I’ve been involved in digital music streaming for around eight or nine years now, but have only truly taken a really deep dive in the last few years. Recently, I’ve had the good fortune to evaluate several exceptionally good turnkey streaming setups, including the Euphony Summus Server/endpoint system that is the cornerstone of my home system. But my initial involvement with digital audio was more as a tinkerer, and there’s a whole cottage industry of home-brew designs (mostly based on Intel NUC or Raspberry PI configurations) out there. At almost ten years into this voyage, my personal experience tells me that the DIY route is not for the faint-of-heart. Dealing with the hardware configurations isn’t horrific, it’s mainly the buggy, mind-numbingly stressful DLNA or UPnP open source software implementations needed to create a working music server. Trust me, I’ve been there, done that!
And along the way, some of the most useful hacks and hardware ancillaries intended to elevate the performance of digital music streaming have seemed to elude me, for whatever reason. One of them involves my current subject, which is employing a Gigabit Ethernet media converter setup in combination with my digital streaming system. I’d been reading about the ease of incorporating one for probably nearly four years now, but for whatever reason, my interest had never been taken to the next level, until just last week, when I caved and spent the $90 necessary to make it happen, at my local Micro Center. And although my previous goal for setting up an Ethernet to fiber conversion scheme was strictly to improve sound quality, that’s recently been augmented by another, previously unknown (to me, at least) rationale for this setup. That is, to implement complete protection of your Ethernet-connected equipment from network-related lightning damage, which can be an extremely vexing problem in any audio setup.
Home Ethernet equipment and cabling can easily introduce unacceptable levels of digital noise into your playback environment. For me, that was initially the big selling point of ethernet to fiber conversion – lowering the noise floor. At the point several years ago when my digital streaming experience had begun to step up to the next level, I still was pretty much oblivious to the need for outboard improvements. Part of my lack of enthusiasm for optical conversion had to do with my perception of optical in general, which was mainly focused around the popular Toslink optical connection. Yes, Toslink is optical, but it has limited bandwidth, which limits its usefulness with uncompressed digital audio signals, and it’s widely considered, by those in the know, to be the very last choice for digital audio connections.
The Sonore Signature Optical Rendu…and the Fidelizer EtherStream Network Switch
I had the pleasure to review the Sonore Signature Optical Rendu SE a couple of years ago, a device that had a built-in multi-mode optical digital input. The unit came supplied with a Gigabit Ethernet media converter, which took the incoming copper cable Ethernet signal and converted it to a multi-mode optical signal. That plugged directly from the media converter, via a multi-mode optical cable, into the Signature Optical Rendu. Sonore touted this connection as being audibly superior, and I couldn’t argue with that based on my evaluation – the resulting sound had a level of transparency and clarity that simply blew me away with its goodness. At the time, that optical connection was the single biggest improvement in digital streaming I’d yet experienced – the Sonore was a digital streaming dream system, without a doubt. I feel pretty certain that systems like the Sonore, with its built-in optical capabilities, are the streamers of the not-too-distant future. But for whatever reason, units with built-in optical ports are few and far between on the current market. By having the optical capability built into the unit, you’re bypassing an entire conversion stage, which ultimately should be a good thing.
Another aspect of optimizing the Sonore setup was that Adrian Lebena (Sonore’s head honcho) insisted that the media converter should be powered by a linear power supply rather than a wall wart. I happened to have an Uptone Audio UltraCap LPS-1 on hand that matched the voltage requirement for their supplied media converter, and it worked perfectly. My experiences with linear power supplies have definitely shown me that using one can significantly enhance the performance of the attached equipment. The LPS-1 in combination with Sonore’s optical module proved to me that Adrian was correct when he insisted that noise from a substandard power supply could degrade the sound of the Signature Optical Rendu. Also, in my recent online explorations of fiber media conversion, where dual converters are employed, I’ve found numerous references to the importance of providing a good power supply to the media converter that’s nearest to your audio stack. The media converter at the beginning of the run is ahead of the fiber cable implementation, so no noise from it can move to the second converter – the optical cable is impervious to electrical noise of any kind.
Despite this positive experience, and for whatever reason, my brain didn’t completely latch onto the whole Ethernet-to-optical conversion concept. Days after the Sonore unit left my system and returned to the manufacturer, I received a new EtherStream network switch from Fidelizer. This takes a top-of-the-line Cisco switch and replaces many of the interior circuit board parts with an upgraded selection of capacitors and resistors, and an improved power supply. Yeah, I know, a switch is a switch, right? That’s not what my ears told me – the EtherStream definitely lifted my streaming experience to a level that came within an eyelash of the goodness of the optical setup with the Signature Optical Rendu. And, when used in combination with my Ethernet-connected Sonore UltraRendu, offered surprisingly good sound. The EtherStream stayed in my system for over a year, and I didn’t spend too much time bemoaning the loss of the Sonore Signature Optical unit. Especially after the Euphony Summus system showed up!
The $400 EtherStream ended up getting returned to the manufacturer about the same time as I experienced my system failure in August, and I replaced it with a $40 managed Ethernet switch. I didn’t think too much at the time about the impact of the new switch on my sound quality, as my system was essentially limping along for months. But when I got the system back up and running again in mid-November, I immediately felt it sounded…less perfect than it did with the EtherStream in place. That’s the point when I really started thinking hard about possibly finally going the fiber media conversion route.
Gigabit Ethernet Media Converters – the Basic Concept
The concept of incorporating a Gigabit media converter revolves around taking the noise-prone copper-wired Ethernet connection and converting it to an optical signal. One media converter unit is required to convert the incoming signal from Ethernet to optical. An optical cable is then connected to the media converter, which in turn connects to a second media converter. The network signal then gets reconverted from optical back to Ethernet. The advantage to this setup is that once the Ethernet signal has been converted to optical, this completely eliminates any electrical noise from the signal path. At the endpoint of the setup, a small length of network cable then connects to the Ethernet port of the second media converter, and in turn to your streaming equipment. The network cable should be as short as possible, so that it won’t act as an antenna for any RF signals that might get attracted by a longer cable. A number of configurations are possible in a Gigabit media conversion setup, and those configurations run a broad range of prices. Mine is at the low end of the spectrum, but you could easily spend several hundred dollars (or more) to accomplish the same thing.
The entry-level setup requires a pair of Gigabit media converters, a pair of SFP transceivers, and a fiber optic cable. Media converters come in iterations from 10Mbps, 100Mbps, 1G, 2.5 G, or 10G, all depending on your available bandwidth and system needs. That said, most high end streaming equipment requires Gigabit capability, so a 1G media converter is the baseline minimum for high-resolution digital file streaming. The converter units come in a surprisingly vast array of configurations; the entry level units feature an ethernet port, an SFP (small form factor pluggable) input cage (more about that later), and a power input port. That said, there are media converters that have the optical transceiver built in, but they’re understandably more expensive. And you have to choose from a couple of optical form factors; your choices include single-mode and multi-mode optical. The single-mode connections have a single-tipped connector, and their cables can carry a robust optical signal for hundreds of feet with no degradation. The multi-mode connections have dual-tipped connectors, but have less robust bandwidth that’s also limited to shorter distances. While greater bandwidth and cable length is important for large-scale optical networks, the needs of most audio implementations of fiber media conversion are much smaller in scale, making the multi-mode transceivers and cabling perfect for audio applications.
The SFP transceiver is a module that plugs into a media converter and facilitates the conversion from Ethernet to fiber optic. SFP transceivers are available in both single- and multi-mode configurations, and have varying speed capabilities. So, if you decide to go with a media converter that only includes an SFP cage input (they’re more common), you’ll need to choose an SFP transceiver module that matches the capabilities of the media converter. As with media converters, pricing for SFP units is all over the place. You’ll need a pair of them, but you can easily pay as much for a single transceiver as I paid for my entire setup!
The only remaining necessary item is a good optical digital cable that matches either the single- or multi-mode setup of your system. The good news is that optical cables tend to all be 10 GbE-capable, so should you decide to later upgrade your system speed (10 GbE and other standards are becoming available), your cabling will already be compatible with your new system. And the length of the cable isn’t important here – it can be 1 meter, or 10 meters, whatever length is necessary to make the setup work in your system. However, it’s definitely preferable to use a longer optical cable and a very short Ethernet cable on the end with the audio equipment.
My Gigabit Media Converter Setup
I obtained everything for this project at my local Micro Center. The pair of TP Link media converters I chose retail for $25 each, and are very similar in design, capabilities, and price to many I’ve seen in online discussions of implementing this process. They’re definitely baseline units, with an SFP cage input that requires the purchase of an SFP transceiver module. The pair of 1 Gigabit, multi-mode SFP transceivers that I chose are manufactured by Ubiquiti, and despite the low $22 price for the pair, they get very high marks for their build quality and functionality. And for $18, the 3-meter Micro Connectors multi-mode optical cable I selected for connecting the SFP transceivers in the two media converters worked perfectly for my particular setup. I decided to go with multi-mode for everything, mainly because single-mode would definitely have been overkill in my system, and the cost savings of multi-mode (about half the cost) versus single-mode was substantial.
Having set up the Sonore Signature Optical Rendu at the time of that review, I basically knew the drill, but with two media converters versus the one needed for the Sonore setup, the process was a bit more involved. It only took about ten minutes tops, though, before I had music up and running again. There were no drivers or anything else software-related that needed any attention during the setup. The voltage requirements of the TP Link media converters were slightly higher (9V) than my UpTone LPS-1 (7V max) was listed as being comfortable with. I reached out to Alex Crespi at UpTone Audio, and he basically told me that I should just try it, it probably would be fine, and if it didn’t have enough juice to power the unit, no harm would be done to either unit. I remembered afterwards that the Keces P8 Linear Power Supply I currently have in the stack had a configurable second output tap, and I was able to set that to 9 volts for the media converter nearest my Euphony Audio stack. I used the Keces P8 to power that unit, and used the supplied wall wart to power the unit on the network side. Everything worked perfectly; afterwards, I looked at the Sonore website, and their media converters (especially the ones that look virtually identical to the TP Link) show acceptable power ranges of 6 – 9 volts, and they were initially designed to work well with UpTone LPS’s. So when the Keces unit needs to go back to the manufacturer, I’ll just sub the UpTone LPS-1 – it should work just fine.
The first media converter unit sits on a table top near all my network equipment, in the room behind my listening room. It’s currently directly connected to my cable modem/router, and I’ve completely bypassed my network switch. The 3-meter multi-mode optical cable runs through an opening in my keystone jack array and into the listening room, where it connects with the second unit that sits beside the Euphony Audio streaming stack. The Keces LPS sits on another rack adjacent to everything else in the streaming stack. It’s convenient, and has thus far has worked perfectly. I plan on playing about with the connection to my cable modem/router, and will take a listen to how everything sounds with the network switch in the loop to see if it has any impact on the sound.
So how does it sound? I’d say that the status quo has definitely been restored. When I replaced the EtherStream switch in my system with the stock switch, I wasn’t too happy with my system’s sound. I then bypassed the switch, using a direct connection from my cable modem/router to the Euphony stack. I wasn’t exactly ecstatic with that, either. But with the fiber media conversion setup now in place, I actually feel the sound quality exceeds any previous incarnation of my digital streaming system. Notice I said feel – this is not just me working all this out in my brain, this is an emotional response to more accurately portrayed music. The noise floor is significantly lower, and there’s zero possibility of any digital hash or noise interfering with the music – the music playing sounds so much more…musical.
As I mentioned earlier, a key advantage of having an optical media converter in the chain is the added benefit of totally isolating your expensive Ethernet-connected audio (and other) equipment from the effects of a lightning strike. The optical cable is non-metallic, and can’t carry a deadly electrical surge, so damage due to lightning via Ethernet will no longer be a consideration. This a welcome level of protection.
Yeah, I know – fire up the flamethrowers! Ethernet is just Ethernet – as with everything digital, it’s just transmitting ones and zeros, and the differences between Ethernet and optical can’t be that significant. Whatever. Is a switch just a switch? Not necessarily. Is a cable just a cable? Not necessarily. It’s all system-dependent, but I firmly believe that inserting an optical connection into a copper-wired network chain will make a significant improvement in any digital streaming system’s musicality. Period. Best $100 I ever spent!
Header image: the Fidelizer EtherStream network switch is a superb – if a bit pricey – performer.
All images courtesy of Sonore, Fidelizer, UpTone Audio, Micro Center and the author.
60 comments on “Improving Digital Streaming with a Gigabit Ethernet Media Converter System”
Leave a Reply
You must be logged in to post a comment.
Good morning Mr. Gibbs!
This mite be a little off topic, but perhaps you mite know something about this.
Just in time for Christmas, I aquired a brand new iPhone 13.
Its storage capacity, is 128 gigs.
I have a sizable collection of DSD files.
I want to put some of those files on my phone.
Just so that, I can listen to them on the go.
I thought about Foobar2000, but it’s an offal lot of work to get Foobar2000 to play DSD files on a smart device.
And on top of all this, I have to rely upon Voiceover.
The reason for this, is because, I’m a blind man.
I don’t have any real usable eye sight to speak of.
Inn spite of the fact that, I can only see lights and shadows.
But I can’t tell collars, and I can’t read printed text either.
In less, I scan it in to a system that can read it out loud to me.
But getting back to my iPhone 13, do you know of any app that can play DSD files without too much of a fuss?
Thanks in advance!
As always, my compliments to your audio pursuits. Your persistence is a shining example for all of us.
Unfortunately, even though I work with both Macs and PCs on an almost daily basis at my day job, my cellphone use has always been limited to Android devices. And with the exception of the Linux-based Euphony equipment in my current stereo setup, I’m strictly PC and Android at home — no idevices of any kind. So, unfortunately, I really can’t answer your question regarding an app for iPhone that will play DSD files — especially without much of a fuss. I sorry I can’t be of any help with this.
Thanks for reading!
How does 2 Media Converters compare to the One Sonore Signature Rendu SE? Big price difference.
The Sonore Signature Rendu SE with its built-in SFP cage — in combination with the necessary first media converter required by their system — really is the equivalent of using two media converters in a system. You’re still using the first media converter in the chain to convert the signal from copper-wired Ethernet to Optical, then the Rendu’s second, built-in media converter reconverts the signal from Optical back to a copper-wired connection. The transfer function that occurs is for all practical purposes identical as the one that occurs in my setup with dual media converters. The Sonore system is definitely more elegantly conceived, but that doesn’t necessarily mean that its sound is significantly better than my dual-unit setup.
That said, in all honesty, it’s been two years now since the Sonore was here for evaluation, so I can’t really give you a head-to-head objective comparison. At the time the Sonore Signature Optical Rendu SE was here, it was head and shoulders above my standard Sonore UltraRendu. But at the point when Sonore supplied the Signature Rendu SE, they also supplied an off-the-shelf media converter to use for the first unit in the chain. They have since started building their own media converters, which are said to be built with an improved selection of parts, and they claim are much better than off-the-shelf units. I have no reason to doubt this, especially based on my experience with the Fidelizer network switch, which was most definitely an improvement over any off-the-shelf switch I’ve ever used. I would venture to guess that Sonore is probably (much like Fidelizer) taking a very good off-the-shelf media converter, upgrading the parts selection (especially in the power supply), and placing it in their own custom-designed outer cases.
However, in all the time that’s passed since I had the Sonore SE here, there’s been a considerable amount of discussion on most of the computer audio boards regarding the use of Ethernet to Optical conversion. As I noted in my article, most folks who’ve experimented with this seem to agree that utilizing a Linear Power Supply — especially with the second media converter in the chain that’s closest to your streamer — provides a significant benefit in terms of overall sound quality. The Sonore SE had its own built-in LPS, and I used the UpTone audio LPS with their supplied off-the-shelf media converter, and the sound was superb. In my current setup, where my first media converter uses the stock wall-wart and the second media converter uses an LPS — I’d be willing to bet that I probably would hear ZERO difference in sound quality between the two.
My use of media converters is as entry-level as it gets — the total cost of the setup was under $100, but the improvement in my digital streaming sound quality is significant. As I noted, you can easily spend $100 for a single optical SFP transceiver — had I gone that route, I could probably have spent closer to $500 for the setup. I really believe that by incorporating an LPS with the second media converter, it will operate at peak efficiency and give you 99.9% of what you’d get with a more expensive setup. I’m currently using a very pricey LPS, but I’ve read countless online entries where audiophiles have employed very reasonably priced LPS’s from Jameco that can be gotten from Amazon for less than $20, and had exceptional results when compared to the cheap wall-wart power supplies of the stock units.
Thanks for reading!
others go ballistic, i went optical some 18 months ago – de rigueur, as the french say, because i went 10G between my fiber-to-the-home wall outlet and my fiber-to-the-desk workstation which is two rooms away. it’s basically a data network, but as it so happens, the wall outlet is right next to my audio system, and the 10G NAS with all my digital music is right next to the work station. i bought two reasonably priced QNAP 408-4c switches with 4 10G SFP ports and 8 1G RJ45. two separate 15 meter multi-mode runs connect the two switches, the MacPro work station was updated to 10G optical with a PCIe card, the Mac Mini i use as audio computer was made 10G compatible with a short SFP>Thunderbolt3 adapter. the Synology NAS already had a 10G RJ45 port which i connected to a copper SFP on the switch. the two redundant fiber runs between the switches are kept separate; one is reserved for audio traffic, the other carries all the other data traffic between the office and the wall outlet.
basically, this means that anything audio (Mac Mini, NAS, Internet for streaming) is galvanically separated from the rest. since the two rooms are on separate electrical circuits, there is no connection between them that could induce ground currents and hum. other than you, my SFP>RJ45 converters are either inside the computer (PCIe card) or bus powered (Thunderbolt).
i experienced similar sonic results before and after. i think it’s less an issue of the Ethernet cable being an antenna guiding RF into your streamers, but instead the complete galvanic separation of the attached gear. no matter what streamer or computer you use, the Ethernet ports carry some of the power supply crud from one box to the next where it pollutes the signal ground. you could achieve the same with unshielded twisted pair as Ethernet cable. (Ethernet transformer couples the data on both ends, but the shield must connect the two chassis in order to act as shield.)
you could try the following experiment to see how much of a difference this galvanic isolation makes: disconnect your cable modem from the TP and plug in a PC or Mac directly via RJ45 Ethernet. send music over the fiber connection to your streamer. listen.
then take the same PC or Mac to your music room and connect it directly via RJ45 to the streamer. listen again. (my bet is that the PC/Mac connected via two media converters and a fiber link to the streamer will sound cleaner, less gritty, less USB-like than with the same computer connected directly via a short piece of Ethernet cable. all you have changed (basically) is to completely isolate the computer’s power supply and high frequency switching noise from your DAC.
> They’re definitely baseline units, with an SFP cage input that requires the purchase of an SFP transceiver module.
having a cage only is not a tell tale sign that it is baseline – depending on the fiber (or copper) connection, you need different SFP, that’s why all fiber switches or converters come as cage-only. (some bundle the converter with an SFP of choice.)
these SFP work quite hard when converting 10G data from light to current. they get hot. and they radiate switching noise. smaller dongle style SFP > USB C, SFP>Thunderbolt, or SFP>RJ45 can get so hot they start to throttle or switch off in a cowardish act of self preservation. i don’t know how it is with 1G, it may be less of an issue. either way, make sure they are well ventilated. i would route other – especially analog – signal cable at a safe distance. you don’t want to go all the way to optical, just to go ballistic because the hash is back in your sound, but this time directly analog…
Wow, you’ve got an impressive setup! When I built my new home five years ago, I really thought I was going cutting-edge by wiring the whole house with Cat 6a cabling — the guys at the local electronic supply thought I was nuts going hard-wired versus the convenience of wireless.
The SFP cages in the two 1G TP Link media converters I use get barely warm to the touch in any stage of operation — I guess with the very limited amount of throughput taking place here, they’re not being overworked. And I totally agree about the galvanic isolation; prior to going with an I2S digital connection this year, I was using strictly USB with a JCAT USB Galvanic isolation unit. As good as that sounded, optical with the I2S is miles beyond.
Thanks for all the input and information, and thanks for reading!
I suggest to all the commenters in this mini thread the book The Extended Mind: The Power of Thinking Outside the Brain.
I have about 15 years of professional experience with computer networking. “Digital noise”, the fundamental reason for your network upgrades, doesn’t exist in network transmission of data using TCP/IP. The basics of streaming media over TCP/IP are pretty straightforward: if the received signal (packet) is not an identical signal to what was sent, then a command is issued to resend that packet. This isn’t analog, and in digital transmission there’s no range between the 1s and the 0s, no distortion which would lead to 1.5 . If you’re using UDP for network transport (probably used on synchronized like Sonos, Google Home, Apple TV, ect…), then there’s a chance you would have packet loss. This would be only sensible element to focus on. If you could spend a small amount of money on sound equipment and provide some quantitative analysis of the audio before and after being subjected to a variety of ~poor Ethernet environments, then your suggested purchases would be a whole lot more credible.
These kinds of arguments will probably continue for generations. I’m not a “network guy,” I’m an audiophile, and I’m not looking at ones and zeros in the data stream — I’m listening to what my ears tell me. And the improved level of galvanic isolation provided by the optical signal provides a greatly reduced noise floor in the audio environment — even at the entry-level of sophistication employed in my current setup.
My current stereo rig retails for around $50K, which is peanuts compared to many high-end setups out there. That said, it’s very resolving compared to many I’ve heard with price tags many multiples of my own. We’re not talking about a Sonos or an Apple TV here, and when you get involved in system enhancements with high-end audio, lowering the noise floor always leads to improved levels of clarity, transparency, improved retrieval of detail, and improved musicality — especially with digital playback equipment.
Just my two cents worth. Thanks for reading!
I’m also an electronic engineer, an audiophile, and a network engineer. I’ve been around “Ethernet” since it was a thick yellow cable with phantom taps, and Cisco gear since major release 7. My main set up is also in the order of many tens of thousands.
There are many flavours of Ethernet but I just don’t buy any of these arguments. For starters, rj45 copper connectors must already be galvanically isolated (for safety reasons). Also the faster you switch, the more noise there is. Gigabit over copper actually runs as 4 “lanes” over the 4 pairs. Each lane has a symbol clock derived from the coding. Gigabit over multimode fiber has to serialize the same data over a single lane, so therefore has to switch much faster. And 10Gb Ethernet requires even more power and fast switching electronics. Higher frequency switching equals more power and more noise. Not less. Anyway all audio that is sent over Ethernet has to be buffered and reclocked, because Ethernet is isochronous, and not fundamentally designed for real time data transfer (there are some niche standards but they aren’t widely used).
If you’re worried about noise IMHO you’d be better off looking at your system earthing and interconnect shielding. The radio spectrum and mains power are much noisier than even 10 years ago: with wifi, lte and all sorts of home automation. But buying top of the range Cisco switches to replace a point to point link? Nah…. a switch can never be faster or more accurate than a simple piece of cable or fiber. And any media converters are going to introduce even more jitter due to the difference in PHY layer coding. Plus of course their own noise.
Finally some common sense and some scientific rationale. The “but I can hear a difference!” faith-based, anti-rational, pseudoscience babble that goes on in the audiophile world makes me literally throw my hands up in despair. How one chooses to spend one’s money is none of anybody else’s business, but trying to hoodwink others into unnecessary and frankly ludicrous expense drives me potty.
Audiophiles in their insistence on the gobbledygook of trying to find and justify getting people to spend money on equipment that is superfluous always makes my brain explode I am an audio enthusiast, I am an electronics engineer but I’m also a music producer that has worked in some of this top music studios in the country and has some hit records to boot. No recording studio goes into the silliness that audio files do so if none of this is done at the source of where the music is composed and recorded what’s the sense of adding it on at the end? The next step is they’re going to come out and tell us that these special hard drives that somebody has thought up to overcharge people for better preserve the clarity and keep the base from becoming too muddy then your regular hard drives because they’re made from specially selected silver nano coated platters and diamond infused actuators and that the motors in the special hard drives don’t induce any low noise vibrations and they make a significant difference that can be heard by audio files only in the music that’s stored on them this is a religion this is not science
Sorry for all the typos I’m driving and had to voice dictate
Based on the autotuned, overprocessed, compressed-as-hell crap that passes for a “hit record” these days, it’s almost impossible to believe that anyone involved in current music production has even the remotest concept of good sound.
Thanks for reading!
Canwire, from what I read and have experienced, as Tom says its not the 1’s and 0’s, but the interference / Noise introduced to the overall environment. Like Cable causes hum, the 1080p picture is the same, but the noise is there and you hear it as ground loop hum. Someone had the good analogy, its like a giant antenna gathering noise and then a physical connection to your very sensitive equipment in your system. The difference is very noticeable.
you don’t need an antenna to pick up noise. your digital gear switches at super high rates, the SMPS in your computer, streamer or whatnot also switches at frequencies >50 kHz. power supplies in computers don’t have to be super quiet, they just have to deliver, with little ripple please. the switching noise is dumped into the (digital) ground. but at some point, analog and digital grounds are joined; maybe with a diode or an RC as “isolation” between them. the shields of ethernet cables, the return on single ended connections or bus powered stuff is connected to the chassis or ground. the noise rides on the shield (in balanced connections) or on the return.
the moment you interconnect two audio devices via ethernet, USB or RCA, the two grounds are connected. the switching noise is directly coupled from one ground plane to the next (as well as capacitively from shield to signal wire). since the signal in single ended audio is always related to ground, your music is enriched with all this noise. the noise per se is not audible; the switching frequency is way outside the audio band. but it creates intermodulation. and since the switching noise is broadband, not a single frequency, the intermodulation is all over the audio band.
balanced signal processing and transfer is /can be floating i.e. not ground related. and the noise is (theoretically, in a perfect world) nulled because one signal is of opposite polarity.
braking up the ground connection between digital devices by optical data transmission and not using ground-referenced (single ended) audio signal transmission is one way to go.
the other would be integrating everything in one chassis without any outside connection; ideally battery powered. (sounds crazy? can/has been done. having all your music inside the streamer is possible. external data would come via WLAN. class D amps are inside the chassis, too. taking a holistic approach, the designer could decide to run *everything* inside the chassis at the same frequency and with one clock: digital audio data would be upsampled to this frequeny, the class D amps would switch at this frequency, the SMPS, too. use 300 amp hour LiFePo batteries (12, 24, or 36 volts, as you please) to power the whole thing.
oh, your turntable, i forgot… put preamps and A/D converter into the turntable chassis. amplify phono signal flat (no RIAA de-emphasis) by 40-some dB, convert un-EQ’d phono signal from A to D either bitstream or PCM, transmit to one-box using either optical or wireless, reclock / upsample inside one-box, apply RIAA-EQ and de-clicking algo with DSP.
ca be done, can be made to sound very good. but cannot be sold commercially because audiophiles don’t want one-box systems. nothing to tinker…
I respect the digital engineers on this subject, but can they rule out analogue ‘noise’ communicating along the ground plane of an ethernet line?
Except there is no “ground” or “shield” on Ethernet over UTP. There might well be some capacitive coupling.
But that’s why Naim streamers have internal isolation between the streaming board (where the audio is decoded and retimed) and the DAC plus analogue audio board.
So any external additional “isolation” is simply unnecessary.
it’s either STP and the grounds are linked, or it’s UTP and there’s your antenna. pick your poison… 😉
for EMI reasons, i wouldn’t want to run UTP (unshielded twistedd pair) over longer distances than, say, 50 centimeters. as far as i know, all Ethernet ports use data transformers on both sides. i don’t know how Ethernet transformers are wound, but i don’t think they will reside in a mu metal can and have two or more electrostatic screens.
what could be done is connect the shield on one side directly and on the other side thru an RC (51 ohm+10 nF). for DC, there’s no connection, but at RF, the shield is grounded. (neutrik has developed an XLR connector with a LC shield filter built-in; the C form a circle around the entire diameter of the shield. i don’t know if RJ-45 connectors exist with a LC or RC coupled shield.
People seem to be under the impression that there needs to be a signal ground for Ethernet over twisted pair to function. There does not.
Ethernet over UTP uses differential signalling. So the tx positive wire is driven high whilst tx negative wire is driven equally far low. No ground reference is required. The pair twisting reduces any antenna effects. The coding scheme is chosen to ensure that there are enough transitions for reliable clock recovery, whilst equally there is zero DC component to the signal (hence no need for ground reference). Noise cancels out at the receiver.
See this link:
Any local digital noise in the streamer hardware (running a general purpose cpu to run a TCP stack and buffer and decode the transmitted audio) is going to be many orders of magnitude higher than anything potentially leaking in on the RJ45 connector to the local digital signal ground. Everything is buffered and retimed so any packet jitter or retransmission is taken care of.
Also as I previously wrote, Naim isolates the streaming board from the DAC board using LVDS I2S signalling (again this is Low Voltage Differential Signalling, with no common ground).
You and everyone else are free to spend your money how you want.
IMHO spending any additional effort and expense in the digital domain is pointless. I invest in high quality streaming hardware and high quality streaming services. If you are really worried about Ethernet then you could anyway plug a flash drive directly into the player….. e.g. Quobuz allows local downloads.
Enjoy the music!
i described differential signalling (and called it balanced, sorry) and said it can but must not be floating. european analog broadcast and studio equipment used to be single ended inside the chassis but transformer coupled to balanced twisted pair as soon as two chassis had to be interconnected. and interestingly, the output transformers often had a center tap that was hard grounded. in a well designed and perfectly executed system, the balance will be so good that no signal flows to ground, grounding the cente tap will define a reference for the signal level and make sure the signalis notmotor boatingup and down.e
transformer coupling became obsolete because it is much cheaper to use a THAT chip. cheaper yes, but not better in terms of CMRR. as the european broadcast industry used 600:600 ohm until the 1980s, they could have gotten away without shields, but they didn’t. even at+4dBu, the signal was cleaner if transmission was done with shielded twisted pair. that was analog, BW limited to 20 kHz because anything higher was not signal but noise.
fast forward to todays”s switching frequency. i would never use UTP for a 1G or 10G data network. does Cat6 even exist as UTP? i don’t think so. so as long as you use your 1G data network also for audio signal transmission (not neccessarily streaming, consider for example a NAS and a PC or Mac used for memory playback (audio data for entire song is loaded into RAM before playback starts). assume furthermore that your hi-res audio files are all 96/24 or 192/24, then you’re glad to have a fast network just for shovelling these files from your NAS to your playback device. also, who would use anything less than 1G for a home data network?) … you’re very likely use STP for transporting your data from one server to the next computer.
and voila. i did say that Ethernet is transformer coupled. but at these switching frequencies–how much is the CMRR of your differential receiver chips at, say, 2.4 ghz? ranging from poor or abysmal. your 50 ft of UTP will act as both, a transmitting antenna and a receiving antenna. and not all all RF interference will be nulled out as common mode; the inevitable imbalancies will treat a part of this RF interference as differential. no, the world is not perfect.
and at 2.4 ghz, even the transformer will couple interference capacitively to the primary side. so yes, with UTP you have galvanically isolated signal transmission, but the shilelds connect every single device in your network together. the cheap data switch with the wallwart, the VoIP telephone with cheap chinese PoE adapters etc. and eventually the digital audio needs to be converted to analog at some point. and as long as there is a digital signal arriving at the DAC chassis, you will be coupling some digital switching noise to the analog ground. as if you have ever designed or built a microphone preamp or a phono stage, you will know how a dirty ground can mess up your SNR.
i built my 10G fiber network for data transmission, not for audio.i used 10G because i work with HUGE image files, sometimes north of 200 GB. the audio system profits from the network being optical by the removal of many old auo files.
thanks for sharing the two files. both make excellent primers. the Naim user manual deserves highest pride. if only every highend manufacturer supplied such valuable prose, thorough design and open information.
how can you completely separate analog, digital and, say, network grounds in a chassis? unless they’re truly floating, they will have some connection, either with a cap (as in the ACME file), side-by-side diodes, a coil or something else that keeps them separate at DC but creates RF ground.
You wrote: i would never use UTP for a 1G or 10G data network. does Cat6 even exist as UTP? i don’t think so
Of course CAT6a UTP exists. And it’s certified for 10Gb Ethernet. A simple Google search will confirm that.
Also if you Google for RJ45 connectors with in built magnetics, you’ll see that they also employ a centre tapped transformer, with the tap grounded.
10Gb is all well and good for faster data transfer but it generates even more noise than 1Gb switching Remember you’ve also got a laser pointing outbound from your streamer to the NAS that is also switching at extremely high power and frequency.
Copper interfaces run into around 100 ohms impedance at around 2V peak to peak. That’s around 40mW of power.
A 10gb laser runs at around 1watt of power (depending on the exact standard/fiber type led/laser diode)
I’d be more worried about Near End XTalk than any noise on the wire.
Also for short fiber runs you need to be aware of the optical link budget to ensure the receiver doesn’t get saturated. Again that depends on the exact version of optical Wthernet and the transceivers you’re using.
Understood. Hi-Fidelity reproduction can be a minefield of misunderstood facts, ideas, and feelings. A lot of money is spent every day on these things, and audiophiles swear that they can hear a reproducible difference. I guess this argument will go on.
I appreciate all the responses, whether supportive or skeptical. I’ve made it pretty clear: I’ve never claimed to be an expert on network systems or engineering. But I’ve had a significant level of involvement with audio industry pros who are either/or both, and they all seem to agree that an optical network is superior to wired copper — period — but especially for digital audio streaming. And that has been my experience.
My current home was built with a dedicated space for listening with two dedicated 10 gauge, 20-amp circuits for AC power and AudioQuest Edison AC outlets for the equipment stack. I also use AudioQuest Niagara 1200 and 3000 units for AC power conditioning — I can honestly say its the quietest dedicated listening room I’ve ever been in. Many of the pieces in my equipment stack employ linear power supplies and upgraded AC cords, etc. The already good level of quiet and exceptionally black backgrounds I normally experience has been enhanced by the use of the Gigabit media conversion setup ahead of my digital streaming stack.
At this level of the audiophile experience, you typically have to spend a boatload of money to get the very slightest incremental improvement in your overall sound quality. But for an improvement like this to be so very obvious — and at a cost of less than $100 — is pretty mind-blowing. Say what you will, but unless you’ve experienced a similar situation yourself, all the theoretical arguments are just that — theoretical.
I’m currently auditioning a Benchmark DAC (I am preferring the Gustard x26 Pro that I also have right now). I called Benchmark and spoke with one of their Engineers (they were extremely nice). I asked him if I would get an improvement in sound with Qobuz if I changed from a Wi Fi connection to an Ethernet connection. He said unless I am experiencing dropouts than the answer is no, Qobuz is getting the data in packets and buffering. It would seem to me that a Wi Fi connection would eliminate noise problems that result from an Ethernet connection. Tom, convince me that I should try your method.
I won’t try and convince you of anything; I’m just relating my experiences, which have been very good with wired connections versus wireless connections. I’ve been involved in digital streaming for almost a decade now, and have had considerable problems with wireless setups — dropouts, hiccups, etc — but nearly none in a wired setup.
I looked back at your posts on my Gustard X16 article; and I’m pretty sure you’re streaming from a PC to your DAC (using Foobar?). I assume that you’re probably using the desktop client for Qobuz, etc., to play music? The setup I’m describing here involves a network-connected streaming setup, where your music sources are derived from the network (Qobuz, Tidal, etc.), your ripped music collection is network-connected, and you’re using a network-connected, dedicated “streamer.” Which can be in the order of a Sonore Rendu unit, or something homebrew, like a Raspberry PI or Intel-NUC based design. This can be accomplished wirelessly, but I personally cannot tolerate anything other than a hard-wired network setup. That can be easily managed by a tablet, iPad, or smart phone, which is a much more elegant streaming solution.
The system I’m describing is not inexpensive, but is also not terribly expensive — if ultimate fidelity is your goal. It’s definitely mine. The insertion of a Gigabit fiber media converter is designed to enhance a hard-wired, network-connected system, so it’s probably not appropriate for your situation. Perhaps if I knew a bit more about your system setup, I could offer a better assessment of your situation.
Tom – It’s a privilege to be be able to converse with you and get you expert advice. I was auditioning the Gustard x16 based on your review and liked it better that the Chord Qutest, which I returned. I figured my budget was $1500 for a DAC so I have tried the Benchmark DAC3 HGC (open box price) and also the Gustard x26 Pro (I was intrigued by the “dual mono” design as I thought it would offer the advantages of monoblocks). As I said, I prefer the x26 to the Benchmark and the x26 is superior to the x16 – is it $1000 superior?, maybe not but I’m thinking I will be keeping it.
The rest of my system (ignoring my record player setup) is an old preamp made by an obscure company called Professional Systems Engineering (PSE). I have found this preamp to be an exceptional sounding unit, and high value for the dollar. The PSE feeds a pair of Bryston ST 7B monoblocks and my speakers are Magnepan Tympani IDs. I also have an Audioquest Niagara power conditioner and Power cord, and an SVS sub. Interconnects are decent, Morrow Audio, Wireworld, Cardas, etc.
As you know, my digital front end is a Windows 10 computer connected via Wi Fi to my DAC via USB (Wireworld USB). I have also been using an Audioquest Jitterbug.
I’m pretty happy with the sound of my WAV rips played by Foobar2000. But I do feel that Qobuz could sound better.
If I told you my budget to improve my streaming was $1000 what would you recommend? One of those options could be returning the x26 and keeping the x16, which I still have. Part of the budget might include costs associated with getting an Ethernet connection near my DAC.
Thank you so much!
You have a very nice system; I have almost 40 years of experience with six different pairs of Maggies and I also regard the Bryston amps very highly.
I don’t know how difficult it will be to get a hard-wired network connection to your listening room. Mine is in my basement, and the room with all my network equipment is in the room directly behind my equipment stack, so it was very easy to make the needed connections. I ran all the network cabling in my house myself, but at that point, it was only bare stud walls, so it was easy to run the wiring at the time to make multiple connection options possible. You may be able to get creative in getting an ethernet signal to your listening environment without too much effort or costing too much money.
I’d recommend a Sonore microRendu 1.5 for a streamer. Here’s a link: https://www.sonore.us/microRendu.html At $399, it’s probably the best affordable high-end streamer available, and will handle any level of resolution of digital files of all types, but it needs a power supply. I’d recommend this one from Small Green Computer ($189): https://www.smallgreencomputer.com/collections/linear-power-supplies/products/7v-linear-power-supply-25w I used a first-generation microRendu for a year in my system, and it worked perfectly during that time, and is remarkably close in sound quality to Sonore’s much more expensive offerings like the UltraRendu. You simply connect the microRendu to a power supply and to your network (it has a built-n RJ-45 ethernet port), then use the available USB port to connect to your DAC. You then connect to Sonore’s website on your PC, where you can easily find your network connected microRendu and use their setup options to program your unit for your needs. If you wanted to go the optical route, you’d connect the media converters between your ethernet connection and the microRendu.
Here’s a potential problem: while the microRendu’s setup menu offers a plethora of connection and streaming capabilities, many of them require the user to create a media server setup from DLNA or UPnP open source software like MinimServer or Linn Kazoo, Logitech Media Server or Squeezelite, or even HQ Player. The best (but costliest) choice would be to get a Roon subscription ($12/month). I went the relatively cost-free route with trying DLNA and UPnP apps, which were buggy and glitchy, and after two or so years of banging my head against the wall, caved in and went with a Roon subscription. That solved all my problems, and worked perfectly with every file type of every resolution, and Roon has the industry’s best graphical user interface. Roon has a core that requires a couple of GB of computer space — you could probably place it on your PC without any issues. Ideally, you’d then use a tablet or iPad to control your music playback wirelessly (that’s how I do it), and it works pretty effortlessly. One of the goals is to not have a noisy computer in your listening room.
Moving to a great high-end streamer will elevate your music enjoyment to a level you couldn’t have previously imagined. After having experienced good high-end streaming, I don’t think you’d ever look back.
Thanks for the reply, Tom – I’m looking into what you suggested — is there a way to use the I2S input on the Gustard with the Sonore?
In the Sonore microRendu stock setup, there’s no way to use I2S. Sonore does make a USB-to-I2S converter, the UltraDigital, but it sells for an extra $300 and also requires a power supply.
I don’t know if you read this from Copper 141: https://www.psaudio.com/copper/article/streaming-digital-audio-via-the-i%c2%b2s-connection/ In this article, I talked about my experiences with the Douk Audio USB-to-I2S converter, which sells on Amazon for about $50. In terms of sound quality, it was off-the-charts great, but there are a few caveats that need to be noted: 1) it doesn’t come with a power supply. I happened to have an available 5V USB power tap on another LPS in my system that matched the Douk power specs and was able to use it, but you’ll need to figure out how to provide power to it should you decide to get one. The Douk will power itself via USB, but I didn’t think it maintained the same level of audio clarity with the internal USB power as it did with a dedicated power suppy (hopefully a linear power supply). 2) With the I2S connection going from the Douk into my PS Audio Stellar Gain Cell DAC, the Douk would pass PCM signals perfectly, but would reverse the channels on DSD. It took a couple of weeks before this became totally obvious to me. A couple of months after getting the Douk unit, Euphony Audio upgraded my Summus Endpoint unit to include I2S, so I haven’t been using the Douk since then.
Since my experience with the Douk unit, I’ve been researching I2S considerably, and the thing that’s really jumped out at me is that the I2S specification doesn’t have a completely clear level of standardization; for example, what might work perfectly for one implementation doesn’t work perfectly for the next. Most uses of I2S employ an HDMI connector, but I2S and HDMI are completely different animals. And different manufacturers use different HDMI pin configurations when using it as I2S. The Douk unit reversed DSD channels with my PS Audio DAC, but it might not with another manufacturer’s DAC. On Sonore’s website, there’s a graphic in the product information for their UltraDigital unit that shows I2S pin configurations, and it seems to me that most manufacturers use a similar standard — except PS Audio. But that’s really an oversimplification, and not exactly true with all equipment.
When I got the Gustard X16 and connected it to the I2S output of the Euphony Summus Endpoint, I immediately noticed that the channels were reversed for both PCM and DSD files. All I had to do to correct for this was to swap the left and right balanced cables going out of the X16 and into my preamp, which fixed everything. In my ongoing research with other manufacturers that utilize I2S, I’ve discovered that many manufacturers have menu options in their DACs that will allow you to adjust the I2S pin configurations via software to compensate for reversed channels, etc. — but the X16 does not. Thankfully, it was consistent with both PCM and DSD files, and only required switching the output cables to fix the problem.
A lot of people apparently bought the Douk unit based on my recommendation, and I got many emails sharing my enthusiasm for the unit, but one reader had the same channel reversal problem I had (he also used a PS Audio DAC), and was unable to resolve his problem. By the time I became aware of his issues, I’d already gotten the I2S upgrade on my Euphony unit and was no longer using the Douk unit. The reader told me that he’d reached out to Douk to try and get a fix for the problem, and they weren’t very responsive (not terribly surprising for a Chinese-based company) and it was too late for him to get a refund on the equipment. An unfortunate situation, but for only $50, not too big of a loss.
I just looked on Amazon, and it appears the price has gone up to $68 (with free shipping). Here’s a link if you’re interested: https://www.amazon.com/Digital-Interface-Coaxial-Optical-Audio/dp/B08HN3VSF8
I haven’t tried using the Douk unit with the X16, so I don’t know if it would have the same channel reversal problem as with the SGCD. And the Euphony people (whom I’ve been working with for several years now) know that I use a PS Audio DAC, and may have adjusted their HDMI/I2S pin configuration to work perfectly with my longtime DAC (I didn’t get the X16 until November). If I get a chance, I’ll see if I can hook up the Douk to the X16 and see what happens.
Thanks, Tom – maybe I’ll give the Douk a try down the road after I get the rest of this worked out. Right now I’m figuring out how to get ethernet near my setup. I have used an ethernet bridge TP-Link) in the past to get ethernet over the house wiring but I got concerned that it was causing “dirty power”.
Hi Tom – I got a microRendu with their power supply. I got it up and running yesterday with a Roon trial. I’m using my Netgear consumer grade router with a 20 foot Cat 5e ethernet cable. Using a nice (and short) Wireworld USB cable to my Gustard x26 Pro DAC. Running Roon on an older HP Laptop with Core 5 processer. I’m “underwhelmed”. Compared to the setup I was using (Foobar2000 for playing WAV files and Qobuz on the HP) I’m missing clarity and sparkle. I listen to a lot of jazz and I’m not getting the ride cymbal as clearly, if you get my drift. This is happening on the files on my drive as well as Qobuz through Roon.
Do you think I will be happier with the i2s connection? I see the Douk converter on Amazon and I found an iFi Power X for $109. There is also the Sonore unit for $499 and it does not appear to need (or have an option for) a power supply. Not sure if the Sonore is worth the extra bucks.
Always sad to anticipate a sonic improvement and not get it. So I guess my options right now are:
1. Return everything and go back to my old setup.
2. Get the i2s converter and see how it sounds.
I think one of the things that drove me to do this was that a lot of Qobuz files didn’t sound so great. Maybe that’s just Qobuz. I also would like to control Qobuz from my listening chair. I can access my files (with a volume control) with the Foobar remote app on Android which works well.
The other thing that annoyed me was that I apparently cannot access KUVO radio through Roon. If I do stick with the microRendu is there a way I can access KUVO on the PC through their website and send it to the DAC?
The other thing is the optical isolation you have described in this article. I’m sure it improves things but I’m skeptical it would restore the “sparkle” I’m missing.
I really appreciate the dialog with you!
Very informative article. And from my experience you are correct. Wired connections for me have always been rock solid even over relatively long distances.
Thanks, and thanks for reading!
I experimented with media converters earlier this year. I was so impressed by the improved sound quality that was provided, that I changed over my entire LAN from copper to fiber. I read all over that if you don’t have fiber in place already, then it’s not worth the cost to convert. I’m here to say that I found the opposite to be the case. When experimenting, I found that each new converter in the chain brought less boomy bass, less midrange bloat and highs that extended more naturally without being excessively strident. Currently my audio chain starts with an Intel NUC, ethernet out to a Sonore media converter and then into a fiber switch. This same switch houses the feed from my NAS which I’ve installed an Intel X520 NIC with a 10G fiber transceiver. Another run from this switch goes to the fiber input of my ASUS RT-AX89X router. And the last run from this first switch goes through 50′ of fiber and into another 8 port 10G fiber switch. This is where all of my media is located. But audio is all we’re discussing here, so I have a short run of fiber to an Uptone EtherRegen switch that I literally just use as a higher end media converter. And then a short run of ethernet to my PS Audio DS DAC. So the only copper ethernet in the entire chain is from the NUC and into PS Audio DAC. All media convers, router, switches, NUC, etc have linear power supplies and they all matter. I’ve A/B’d them in and out of my system and it’s all subtle but noticeable. I’m fortunate to have fiber running into my home as well… Anyway, to say I’m a fan of fiber in the digital audio chain is an understatement. Nice article and hopefully others reading this will try the fiber isolation tweak.
It sounds like you’ve got a really great system in place! I’m a big fan of the UpTone Audio equipment, and would love to have the newest Ether Regen, but they’ve been out of stock waiting for parts with the pandemic going on. Maybe later this year!
Thanks for reading, and enjoy the music!
Very cool project. Thanks for sharing. This project is affordable enough that I’ll try it. I want to see if I can replicate your results on my much more affordable system. Give me some time, but I’ll report back with my findings.
My question for you is whether you believe that, once optical internet is in place, whether anything upstream from the optical connection will make a difference. Examples of upstream gear may include low noise power supplies on the modem and router, Roon Core quality when the Core is separate from the Bridge, or upgrading the coax internet cable that feeds into the modem.
Like you, I like to tinker, and I am looking for budget tweaks that will make a big difference for A/V reproduction. While in the market for a shorter length of coax internet cable to hide my modem behind the furniture, I recently stumbled upon the fact that coax internet cable is usually made of copper plated steel. It seems to me that upgrading the coax internet cable throughout the house to at least copper-only coaxial cable could be worthwhile. Since the work involves the attic, I might as well install a dedicated power line at the same time. But your optical line is even more affordable with equal or less work, so I’ll move an optical line ahead in the queue.
Here’s the recent discussion I had on the coax internet cable in case you’re interested: https://forum.audiogon.com/discussions/coax-internet-cable-if-i-have-to-build-one-what-am-i-looking-for-exac
From most of my explorations online, the most important point in the chain is the media converter that’s located nearest your streamer setup. Once the data stream reaches that location, its via the optical cable, so none of the grunge or digital noise that might have been present on the copper ethernet line will already have been eliminated. If you have a linear power supply for that location, that will optimize the performance of the media converter, and enhance your overall sound quality. I’ve read countless online entries where most of the participants encourage the use of linear power supplies for routers, modems, network switches, NAS, etc. I haven’t done that yet, but I’ve read many positive reviews of setups that used linear power supplies with that type equipment from Jameco that were gotten from Amazon or elsewhere, that typically sell for less than $20 each, and do add an additional level of enhancement to your system. While I’m perfectly happy with what I’m currently hearing, at that price point, I’m definitely going to take the leap!
Thanks for reading!
Hey Tom, I got the fiber line up and have been listening for almost a week now. I am pleased to tell you that I feel the upgrade in sound quality is significant. I’m going to continue to listen with the optical line in place for at least a month before I switch back to see if I can quantify the difference it made.
My system’s new retail price comes out to 5k, so about a 1/10th the value of your system. I would think that the difference I heard would be even more pronounced on your system. The noise floor went down a level. I’m hearing more micro-details. And emotionally engaging with the music is more effortless. I recently tried upsampling in Roon, but I can’t say I heard a difference there. But with the fiber line, the difference was immediately apparent. So, nice find on this tweak.
I started researching with your gear recommendations. I thought I knew networking gear, but optical is another animal: the SFP+ standard isn’t backwards compatible, not all transceivers fit into all converters, and you have to make sure you buy the corresponding single or multi-mode cable to match the transceivers. That is to say, I’ve learned a lot.
I like TP-Link as a brand, so I copied you and used their converter. But, I also used their multi-mode transceivers to keep things easy. And for the fiber cable, I went with OM5 (the highest standard of multi-mode) from Monoprice, who I also like as a brand. If I’m itching to try another affordable test in 6 months, I could A-B test multi vs single fiber with other offerings from TP-Link and Monoprice. Because transceivers are hot-swappable, switching out the cables would be quick.
I also had an unused iFi 9V low noise power supply, so I placed it on the converter nearest the streamer. At the moment, I don’t feel the need to A-B test the iFi against the converter’s stock power brick.
I didn’t run the optical through the walls to start. My converters sit next to each other. But now that I’ve learned this, I may run optical through the attic to the home theater area to see what happens to the picture.
I don’t know of too many HiFi manufacturers who build fiber converters into their streamers or switches. I’ve seen what a streamer upgrade could do, so I will eventually try a switch or converter. UpTone and Sonore seem to be the only options.
I’ve looked at that UpTone etherRegen unit several times, but haven’t pulled the trigger. I’m kind of intrigued by its “B” output port that’s supposed to greatly improve the signal going to your streaming setup. I’ve owned one of their LPS-1 power supplies for years and love it.
It looks like UpTone and Sonore components are both unavailable right now, possibly due to supply chain issues. On their site, UpTone even says that they are attempting to source parts elsewhere now. Fingers crossed that they’ll put together a switch that can handle at least 1Gb speeds next. The prior switch was limited to 100Mbps. Another issue with the UpTone switch is that it would only partially work with optical, unless I didn’t use it for the entire home network.
I recently talked with iFi about their Zen Stream, and the rep told me they were looking into ways to incorporate optical with a future iteration of a streamer.
@ClassDStreamer I’m not Tom but I have gone through all that you ask about. If you read my post above, you can find my opinions on the subject at hand. I say opinions because I am not here to try to convince you that upstream LPS’s are guaranteed to make a difference in your setup. But they certainly did in mine. I will not go back to cheap, wall-wart PSUs. Tom gave excellent advice in going with some Jameco transformer PSUs for the media converters which cost about $15/ea. Just make sure the center pin is hot and that they’re regulated power supplies (they also make 5v unregulated). I still use several Jameco PSUs in my video streaming setup (on the media converters). And surprisingly, the linear supply made a bigger difference in my router than it did in my NUC running Roon ROCK. I try to only use R-core transformers…an OCD thing I guess.
You have a lot of Gizmos in your system, Cliff. You say you have a fiber internet line running into your house. I’m assuming that my internet line is coaxial. I’ll have to take a look at the box in the attic to verify. I do know I have coax running through the attic to the wall where I connect the modem.
To start, I will try placing an optical line in front of HiFi streamer to see what happens. If I like the results, I’ll consider running an optical line from the switch to the home theater system too.
I don’t have any local servers like you do. I stream all of my audio and video content from the cloud. So I don’t believe I would benefit from converting my mesh network to optical lines. As I understand your and Tom’s recommendation, you guys are just placing optical right before any A/V streamers.
Great post and your summary is consistent with my results of implementing a fiber connection to my streamer
Vastly superior sound presentation than the copper connection
The engineers and network purists are really busting your chops and are skeptical of the benefits that you’ve accurately articulated
After all it’s just 1s and 0s and layer0 medium choices
I wanted to up my game in digital playback and my options were to upgrade the transport or improve my streamer
With the advent of hi-res streaming formats and vast libraries of Tidal and Qobuz, investing in a new transport struck me as costly and perhaps not quite yesterday’s news, though playing CDs has peaked and definently in decline
I went the Sonore route as well with Small Green Computer
I ordered a network switch on Amazon, recommended by SGC, with a fiber port on the switch, thus no media converter needed
Fiber from the switch to the Sonore Optical Rendu, the output on the Rendu is USB
I could’ve gone USB from the Rendu to the DSD DAC but instead added another Sonore box, the ultraDigital, that’s USB in and i2s out
The ultra digital allows me to leverage the i2s input on the DSD
The i2s connection shows up as another endpoint in the Roon core
I can A/B compare the fiber and copper connections by simply changing endpoints in Roon
The digital playback SQ hierarchy on the DSD connected devices are as follows, in order of good to best
1. Transport to coax
2. Copper to Network Bridge II
3. Fiber to streamer, USB to DSD
4. Fiber to streamer, i2s to DSD
The first 3 do a great job of playing the music but the i2s playback is a step up and sounds more like the artist or the band is being presented
All the instruments and vocals sound harmonious and are complementarily working together
The sum of the parts has exceeded my expectations while providing exceptional return on price performance
I do not know the science but I do trust my ears
Thanks for your reply. It’s really great with these network guys and their insistence that “it can’t possibly sound any better” when they’ve never actually listened to it. Whatever!
I also use I2S in my system; my Euphony Summus Streaming system was updated to I2S last August, and both my PS Audio Stellar Gain Cell DAC and my Gustard X16 DAC have I2S inputs. The difference in sound quality via I2S versus USB is astonishingly good, and for me, there’s no going back.
Thanks for reading, and enjoy the music!
Also many thanks for the tip on the after market Jameco linear power supplies
No doubt the router and switch will benefit
Well done Sir
Thank you! Just doing what I love!
I copied this, below, from the comments section of the next page as I think it probably belongs here.
I agree with the sentiment though as I understand and can remember concepts as pictures/diagrams much more quickly than just as words. The article was great, by the way and I intend to replicate your efforts.
Please, please, please use simple block diagrams to illustrate all the various elements of system interconnections/devices. The problem with experts writing is that they are experts.
It was from a reader named daddyo.
Tom, I’d love to see a diagram of what you described. For me, at least, relatively uninitiated, it’s hard to follow a text description.
Btw, I also had a heck of a time with DNLA and UPnP between a router and the (then) Directstream dac on my Win 8 (or 10?) desktop. Lots of connection drop outs. I wasn’t alone in that regard, as I seem to recall PS Audio taking steps to overcome it subsequently.
Great article ! I quickly implemented this conversion to feed the Bridge II network card in my Perfect Wave MkII DAC. I used a 9v iFi Silent Power 2 on the end converter which measures noise on the order of a few micro volts as opposed to many linear supplies which measure in millivolts. The iFi cost me $75 delivered. I have been really rather amazed in the difference this has made ! Many Thanks
That’s awesome! Glad it worked out for you. And thanks for reading!
Thanks for this. I have been wanting to try out fiber optics but didn’t want to incur the expense of replacing my Ultrarendu. I bought the exact equipment you did and it worked out great. It definitely seems like a quieter blacker background. I didn’t upgrade the power supply on the last converter yet. I took pwayland’s suggestion on the iFi Silent Power 2. Hopefully, that will add something, too. Thanks, again.
Glad to hear that’s working out for you! Let me know how the iFi unit works out.
Thanks, for reading!
Tom, I do think the iFi adds something. I will probably get another one when I add fiber to my 2 channel system. Headphone system is PC>Ultrarendu>Yggy>Kenzie Encore>JOTR>SR1a. Thanks again for explaining how to do this on the cheap!
So glad to hear that it’s working for you! It’s really great when a relatively inexpensive tweak makes a big difference in your sound quality.
I’m Pino and I’m reading you from Venice, Italy.
I would like to thank you for this article, I decided to make the test buying exactly the same components and in my case it worked, even if I didn’t install yet a linear PSU on the second media converter.
I used your configuration to connect my Roon ROCK server to my streamer, an Ifi Zen Stream + Ifi iPower Elite, the streamer is then connected via USB to my Lector Digitube SD192 DAC.
I have to say that my ROCK server configuration is a bit different than the standard recommended by Roon, I use a standard i5 desktop with 12GB RAM and 512GB SSD, it is directly connected to my ISP router via ethernet; then I installed a second PCIe network card that communicate only with the streamer on a private IP (doing several tests, I found this configuration much better than using the ISP router ethernet ports to connect my Roon devices).
Using the media converters between the server and the streamer I immediately noticed a much more clear sound, more details, it is like having changed a much more expensive device in the sound chain.
So thanks again for sharing your experience with us.
Glad that it worked for you! At this level of high-end audio, you usually have to pay a lot to get a small increment of improvement, but with this tweak, you get a lot, and it costs nearly nothing!
Thanks for reading,
I suppose you tried the second media converter also with it’s own cheap PSU: could you tell me what should I expect replacing it with a high quality device?
Honestly, I did not try the cheap wall-wart power supply with the second media converter. I happened to have a correctly sized LPS on hand that had an available tap, so I made the connection there right out of the gate.
I probably researched this for nearly two years before I finally pulled the trigger. And at virtually every site I checked out, everyone who’d added the media conversion setup noted that a step-up in performance was achieved by using an LPS with the second media converter. With virtually every high-end digital streaming setup I’ve seen in audiophile systems, most everyone absolutely insists that you need to use an LPS with every piece of network equipment, including routers, switches, etc. That doesn’t mean that you need to spend huge bucks for LPS’s — you can get the Jameco LPS’s online for about $15-20 each via Amazon or elsewhere. And from what I read, they give you much of what you’d expect to get from a $200 LPS. I’m in the process of ordering Jameco LPS’s for my router, the first media converter, and the couple of switches in my home network setup. Just make certain (if you go that route) to order the correct LPS for your equipment — it can be a little tricky, especially when you generally need a magnifying glass to read the specs on the wall warts that come with everything.
I have LPS units in place with all my digital equipment as well as with the two phono preamps I currently use (I have two turntables, one for mono and one for stereo). Adding the LPS’s to the phono preamps was a shocking improvement in every area of sound quality — I can’t recommend LPS’s highly enough.