can be changed another ways sometimes to get different feeling .