With the advent and rise of steaming music services (e.g., Spotify, Tidal, and Qobuz) in recent years, along with the ever-increasing use of music servers in conjunction with ripped or downloaded music files in lieu of traditional physical media, more and more audiophiles have come to rely on LAN-type network connections for providing content to digital components. While Wi-Fi-based connections can work well for transmitting digital files or streaming music files from service providers, most listeners are finding fewer dropouts, more consistent connectivity, and higher audio quality is obtained by using physical LAN-based connections. Copper Ethernet cables have been the traditional means for physical LAN-based connections, but recently, more audio companies and listeners are utilizing components that support fiber-optic-based connections. This article discusses the foundation, advantages, and components of fiber-optical LAN connections for digital streaming front ends in high-end audio systems.
Background and USB DACs
The advent of asynchronous USB DACs and streamers about 10 years caused a revolution in digital music reproduction. No longer shackled by physical media and disc players to provide digital content, a lot of folks, including me, started connecting computers directly to USB-capable DACs. While the ultimate quality of reproduction was arguably not as good as with disc players or LPs, it was good enough in most circumstances. And, for many, the convenience of having hundreds or thousands of recordings readily available from an iPhone, iPad, or tablet outweighed any downsides with respect to “ultimate” audio quality.
As we started connecting our computers directly to USB-capable DACs, however, we discovered that USB as a communications protocol and a digital interface had ”issues.” While USB was really convenient, it often didn’t sound quite as good as other types of digital connections, e.g., a SPDIF interface via coax. Gordon Rankin’s development of the Streamlength code showed the benefits of asynchronous (rather than isochronous) USB with respect to improved timing and the concomitant increase in audio quality. It was also discovered that the USB receivers in computers were “dirty,” and that USB cables, which folks had mistakenly thought were only transmitting 0s and 1s, had a significant impact on audio quality. It turned out digital streaming wasn’t, as many had initially thought, a bit-perfect stream immune to noise, but a system where everything mattered.
Enter the Network Bridge
Jump ahead about five years, and we’ve seen the development of the digital streamer and network bridge for high-end audio. (See the article by Jeffrey Barish in The Absolute Sound at https://www.theabsolutesound.com/articles/understanding-digital-music-systems.) These components are purpose-designed to provide a dedicated interface for connecting with home computer networks to stream digital content, and they bring notable advantages (e.g., improved functionality, flexibility, and audio quality) over connecting a computer directly to a DAC or to integrated amps with digital inputs.
A product that debuted a few years ago which raised the bar with respect to digital streaming was the microRendu network bridge, launched by Sonore and Small Green Computer in 2016 (reviewed in Issue 218). For the first time, here was a small computer and custom operating system purpose-built for streaming digital music files specifically for high-end audio reproduction. Without frills and frou-frou, what the microRendu did was remove a number of noise sources that degraded digital music reproduction, in addition to re-clocking the digital bitstream and thereby reducing timing errors. And timing errors, most notably in the form of clock phase noise, would prove to be a key metric for digital streaming front ends.
I’ll share my own journey with the micoRendu as an example.
Prior to getting the microRendu, I used a Mac Mini sitting in my audio rack as a music server, connected directly to my Schiit Gungnir DAC with an AudioQuest Diamond USB cable. Without any other reference, I thought it sounded great. Little did I know!
In December 2016, I bought a Sonore microRendu so that I could use it as an endpoint for Roon, which had also recently come on the market and was making a very big splash in the audio world. After installing the microRendu, I moved the Mac Mini and the hard drive that contained my music files into my bedroom/study, approximately forty feet away from the audio rack. The microRendu requires a copper Ethernet cable connection to work, so initially I used an Apple Airport Express in the audio rack with the microRendu. The Airport Express connected to the Mac Mini music server via my Wi-Fi network and to the microRendu via an AudioQuest Cinnamon Ethernet cable. With this setup, I used Roon to stream files from the music server to the microRendu, where they were re-clocked and subsequently passed on to the Gungnir for D/A conversion. I immediately noted significant improvements in audio quality: a lower noise floor; a wider, more spacious soundstage; less digital glare; and, overall, a notably more accurate, more natural, and more lifelike presentation.
All Is Not Well In Paradise
All was well and good until about mid-2018, when my system started having frequent problems with dropouts and streaming interruptions. I finally traced the problem to a recently installed Wi-Fi-connected video doorbell. When the doorbell’s motion sensor was triggered, it would usually result in drop-outs and interruptions in my digital stereo system. The situation became so bad that occasionally I couldn’t get a single song to play all the way through without disruption.
At this point, I decided that it would be worthwhile to consider running a direct physical LAN connection from the Mac Mini in the study to the network bridge in the audio rack. This should result in better connectivity, I thought, as I had read in a support bulletin from Sonore that a direct Ethernet connection from the server to the network bridge would result in better overall performance, as well as higher audio quality than Wi-Fi. Around this time, I also become aware of a new product from Sonore called the OpticalRendu, which used “optical Ethernet.” Further research showed “optical Ethernet” is an optical fiber often used for very long runs (in the 100m to 1km range) of Ethernet connectivity, as it has significantly less insertion loss than copper Ethernet.
Advantages of Optical Fiber
Putting my scientist hat on, I went into full research mode and started reading everything I could find about digital streaming using copper Ethernent versus optical fiber. Here’s what I learned.
Consumer-grade computers contribute significant high-bandwidth RF and impulse noise from their CPUs and GPUs that audibly impacts and degrades the sound quality of a stereo system. Additionally, this high-bandwidth noise can be picked up by many speaker cables (most of which are unshielded for sound engineering reasons), which literally function as antennas for high-bandwidth noise components that are then fed backwards into the power amplifier, to be amplified as noise. Moreover, any smart devices in the home (mobile phones, Wi-Fi routers, tablets, non-audio computers, video doorbells, thermostats) also contribute to the high-bandwidth RF and impulse noise in listening rooms. Consequently, one of the most significant things you can do to improve your digital audio system is to move any computer-based music servers (laptops, Mac Minis, Intel NUCs, etc.) out of the audio rack and well away from the main system, as these devices are very dirty with respect to the noise they create. The inverse-square law pays big dividends here.
The “el cheapo” clocks in consumer-grade cable modems, network routers, Ethernet switches, and fiber media converters also contribute notable clock phase noise to the analog square wave voltages that actually comprise the digital bitstream, and the more of these devices in the configuration, the more the original signal is degraded.
The dreaded switch-mode power supply (aka SMPS)—the ubiquitous device that powers almost everything from a computer’s internal power supply to streamers, network bridges, routers, NAS’s, external hard drives, switches, fiber media converters, etc.—are very dirty and nasty sources of noise, as they create both low-impedance and high-impedance AC leakage currents, which travel down DC power busses and lines, and ultimately into our DACs. High-impedance leakage currents arising from SMPS are particularly insidious, as they cause increased jitter and clock phase noise.
Copper Ethernet cables are also susceptible to a number of noise factors, including RF, EMI, and the low- and high-impedance leakage currents described above; in addition, they suffer from a lack of galvanic isolation and common-mode noise rejection. In particular, shielded Cat 7 and Cat 8 Ethernet cables that are connected at both ends actually serve as conductors for high-impedance leakage currents.
A good mitigation strategy for these problems is to use a run of optical fiber between the music server and the network bridge, streamer, or DAC. Optical fiber has a number of advantages over copper Ethernet: It is inexpensive, thin, flexible, and very easy to route. Most importantly, because the digital signals are transmitted as light, optical fiber is immune to RF and EMI and will not pass high-impedance leakage currents from computers, NAS’s, and routers to network bridges, streamers, and DACs. This results in a significant reduction in noise across the entire chain of streamer components, and notably cleaner, quieter, more transparent, and more natural-sounding digital music reproduction.
The nice thing is that it’s easy and straightforward to set up an optical-fiber-based network connection. All you need are two fiber media convertors (FMCs), optical transceivers (if not provided with the FMCs), and a length of OM-1 specification optical fiber to run between them.
The fiber media converters are active devices, so you’ll need a power supply for them. You can use the power supplies that are typically supplied with the FMCs, but as these are typically switch-mode supplies, I strongly recommend ordering a good linear supply that matches the voltage requirements of your specific FMC. I’ve used inexpensive Reliapro linear power supplies from Jameco Electronics (approx. $11 each) very successfully, as well as power supplies purpose-built for audio applications, e.g. the Uptone Audio LPS-1.2 and the Keces P3. These power supplies support a range of voltages so you can set them to match your FMC’s voltage requirements.
Below is a list of equipment for a basic setup:
1.2x TP-LINK MC200L Gigabit Media Converter, 1000Mbps RJ45 to 1000M multi-mode fiber, up to 550m/1800ft.
2.2x TP-Link TL-SM311LM 1000-Base 850nM MMF LC/LC optical transceivers.
3.Cat 6 Ethernet cables. (I use Shunyata Research Venom as a minimum spec Ethernet cable; Shunyata Alpha or Sigma Ethernet cables will deliver notably improved audio quality.) The number required is system-dependent.
4.Tripp-Lite Duplex Multimode OM-1 62.5/125 Fiber Patch Cable (LC/LC termination). The length will vary with your application, but the fiber optic cable is inexpensive. For example, 30 meters costs approximately $40.
5.2x 4.5W Jameco Reliapro AC-to-DC Regulated Linear Wall Adapter 9 Volt, PN:1953639
6.Optional Cisco WS-C2960L-8TS-LL Catalyst Ethernet Switch (8-port).
Here are the steps to connect a basic fiber optical network:
1.Connect your music server/NUC/Roon Core to your router with an Ethernet cable.
2.Connect the Ethernet cable from the Router to the Ethernet switch.
3.Connect the Ethernet cable from the Ethernet switch to the fiber media convertor.
4.Connect the optical-fiber patch cable from the upstream FMC to the downstream FMC 2 that will connect to the streamer or network bridge (length/run will depend on your setup and application).
5.Connect Ethernet cable from the downstream FMC to the streamer or network bridge.
A representative basic configuration is shown here in Figure 1.
Sonore Optical Module
If you like the benefits obtained from the basic configuration, you can go “high-end” and replace the generic FMCs with Sonore’s OpticalModule and use higher-quality power supplies. The Sonore OpticalModule was designed specifically for high-end audio applications, and features a considerably higher-specification ultra-low-jitter femto-oscillator (i.e., “clock”) and four high-quality, ultra-low-noise linear regulators. Using a OpticalModule FMC will result in considerably less jitter, lower clock phase noise, and a notable decrease in the noise floor compared to generic FMCs. Most notably, the lower clock phase noise results in more precise and accurate timing, and a concomitant increase in overall audio quality.
Below is a list of equipment for a setup using the Sonore OpticalModule:
1.2x Sonore OpticalModules.
2.2x TP-Link TL-SM311LM 1000-Base 850nM MMF LC/LC optical transceivers.
3.2x Uptone Audio LPS-1.2, or Keces linear power supply.
4.Cat 6 Ethernet cables. (I use Shunyata Research Venom as a minimum spec Ethernet cable; Shunyata Alpha or Sigma Ethernet cables will deliver notably improved audio quality.)
5.Tripp-Lite Duplex Multimode OM-1 62.5/125 Fiber Patch Cable (LC/LC termination). Length will vary per your application, but the fiber optic cable is inexpensive. For example, 30 meters costs approx. $40.
6.Optional Uptone Audio EtherREGEN Ethernet switch.
Figure 2 depicts an optical fiber configuration using the Sonore OpticalModule.
Less is More
The purpose of using optical fiber and networking devices with better clocks and power supplies is to reduce sources of noise and improve timing of bitstreams. It was originally thought that digital sources and connections were impervious to the effect of noise on audio quality; after all, they’re just 1’s and 0’s, right? What can go wrong? Well, it turns out that bitstreams are not composed of 1’s and 0’s; that is only how the data comprising the music file is encoded. What is actually transmitted from server to DAC are analog square wave voltages. As result, it turns out that…everything matters.
In my home network and system, adding the FMCs and a run of optical fiber from my router to the downstream network bridge improved the sound of my digital streaming system in notable ways. The noise floor was appreciably lower, which allowed me to hear more deeply into the music. Interestingly, I could lower the volume setting on the preamp, yet still hear as much detail and musical information as I could at louder levels. Compared to Wi-Fi or a long run of generic copper Ethernet, the soundstage was also notably more spacious, open, and airy-sounding with improved focus on individual instruments and voices. On the whole, the overall presentation was a notably more lifelike and natural-sounding digital streaming system.
The advent of the Sonore OpticalModule in the market around this time provided yet another audible level of improvement with a fiber-optic-based network. The higher-specification clock (crystal oscillator) and power supply of the OpticalModule provides improved timing and lowers the noise floor more than generic FMCs, so you get a lot less of the bad stuff and lot more of the good stuff described above. Power the OpticalModule with a good linear power supply, e.g., an Uptone Audio LPS-1.2, and you’ll effectively have a state-of-the-art network system for digital streaming that will be clean, quiet, transparent, and lifelike.
We’ve come a long way with respect to digital music reproduction, particularly in the last five years. Streaming services like Tidal and Qobuz have become mainstream as viable sources of musical content, and components specifically designed for networking digital streaming have become available to provide superior-quality audio reproduction. We’re no longer limited to physical media and disc players, and the convenience and ease of use that streaming offers is matched by equipment and components that provide a concomitant level of superb audio performance. In my experience and in my system, this performance is best realized by direct network connections from streamer to DAC utilizing the advantages that a fiber-optic connection has over traditional copper Ethernet, including immunity from RF, EMI, and high-impedance leakage currents from SMPS and network component clocks that increase jitter and clock phase noise.
Everyone’s system is different, and the degree of benefits in any system can be tricky to assess, but the components and cabling for a basic configuration are inexpensive enough to try for yourself, and see if they work for you. They certainly worked very well in my system. If you like the results, and want to upgrade these networking components, there are now audio-application-specific fiber media convertors, e.g. the Sonore OpticalModule, and power supplies (e.g., the Uptone Audio LPS-1.2) that provide higher performance and audio quality. Moreover, if you find that you also need an audiophile-grade Ethernet switch, the AQVox and SOtM Ethernet switches have been on the market for over a year, and the recently released Uptone Audio EtherREGEN Ethernet switch shipped in Q4, 2019. The EtherREGEN not only provides much higher-quality audio performance than other Ethernet switches, but also supports fiber-optic connections.
Bottom-line: Digital streaming is here to stay. We’ve now got amazing streaming service providers at our beck and call, and we also now have audiophile-grade networking components that lets us realize the full potential of the amazing content we have at our fingertips. My advice: Jump in! The water’s fine.
By Stephen Scharf
I’ve worked as a molecular biologist for my entire professional career. As one of the inventors of Polymerase Chain Reaction (PCR), I worked in human molecular genetics and developing molecular-diagnostic and DNA forensic testsMore articles from this editor
Read Next From BlogSee all
How to Build Your Own Low-Cost, High-Performance Streamer
There is no product category with a wider range of […]
- by Steven Stone
- Jan 25th, 2021