Switch for streaming from NAS
- Thread starter ava1962
- Start date
Everything you mention also applies when streaming music from a local network such as a NAS server???
Hi Kevin
I have spent the past 4 years refining my network and made a few discoveries along the way.
For audiophiles, the network serves 2 purposes, moving data files from 1 place to another and refining those files to achieve better sound quality. With analog, the signal is generated once and must be thereafter protected against any form of noise addition and contamination in order to preserve sound quality. Each step of the process is a potential entry point for further contamination. In a digital network, the data stream is converted multiple times (ethernet, radio waves, light pulses, data packets, USB etc.) and with each conversion the data stream is regenerated, providing multiple opportunities for improving the signal and therein lies some important logic. If I replace the modem or routers’ cheap wall wart power supply with a superior, low noise, low impedance linear supply I hear a improvement in SQ. That means that to be heard, the improvement at the modem/router has to pass through the entire network and the only way for that to happen across all the multiple steps is for the improved input signal to result in a better output at every stage, so by definition, the network must operate on a ’Better input = better output’ logic, ~BUT that same logic has an equally valid inverse….‘worse input = worse output’. With its improved LPS, the modem/router produces a low noise output but if I take that output into let’s say a bridge with a high noise SMPS, I loose everything I gained at the modem, but if I give the bridge an even lower noise/impedance LPS, the improved modem output is once again improved by the bridge. If I then take the bridge output into a switch, the stream quality will again improve IF the switch has an even better power supply. I could also take the opportunity to further improve the data stream if the clock in the switch has a superior specification compared to the upstream components. So, if I build my network from components whose specifications get better at every successive stage, the initial modem power supply improvement will improve the input to the bridge, resulting in an improved output and therefore input to the switch and so on. In other words, the modem improvement will be compounded as it travels through the network.
So my fist discovery, a network should comprise a series of improving specifications
Next, where can SQ improvements be generated? Which specs are important?
Power supply noise, stability, impedance
DC cables impedance, screening
Network cables
Oscillator/clock accuracy and stability
Vibration/resonance control….internal grounding, external isolation
Network traffic management…ideally only audio related traffic
Galvanic isolation
Mains power to all components (isolation, noise, cables, connectors)
In order to fully implement the above concepts I use the same brand of power supplies throughout to cut down on noise harmonics that could result from using PSs from different suppliers. I also use the same brand of cables throughout so I can gradually improve the cables along the chain in a more precise and controlled way
My cable screens are all star earthed to avoid unintended continuity and resultant interferring current flows between components
I have been involved with listening to and upgrading various hi-fi systems for over 50 years and I have yet to mine a richer vein of SQ goodness than the one I discovered in optimizing my network for local and remote streaming. The improvements I get following the above concept are predictable and entirely scalable and so far I have encountered no law of diminishing returns In that as the improvements become more expensive the jump in SQ also increases.
Last edited:
Thank you for your question. Yes, in some cases this can be even more significant depending on the quality of your music files. Noise entering the system will affect both streaming services and local file playback.
Dear all, my question might sound dumb but I am going to ask anyway...
When I play files - Melco NAS to C1.1 via Ethernet, since one of the main reasons I went with Melco was their NAS have 2 Ethernet ports, one connects directly to the DAC (C1.1) and the other to a WiFi router for controling playback only, do I still need to consider about switches (such as SOtM)? I don't stream online music at all.
When I play files - Melco NAS to C1.1 via Ethernet, since one of the main reasons I went with Melco was their NAS have 2 Ethernet ports, one connects directly to the DAC (C1.1) and the other to a WiFi router for controling playback only, do I still need to consider about switches (such as SOtM)? I don't stream online music at all.
@zbub I can not answer your queation but let me tell you about my experience.
I have probably one of the best switches on the market a Melco S100 modified by Pink Faun powered by two Farad Super3. My NAS is the first generation of Melco N1Z HDD. (The new ones are much better). My Streamer/DAC is the Naim ND555.
In my case when playing music from the N1Z it sounds better (cleaner/clearer) with the switch between the N1Z and the ND555 than having the switch prior the N1Z. So in my case yes. A switch improves the sound. Prior the S100 I had a Etherregen with a Super3. I also prefered that switch between.
In your case I think it depends on which Melco NAS you have. My best guess is that if you have the top N10/2 you will have less to none improvement. But just a guess.
And a switch needs a good powersupply and a good powercord. That's my experience. And not to forget good Ethernetcables. Never ending story...
Have you the possibilty to try a SOtM or someting similar?
Gunnar
I have probably one of the best switches on the market a Melco S100 modified by Pink Faun powered by two Farad Super3. My NAS is the first generation of Melco N1Z HDD. (The new ones are much better). My Streamer/DAC is the Naim ND555.
In my case when playing music from the N1Z it sounds better (cleaner/clearer) with the switch between the N1Z and the ND555 than having the switch prior the N1Z. So in my case yes. A switch improves the sound. Prior the S100 I had a Etherregen with a Super3. I also prefered that switch between.
In your case I think it depends on which Melco NAS you have. My best guess is that if you have the top N10/2 you will have less to none improvement. But just a guess.
And a switch needs a good powersupply and a good powercord. That's my experience. And not to forget good Ethernetcables. Never ending story...
Have you the possibilty to try a SOtM or someting similar?
Gunnar
Hi Gunnar thank you for the message. I have the current N1ZH (with 5TB HDD) which replaced the first generation N1A. I haven't tried any switches because I thought there wouldn't be a need for it if I have a direct Ethernet connection to the DAC.
From your descriptions it is beneficial to have a high quality switch between N1Z and Naim over direct connection with nothing in between.
Hi Gunnar,
In my experience, a network operates on the basis of ~”better in = better out” ….its how an improvement made at the router can traverse the network and he heard as an improvement to the music. So on the logic of better in = better out you should hear an even bigger improvement with the better NAS
Interesting. If I understand you right. A better NAS would make the music even better. Make sense.
And what about “bigger or smaller“ difference if the switch is identical and the only change would be a better NAS. My guess is that the difference would be smaller with a better NAS than with one not as good. Agree? Or???
The argument of better in = better out seems logical and tempting. But when thinking about stereo reproduction what means "better" in terms of digital storage and transmission? In terms of digital data we know well what it means - faster, more reliable, sometimes just smaller. But are we addressing when we refer to a better NAS?
"Better NAS". I thought of for example Melco. They have 4 or 5 different models. From N100 up to N10/2. For me the better would be N10/2. You see?
Hi microstrip,
what does ‘better’ mean in terms of digital storage and transmission?.
The quality of the input and output (transmission) cables
The quality of the power supplies used (noise, ripple, impedance) to create data streams
The degree and effectiveness of internal and external vibration control
The accuracy of any timing clocks when creating an output in any format
The amount of noise and emi created when retrieving and transmitting a file in any format
And in terms of the actual data stream:
The bit-level voltage profile and degree of perfection across the network
The amount of jitter and phase noise inherent in the generated data stream in any format
The amount of EM noise included at all stages of the network
The losses, distortions and noise added by the data stream transmission cables
The sonic signature of actual network devices like routers, bridges, switches etc.
Lets say you place a new switch anywhere in your network. If the switch has a better PS, noise and timing spec than preceeeding stages, and if all subsequent stages have a better spec you get the maximum from the switch. If the switch has worse specs than the preceding components, SQ will go down, logically.
The point is, its not only the fact that the bits are ’perfect’ that matters. Also the inherent ‘structure’ or ‘fabric’ of the network stream in whatever form has a major influence on how the eventual music is processed and presented by the DAC. At each stage or step of the network, the better the input, the better its output, BUT that only works to our advantage when each successive stage or step has the same or preferably better specification (noise, timing, vibration mitigation, transmission cabling) than all previous stages. For example if you create a data stream based on a 10ppb clock and send that stream to a device with a 100ppm clock, the data stream gets worse, not better. The reverse, if you transmit a 100ppm stream then process it in a 10ppb device, you uplift the quality of the stream, so from a sonic point of view, a network that comprises a series of improvements will logically improve the quality of the stream, in whatever form it is in. And the better the stream at any stage, the better the output. Not only that! In a properly organised network, improved outputs from say the modem are improved at every subsequent step….in other words that single improvement compounds as it travels along the network.
Take this whole concept to the extreme and what you get if the rest of your system DAC, Amps and speakers are up to it, is sound quality that truly sounds like 4 dimensional musicians playing their instruments in a natural or studio-created venue. For good recordings, you get what genuinely feels like a performance rather than playing a recording. You’ll know this when you spontaneously applaud at the end of piece or feel tears pricking your eyes at the sheer beauty of the music.
The point of all this is; the data on the network doesn’t care a damn about all this nonsense and music will play ‘perfectly well’ as long as the network functions as designed. The BIG difference comes when you judge the network in terms of the quality of the sounds it produces. When SQ is the criterion, ‘bit perfect’ is only one of several other criteria that have a major impact.
Last edited:
You just highlighting my question. I went in the Melco page and read about their switches. What separates them according to Melco: exactly the same aspects that audiophiles love in analog electronics, and have no formal relevance in digital data. For example , "external power supplies" , "Complex and rigid vibration-reducing internal chassis". And yes, you get more of it as price increases.
Buying them is an act of faith. BTW, I am not questioning the quality of the product - I have never listened to such units.
Reading from your post we would conclude that what matters is creating a perfect, text-book digital signal with perfect timing. IMO this would be easily measurable and we could easily rank switches according to measurements. I think that it is a lot more than that. I am not an expert in streaming data or networks, but I think that a lot has to do with how the data is repacked and the software of the switch. At current state of affairs all we can do is listening to their sound effect in particular systems.
Yes, what we need is a perfect text-book digital signal with perfect timing. That way, whatever processes are conducted on that signal, the noise is minimal, the jitter is minimal and the number of errors are reduced. The fact that an improvement at the beginning of a LAN, for example the modem, can be heard in the resulting music means that the improvement has to travel all the way through the network and the only logical way that can happen is if better in = better out. So the better the data file at all stages of the network, the better each part of the network process runs and the better the outcome is in terms of SQ. But I also think there’s a lot more than that. Things like the correct buffer sizes, the SW running on various components, the amount of network traffic, the number of times processes are interrupted, the number of errors that need correction, the way the data stream is handled, especially how the SW at the various stages works together, for example how the results of SW running on the router interacts with SW running on the server and DAC
In a network switch for example, you can optimize quite a bit that impacts SQ, for example:
Internal and external vibration control
Noise generated by the switch chip
AC power supply quality
DC power supply quality, noise and impedance
DC wiring and screening
Input and output (transmission) cable screening
Internal generation of EMI
External EMI ingress
Quality of RJ45 connectors and isolation from vibration
Quality of output isolation transformers
Accurate clocking of incoming and outgoing ethernet streams
Amount of traffic passing through the switch
Clock power supply noise
Most of these effects can be measured, but the problem is, those measurements will change on every single network. We’ve already noted above that the output of a device is totally dependent on its input so most improvements can only be measured as system differentials i.e before and after but the point is, what you measure on one system may not be the same as another. Manufacturers like Taiko or Innuos are making these measurements all the time and building hardware and writing SW around their findings. But those findings and the resulting improvements are valuable IP, so you’re not going to find them in the public domain. Take Taiko for example. They are about to release a new network card, modem and switch. The main thing about the switch is that its software is specifically designed to complement the Extreme SW, but Taiko are unlikely to be posting measurements and details on what amounts to their competitive advantage.
I would summarize by saying that the network has the ability to massively affect the quality of the final music we hear. We understand the basis for and can influence some of those variables….certain manufacturers and our DIY gurus understand a lot more, so we have the capability to improve SQ massively ourselves and we can buy products commercially that will give us yet more gains.
In digital there are 2 levels of quality….the bit integrity (are all bits present and correct) and the bit quality (how close to perfect are bit timing, voltage profiles, noise content, error rate etc.) Every single component in a network has a sonic identity and sonic influences. The more of those we understand, the more we can modify our networks and improve SQ
Similar threads
 | Steve Williams Site Founder | Site Owner | Administrator |  | Ron Resnick Site Co-Owner | Administrator |  | Julian (The Fixer) Website Build | Marketing Managersing |