Anybody use Innuos Zenith or similar server with Bartok or Rossini?

Audio Systems
21

Thanks. However the reason for the diagram showing that configuration is that Melco are proffering the switch as something that can be used universally and not as exclusive to existing Melco based systems. As you can see neither music player nor network player are identified as being Melco products.

As far as Melco are concerned their direct connection from Melco to network player is unique to them ( it isn’t as we seem to have found that Innuos offer a similar feature). You will see this from this other diagram from Melco showing the preferred connection from the Melco NAS/player device and streamer. The S100 is substituted for the generic switch shown in the right hand diagram

1511932259
15119322591363×594 73.2 KB

It seems that the S100 switch purifies further than the existing purification :thinking:.

However given some time maybe later this coming week I may give your suggestion a try.

22

I tend to agree with this . It is certainly never a substitute for a proper backup. Or, as I believe, three of them ( two at your location and one off site). It only functions for continuous provision of service if a drive fails which may be essential in a business context but not so much for home use where service loss is an irritation rather than something that may have huge economic consequences. Of course you are subsequently going to have to replace the failed disc anyway and then reinstall the full music library ( and everything else stored across the drives).

Actually it is interesting that my particular Melco model has a prebuilt Raid configuration across its two drives. However RAID 0 ( which is effectively no RAID in terms of recovery) clearly sounds best ( as noted in at least one review and also given to me as a tip by a Melco engineer).

1 Like
23

Yep, that’s the gist I’ve gotten while reading quite a bit about the S100. A few articles, like the one below, have quotes from Melco leaders who basically say, “We made a big improvement with the server’s player port, but there were still issues. The S100 is the next effort to improve that even further.”

At the same time, I’ve seen comments from users that the S100 may “overpurify” the signal and the softening or smoothing effect isn’t necessarily to their liking. So, as usual, your ears will tell you what you need to know. And even though they’re only your ears, many of us will still be curious what they have to say. :slight_smile:

24

Well I haven’t experienced any softening effects. Quite the opposite. More a case of increased cohesion of the musical phrasing, detail and way more vibrancy. BTW I have just replaced the supplied switching PS with an SBooster linear one. Oh Boy!!

25

Wow, this is a much stronger reaction than I expected. I was actually expecting it to be more of a toss up. I don’t see a sarcasm switch, so I’ll assume it’s sincere. I’m thrilled you’ve apparently unlocked your next level. Please do keep us informed with your discoveries and evolving impressions after some time. I’m sure many will be interested to know if the S100 (or even other devices with similar goals) is a must-have item in a system as resolving as yours.

26

(I promise, I tried to stay away from this thread :grin:)

Before anyone decides to spend £1,999 (!) on an Ethernet switch, I seriously urge proper due diligence.

Here’s a starting point:

The Melco S100 is built on the back of a single Broadcom BCM53342 SoC, ancillary support circuits, BothHands GH5210-LF quad Ethernet port line-transformers, a power supply, and a pretty chassis.

Melco-S100_04
Melco-S100_041333×1000 263 KB

(Image credit: Lowbeats)

From a data integrity standpoint - in both bit/packet accuracy, and timing - there’s absolutely nothing that the S100 does that can’t be had for under $200 from elsewhere. In fact, the D-Link DGS-1210-08P uses the exact same chip architecture and offers the exact same port configuration for £95.99 :slight_smile:

If you’re worried about data integrity on your home LAN, here’s a quick way to verify that you have no issues; using Roon when it shows “Signal Path: Lossless” that means complete bit-perfect stream end-to-end from your Server right into your dCS component - Roon’s RAAT embeds data integrity validation to ensure bit transparency end-to-end (regardless of any Ethernet cables or Switches/Routers in between!).

If noise or grounding or weak/damaged digital bit voltage levels are causing bits/packets to be altered or misinterpreted in any way by your dCS component (thats the only way you can perceive sound quality changes in a digital streamed playback chain) then Roon/RAAT will let you know via the Signal Path function that your bitstream is not lossless (if you’re still unconvinced, I’ll explain other methods to verify data integrity in a separate post elsewhere sometime in the near future).

As to the “purify” discussion, I’m afraid that’s just pure marketing speak. In the Melco context it could possibly refer to two things;

  1. The individually isolated Ethernet ports via separate (Marvell) Ethernet chips on their N1x-series Servers - which is defeated as soon as you switch the Server into “Network mode” (necessary to stream from Qobuz/Tidal) as the two ports are then completely bridge at that point. The benefits of Melco Server’s “Direct Mode” is purely theoretical, dCS component are not bothered in the least by miscellaneous LAN traffic on its port not directly addressed to it.

  2. The discrete Ethernet port transformers (TDK pulse-transformers on the N1x Servers and the BothHands ones on the S100), but every Ethernet switch comes with balanced transformers on the Ethernet ports (often referred to as “Magnetics” - discrete or otherwise), that ensures high levels of common-mode noise rejection to ensure data integrity, it’s part of the Ethernet physical interface standard!

Ultimately, if you’re still worried about ground noise getting into your dCS components from your Ethernet switch, say, via the tiny amounts of parasitic capacitance on the Ethernet port transformers, then go full optical isolation; with something like a Mikrotik CRS309 with non-blocking 8 x 1G/10G SFP+ ports for a mere $238.60, and populate the switch with SFPs of your choice for complete optical isolation between all key components on your LAN.

No doubt Melco products are well built physically/ergonomically - as a matter of fact, I have an N1x Server in my system - but you really should know exactly what you’re paying for before you buy it. If you like the S100’s aesthetics, and can’t think of a better way to spend £1,999, then by all means knock yourself out, but don’t imagine it does anything extra special to improve data integrity.

Disclosure: I work for an Ethernet switch manufacturer.

2 Likes
27

Data integrity is not an issue, but there is a real case for using fiber optic connections because electrical noise leakage still seems to be a problem for listening. This would be most ideal with streamers like Lumin that have optical inputs. I’ve heard the issue of phase noise from clocking can still be an issue, and converting back from optical can generate noise itself.

Ultimately, what I know is that standard ethernet network connections actually sound pretty cloudy compared to a high end USB, I2S or AES setup, even with the Bartok I demoed. Something needs to be improved, and the only guidance I really have is my own ears and the experience of other people. If my router and streamer both used fiber optics and maintained low phase noise, it might be a much simpler thing, in terms of sonics.

Of course there are other considerations that have to do with system complexity, user experience, and mechanical noise.

28

Thanks Anup.

As you have used a picture from the Lowbeats review of the S100 switch perhaps folk may be interested in reading the full review:

NB: Readers may want to use a web translator.

I appreciate your disclosure that you work for an ethernet switch manufacturer. Melco is, of course, a division of Buffalo who are themselves ethernet switch manufacturers for IT applications.

29

RAID 0 is the less safe option :slight_smile: but if it sounds better perhaps makes sense to take risk.
Anyway, I’m old school and still like to have files. I tried to move step by step to Qobuz streaming but because of internet bandwidth limitation, still buying files…

Concerning the MELCO switch…perhaps to stabilize the system before. My 2 cents

30

Interesting. The Qobuz hi-res streaming strategy seems very efficient. About a year ago I was up in the area bordering Norfolk, Cambridge and Lincolnshire (UK) visiting a friend. Very rural in a village with one pub, one shop and one church. With, as is the case in so many rural areas of the UK, poor internet bandwidth .

I logged into my friend’s wifi LAN with my Samsung Galaxy 7 phone. The connection to the internet was only 7Mb/s. However my AQ Dragonlfy was indicating it was successfully playing files < 24/96 from Qobuz. That being the limit the little DAC will display for non-MQA and the connection may well have been able to support even higher resolutions for all I know.

31

By “electrical noise leakage” do you mean noise into the Streamer/DAC? Do you have any objective pointers to that happening in the real world?

Here’s the thing; while electrical phenomena like parasitic capacitance on an Ethernet port magnetics is real and measurable, no one has ever shown how that actually impacts the final D-to-A conversion in an actual Streamer, especially in terms of any mitigation techniques and/or consequential improved sound quality. Not even the strongest advocates of solving the “issue”. Switch manufacturers like Melco or Uptone like to talk about such issues, but completely avoid showing any proof of effect and/or mitigation.

Have you ever wondered why? :thinking:

If by “phase noise” you mean the commonly referenced clock jitter spectrum measure, I’m afraid you’re conflating issues - the clocking required on Ethernet Switches and their operating characteristics/issues are completely different from that of precision clocking required inside DACs for digital-to-analog conversion.

And I’m sorry but it would be impossible for anyone to be able to hear Ethernet Switch clock jitter because (a) Ethernet switches have packet buffering as an essential part of how they function, and (b) the layers of abstractions and additional buffers involved in a typical TCP stream before final conversion to analog means you will never hear any clock jitter involved on a layer2 Switch.

If you don’t believe me, I’m sure the dCS folks would be happy to chime in. Alternatively, you might believe Roon’s CTO, here’s an excellent post from 2 years ago when he posted about the RAAT protocol & Clocking; https://community.roonlabs.com/t/raat-and-clock-ownership/6915/38

On the issue of noise generated from electrical-to-optical conversion on SFPs, they’re trivially filtered out on even the most basic Ethernet Switches that support SFPs (see SFP MSA page 23, Figure 2A on how) after which the data stream is passed onto the host board where rest of the Ethernet handling circuitry are, which by the way is magnitudes noisier! :slight_smile:

1 Like
32

Pete, the reason I didn’t link to the review is that a reader would be led to believe that it’s common for standard Ethernet switches to have lots of data errors, basically compromising data integrity.

Nothing could be further from the truth, even with the cheapest of modern Ethernet switches :slightly_smiling_face:

1 Like
33

Blockquote
By “electrical noise leakage” do you mean noise into the Streamer/DAC? Do you have any objective pointers to that happening in the real world? > Blockquote

Absolutely not, nor do I have any proof of this happening in USB cables, HDMI cables, or AES cables. But in each case it is easily audible. USB allows asynchronous clocking and supposed error correction, perfect data integrity. But stock USB cables sound like garbage and it takes all kinds of source upgrades, cable upgrades, decrapifiers, and even format converters to remove all the noise and grunge. Toslink and AES both have jitter and bandwidth issues, but it’s pretty easy to hear the noise differences, especially with Toslink. I don’t see much of a difference from where we were in the early days of USB when the sound quality issues were not yet understood and engineers were (and many are) still adamant that the format was perfect because it worked for printers.

My experiences with audio over standard ethernet sounds worse than the signal chains I’ve built for USB, I2S, Toslink, and AES. Ethernet doesn’t have the same sound signature as the others, but is just as distracting when you are used to a clean digital input.

I can’t wait around for someone’s objective measurements or the theoretical consensus to catch up to my hearing.

Unfortunately consumers are always stuck between engineers who are invested in not seeing (or admiting) flaws with standard designs as they relate to audio playback, and people who build black box products that we don’t totally understand, but seem to work. But that’s just the way it is. Give me something that accomplishes the specific thing I want.

34

Just to be clear, my posts in this thread are all only in relation to Ethernet Switching/streaming. It has nothing to do with USB, AES, or any other interfaces types which have their own separate and distinct issues. I wouldn’t bunch them together.

35

Ok, but the empirical process towards issues of playback sound quality is essentially the same. I notice more similarities between the sound of ethernet and the sound signature of noisy USB than I do noise-free Toslink. This leads me to believe there is a noise source down the line, and to begin looking for a solution. It could be that streamers themselves produce that noise, but one thing I have seen is that different ethernet cables do sound different, although the ones I heard were not hugely different. That again tells me that some kind of noise contamination is happening down the line. I admit that I don’t really understand what is going on in the most technical sense, but this is probably the best methodology I can think of when building a sound system that sounds right to my ears.

1 Like
36

There’s “anecdotal evidence” (if you could call it that) from across the ‘net, but it’s far from an established fact that Ethernet cables make any difference to sound quality. The jury’s probably out 50/50 :grin:

It’s actually very easy to measure electrical/electromagnetic characteristic differences between various Ethernet cables, but when inserted into a system thats designed from the onset with layered abstraction, the physical layer/cable becomes insignificant, especially in home audio streaming setups.

Unfortunately, there’s a fixation by many on ascribing "analog interconnect like” characteristics to Ethernet cables, completely ignoring the abstraction/isolation involved. If audio streaming applications where affected by noise over Ethernet cables/interfaces as you suggest, the Internet would stop functioning :woozy_face:

37

I don’t know about anybody else’s anecdotes, but I heard what I heard. Granted the difference with this particular cable was not on par with swapping a USB cable, it was still there.

I can say that over the last few years I’ve found that a lot of things that have nothing to do with data transmission including digital cables, power cables, power conditioners, vibration feet, all sonically impacted DACs and ADCs I’ve owned. I’m not sure if any of this would be measurable, or if measured accepted as being of any significance. At this point I’m pretty comfortable believing that any input wire I attach to my DAC has the potential of bringing in EFI/RMI, ground loops etc.

In the end I’m hedging my bets, because if ethernet can’t be improved, I can always use the Innuos as a USB streamer and get a sound I am more comfortable with.

38

No doubt all wires do attract EMI/RFI (which is generally quite easily measured). The key question though is; to what extent do those interferences actually audibly affect the final analog output of a properly designed/built DAC?

In the case of dCS, the answer is fairly obvious (at least to most of us :grin:).

All things being equal, Ethernet/TCP will better USB every.single.time.

It’s taken Ethernet and TCP/IP close to 50 years to go through a “natural selection” process to emerge as the most widely used physical interface/protocol on the Planet!

Think about that for a minute before you suggest it’s so easily susceptible to “noise problems” :wink:

1 Like
39

Sticking with USB, if a simple in line USB noise filter such as an ifi audio iDefender is inserted between a PC USB output and Vivaldi USB input there is an audible and repeatable effect. Does that answer your question?

40

Pretty sure thats electrical noise carried across the USB cable/interface into the Vivaldi, not EMI/RFI.