With the increasingly widespread streaming of digital music content to our audio systems, either from discs or streaming services (e.g., Tidal or Qobuz), many audiophiles are finding the need for an audio-grade Ethernet switch to connect their home computer network modems, routers, computers, drive arrays, etc. to an increasingly wide array of devices to mediate the transfer of digital content to digital-to-analog converters. As my digital streaming “front-end” has been in development for the last year and half or so, I’m finding there are now better audiophile-grade solutions that provide network access to my nearly 4TB of disc-based digital content, and to new content via streaming services.
UpTone Audio is a small audio-products firm based in Northern California. Company principal and founder, Alex Crespi, has been in the audio business for over 30 years, and was one of the co-founders of the well-regarded audio products company, Hovland, which was started in 1999 and closed in 2009. UpTone was founded in 2010, and all of its products are developed in close partnership with professional electronics engineer, John Swenson, who has had a long career designing the power-distribution networks deep inside large custom computer chips (the sort of devices that run high-speed network data centers). As UpTone Audio, Crespi and Swenson have developed a successful range of products for digital playback including the REGEN line of digital audio products and the UltraCap line of linear power supplies for digital components and music servers. John is also well-known as the hardware engineer for the popular Sonore Rendu series of products.
I purchased my $640 UpTone Audio EtherREGEN in October, 2019, and I can unequivocally say it has been instrumental in getting the audio performance of my digital streaming front end to a level where it is now as fully engaging, immersive, and enjoyable a musical experience as my turntable-based analog system.
Description and Features
The UpTone Audio EtherREGEN Ethernet switch utilizes the same oval-shaped aluminum case used for many UpTone Audio products, and measures 112mm x 110mm x 30mm. Of note, UpTone has specifically designed EtherREGEN so that it does not require an external linear power supply to provide nominal performance. Rather, it comes with an UpTone Audio switch-mode power supply specially designed to prevent the passage of high source-impedance leakage current to the EtherREGEN (more on this later).
The EtherREGEN is unique among Ethernet switches in that it uses a circuit topology called the Active Differential Isolation Moat or ADIM that completely isolates the side of the switch used for connecting network devices (typically the “A” side) from the side connected to the renderer, network bridge, or streamer (typically the “B” side). The ADIM provides electrical, galvanic, and noise isolation between the two sides. I’ll discuss the rationale and function of the “moat” in more detail in its own section in this review.
On the “A” side of EtherREGEN is a SFP (small form-factor, pluggable) cage for making fiber optic connections using an optical transceiver (Figure 1) and an RJ45 module composed of four specially manufactured copper Ethernet ports with status lights. There are 12 transformer cores in each port and the center-taps of these cores are grounded using capacitors to block port-to-port leakage current.
The “B” side of EtherREGEN (Figure 2) has a DC-power barrel jack, ground post, a single 100-Megabit RJ45 port with status lights, a switch for selecting the internal or an external clock, and a BNC connector for connecting an external clock, if hot-rodding your EtherREGEN is desired. The standard impedance for the clock connector is 75-ohm, but a 50-ohm clock connection can be ordered from UpTone Audio at no additional charge. While EtherREGEN works identically in both directions, the “B” side is typically connected to the computer, streamer, network bridge, endpoint renderer, etc., with a copper Ethernet cable. DACs that have an Ethernet input can be directly connected to the “B” port.
Technical Considerations and Design Features
The Active Differential Isolation Moat (ADIM) design is unique to the EtherREGEN, as it provides data-transfer functionality and audio performance that significantly differentiates it from all other Ethernet switches, including those modified for audio applications. EtherREGEN has two data/power/clock domains which are isolated from each other by the ADIM. These two domains are referred to the as “A” side and “B” side, respectively. Each isolated domain is re-clocked, and the clocking system runs from an advanced, programmable, jitter-attenuating clock synthesizer with four differential outputs. It is referenced to an ultra-low-jitter/phase-noise Crystek CCHD-575 oscillator, and the clock distribution system is run differentially throughout. UpTone Audio colloquially refers to the separation provided by the ADIM between the sides of its circuit board as “the moat.”
The rationale for the moat is important because every signal “edge” coming out of any digital device carries the jitter and phase-noise of the clock used to “clock out” that edge. When dealing with Ethernet-based signals, specifically for audio applications, where timing is critical, there are two types or classes of sound-degrading noises that EtherREGEN was designed to mitigate: 1) low and high source-impedance leakage current, and 2) clock phase noise. Clock phase noise travels on the Ethernet signal and on power and ground planes. This phase noise affects the threshold jitter of chips’ clock inputs.
The differential circuitry was specifically designed to eliminate the signal-borne phase noise going from one side of the moat to the other. EtherREGEN is mostly symmetrical—technically, there is no “dirty side” or “clean side.” While ER works identically in both directions, it is best to have the streaming endpoint device (renderer/streamer/network bridge) that connects to the DAC be the only device that is connected to that particular side (typically the “B” side). While the circuitry between the individual RJ45 ports or SFP cage on the “A” side decreases phase-noise effects to some degree, it does not decrease it to the same degree obtained by crossing the moat. This is because crossing the moat makes use of differential isolators in conjunction with differential flip-flops. (Refer to the engineering white paper by John Swenson, Understanding how perturbations on digital signals can affect sound quality without changing bits, and how these issues are addressed by the UpTone EtherREGEN, https://cdn.shopify.com/s/files/1/0660/6121/files/UpTone-J.Swenson_EtherREGEN_white_paper.pdf?v=1583429386.) And this is the big advantage: The differential isolators prevent the data-borne clock signature from getting onto the ground plane of the circuit board, while the differential flip-flops prevent the clock phase-noise signature from getting into the flip-flop’s own internal ground network, as well as the ground plane of the rest of the board. It is the use of these two differential isolation methodologies together that provides the performance advantage of EtherREGEN compared to other Ethernet switches.
Physically, EtherREGEN circuitry is mounted on a single, premium, six-layer circuit board, but electronically it is actually two boards (Figure 3). Looking at the EtherREGEN circuit board, we see it is galvanically divided into two parts, the “A” side (lower half of Figure 3) that typically handles network connections, and the “B” side (upper portion of Figure 3) across the moat, which provides the fully isolated signal to the renderer/endpoint.
Transfer of data from one side to the other is mediated using high-speed Analog Devices differential isolators. When the network signal has passed to the other side of the moat via the differential isolators, it is sent to the “B” side electronics that re-clock the data with an ultra-low phase noise Crystek CCHD-575 Crystal oscillator and Silicon Labs jitter-attenuating clock synthesizer (unless the clock switch is set to external, in which case, the external clock signal is used).
This is a key point: Because all the clocking in EtherREGEN is run with differential lines, and the isolators and flip-flops are fully differential, there is no noise overlaid on EtherREGEN’s ground plane by any of the el-cheapo clocks from network devices (e.g. routers, music servers, NAS’s, generic switches, etc.) upstream. Furthermore, no leakage current from switch-mode power supplies from these upstream devices can pass across the moat, either. No other Ethernet switch design does this, and these features are key aspects of EtherREGEN’s unique design because they prevent the data-borne clock signature and leakage currents from getting onto the ground plane of the circuit board and into the DAC, where they will increase jitter in the circuitry. Reduction of phase noise, leakage current, and concomitantly, jitter, are the significant audible benefits that EtherREGEN offers when used in a digital music streaming setup.
Set-up and Configuration
The setup and configuration of EtherREGEN is simple and straightforward. Connect a single copper Ethernet cable from their main network router or switch to any of the RJ45 ports on the “A” side (Figure 4) of EtherREGEN. The remaining three RJ45 ports can be used for other connections, e.g. a music server, NAS, or other network device. If an optical-fiber connection is used, it can be connected to an optical transceiver in the SFP cage on the “A” side.
Depending on the user’s network or requirements, different connection configurations are possible. For example, one could leave the network router music server, NAS, etc. connected to another switch, and make a single Ethernet connection to EtherREGEN as shown in Figure 4. How the connections are configured are primarily a matter of convenience and/or the requirements of the specific network, and UpTone says the effects of different configurations on sonic performance are likely to be small, if any.
Networking Setup for Evaluation
I made my evaluations of the EtherREGEN compared to a consumer-grade TP-Link Ethernet switch, as I did not have access to any other “audiophile-grade” Ethernet switches. Since I am using a run of fiber from the bedroom study, where the music server, router, and switches are located, to the main audio rack, I also used a consumer-grade TP-link fiber-media converter (FMC) in conjunction with the TP-link Ethernet switch. It’s important to understand that the base configuration was composed of two networking devices used together to stream the data: the consumer Ethernet switch and a fiber-media converter. The run of fiber connected at the downstream end in the main audio rack to a Sonore Optical Module that was connected by a Shunyata Sigma copper Ethernet cable to my SOtM SMS-200 UltraNeo network bridge, which serves as my endpoint/renderer, as shown in the base networking configuration in Figure 5.
For my configuration with EtherREGEN, I replaced the generic TP-link FMC with the Sonore OpticalModule at the upstream end; the Mac Mini music server and OpticalModule were connected to the RJ45 ports on my network router with Shunyata Sigma Ethernet cables. The OpticalModule was connected with a forty-foot run of optical fiber downstream to EtherREGEN’s “A” side via it’s SFP cage in my main audio rack. A Shunyata Sigma Ethernet cable from the “B” side of EtherREGEN was connected to my SOtM SMS-200 UltraNeo renderer/endpoint, which was connected to my Schiit Gungnir Multibit DAC with a Shunyata Alpha USB cable. The configuration using EtherREGEN is shown in Figure 6.
Power Connections for Network Devices
Because of the insidious and audible problems of leakage current, caused by the use of switch-mode power supplies in virtually almost all network devices, I paid special attention to the quality of power for the devices in the digital streaming chain. In my base configuration, when the TP-link Ethernet switch and FMC were used, they were powered by Jameco Reliapro linear power supplies; my Pace router used its standard switch-mode power supply with a home-made connector that shunts high source-impedance leakage current to ground. For the purposes of this review, the Mac Mini music server was powered with a Shunyata Venom V14D digital power cord, and the Sonore OpticalModule was powered with an UpTone Audio LPS-1 power supply. Both the UpTone LPS-1 and the Mac Mini were plugged into a Cryoparts power strip powered by a Shunyata V12 NR power cord. The Shunyata V12 NR power cord provides approximately 12dB of noise reduction for the power strip, and the Venom V14D digital power cord, specifically designed for devices that utilize switch-mode power supplies (e.g. the Mac Mini), provides another 12dB of noise reduction. In the main audio rack, the EtherREGEN was evaluated using the supplied UpTone Audio power supply, and the SOtM SMS-200 UltraNeo endpoint renderer was powered by a Keces P3 linear power supply. Both of these power supplies were powered by Shunyata Venom V14D digital power cords plugged into a Shunyata Everest power distributor. All listening evaluations were conducted using my two-channel loudspeaker system described in the addendum.
My journey with my digital streaming front end has been one of discovery, experimentation, and reflection. This journey has taught me new ways to listen to digital content that are much more convenient, flexible, and enlightening, specifically when I use the Radio mode of Roon and Qobuz to discover new music. While I was reasonably satisfied with my first digital front end, when I made the decision to run a direct optical fiber cable connection from the music server to the DAC (see my article on streaming via optical fiber in the May issue of TAS), I needed an Ethernet switch. I bought a generic, consumer-grade TP-link switch to connect the router, music server, and my fiber media converter to the network. After installing it, there were aspects of the presentation that I just didn’t like. Upon first listen, the TP-link switch brought improved detail, but upon extended listening it also brought other attributes were very decidedly un-musical.
What I found was, despite the increased detail, there were “components” to the music and presentation that became increasingly irritating, fatiguing, and over longer listening sessions, quite unpleasant to listen to. I found that I couldn’t listen to the system for more than an hour without wanting to leave the room. It turns out that my Pace network router had four RJ45 ports on it, and I found that by not using the TP-link Ethernet switch (and just using the RJ-45 ports in my router) the presentation was actually pretty good. I would have thought that installing a dedicated Ethernet switch should have sounded better than the ports in a generic Internet router, but every time I put the TP-link switch back into the network (regardless of recording) the system sounded…ghastly. Harsh, strident, bright, irritating, with shrinking of the soundstage and spatiality, a crushing down of dynamic range, and a loss of nuance and gesture in the artists’ singing or playing. By contrast, I found when just using the router’s RJ-45 ports to make connections the sound was acceptably good; I’ll circle back to this observation in the conclusion.
Installing the EtherREGEN Ethernet switch into the system (Fig. 6) brought a significant improvement to the system overall—so much so that it doesn’t even sound like the same system compared to the base configurations. I heard these improvements immediately upon installation, but also noted that, as with almost every other component I’ve owned, audio performance was a bit uneven for the first 200 hours. After that burn-in period, EtherREGEN really smoothed out tonally and opened up spatially with a very low noise floor, providing an exceptionally clean space for the music to emerge into. Most importantly, there was virtually no digital glare, hash, or other nasties that induce listener fatigue. Rather, the presentation was organic, relaxed, and natural to the degree that my digital streaming front end fully rivaled my analog setup.
With EtherREGEN in the system, the Spanish guitar solo fantasia “Malagueña,” composed by legendary guitarist Celedonio Romero, and performed by his son Pepe Romero, had incredibly energetic pacing, a fulsome woody timbre, and finely delineated harmonic structures, as Romero ripped his way through this classic with stunning virtuosity and impeccable technique. There was so much resolution and definition and sheer virtuosity on display that, the first time I heard it, I thought two guitarists, Celedonio and Pepe Romero, were playing…but no, it was just Pepe. As he is the only musician, the stereo image is precisely focused on Romero, but the trailing edges of notes can clearly be heard decaying into the space of the room. Romero moves from a forceful, fast, and dynamic passage to a very beautiful slow, refined, and delicate passage with exquisite phrasing, nuance, and control, and you can hear deeply into the recording, the subtle shadings of timbres and notes flying with perfect articulation off the sound box of his guitar. It was dazzling and incredibly beautiful.
“Falling out of Love” by Mary Gauthier (Mercy Now), is skillfully recorded and mastered with a rich tapestry of sounds, from Gauthier’s gruff voice to her clear, sonorous, and pitch-perfect acoustic guitar in counterpoint to a background harmonica, all fully grounded by a bass drum with a steel guitar layering a plaintive whine on top. Everything comes together as a darkly beautiful song, full of pathos and loss. One of the things that I heard for the first time with EtherREGEN was the gourd instrument, the shakere, driving the rhythm forward, fully resolved but played with just enough restraint not to overpower Gauthier’s voice.
The timeless standard “Cry Me a River” with bassist Ray Brown (Soular Energy) featuring Gene Harris on piano, was beautifully presented with a deep black background against which Brown’s finely articulated string bass provided counterpoint to Harris’ exquisite piano playing. This is one of the most natural and organic-sounding jazz recordings I’ve ever heard, and the clean and quiet background provided by EtherREGEN let the simple beauty of the mastering shine through with subtlety, texture, and naturally rendered tone colors from the piano and bass.
I’ll have to admit I haven’t followed Adele’s studio recordings all that closely—a bit too much commercial overexposure for my sensibilities—but the first time I heard her live recording of “Lovesong” (Adele: Live at the Royal Albert Hall) on my Harbeths, I was smitten. With an exquisitely beautiful vocal performance by Adele, “Lovesong” is exceptionally well-recorded and mastered for a live recording in a large venue. Its starts with the two guitarists, Ben Thomas and Tim Van Der Kull, left and right, and Miles Robertson’s organ laying out the melody and rhythm as her voice comes in over the top, followed by the crispness of Derek Wrights’ drumming and the swell from the string section. The song builds and develops slowly and beautifully as Adele’s incredible voice, finally set free, soars above the string section. It is an absolute knockout. The feel, ambience, and atmosphere of an outstanding live performance in front of a enraptured audience is captured on the recording, and listening to it with EtherREGEN in the system I felt as if I could dive into the experience along with the audience.
After 40 years as an audiophile, I’ve never tired of listening to J.S. Bach’s Concerto for Two Violins, strings, and continuo in D Minor, particularly when Belgian violinist Arthur Grumiaux is playing, as he is on the Decca recording with Les Solistes Romands. Here Grumiaux used his Guanerius violin, Rose, made by Giuseppi Guaneri in 1744. All the classic Guarnerius violin attributes—richness, sonority, timbral weight, and exquisite tone—are in full evidence, and in Grumiaux’s hands the virtuosity is fully resplendent, bright, and energetic, never harsh or strident. Each musician in the chamber orchestra is precisely positioned in the soundstage, and with the inner detail and spatial accuracy that EtherREGEN brought, you could focus in on each instrument in the orchestra and hear it fully resolved against the background of the orchestra as a whole. For example, during the Allegro on Concerto No. 2 for Violin, strings, and continuo in E Major on the same recording, as Grumiaux winds down a solo, I could clearly hear the first cello come in on the right, providing counterpoint to Grumiaux’s violin.
A recent find from Roon’s Radio mode and Qobuz is Yeol Eum Son and The Academy of St Martin in the Fields, playing Mozart piano concertos. In particular, Piano Concerto No. 21 in C Major (“Elvira Madigan”) is a standout. Son’s sweet, light touch and virtuosic technique remind me of Julian Katchen, and give full credit to Mozart’s genius as well as his musical “sense of humor” in this, his most popular piano concerto. The vibe that comes across is that the musicians are having an absolute blast playing this delightful concerto. This is most evident in Movement III: Allegro Vivace Assai; where both Son and the Academy engage in the classical music version of “call and response,” both pianist and orchestra playing with incredible speed, delicacy, and touch in this challenging movement, the woodwinds cleanly interlaced with the piano. The piano’s full dynamic range, power, weight, and timbre were evident, yet resolved and defined, as were the contributions of every musician and instrument in the orchestra. A fabulous recording by Onyx Records, Son and the Academy’s world-famous super-clean, super-light, and super-tight playing is sensational. Just a fantastic recording and performance.
As I mentioned in Listening Impressions, my journey with developing my digital streaming front end and the UpTone Audio EtherREGEN has been an interesting and informative one. I’ve learned that digital streaming, either via a music server or a streaming service (especially Roon), has clear advantages with respect to accessing and listening to digital content. But, as we add more networking devices to take advantage of the convenience, functions, and features, there is a price to paid. Every consumer-grade digital device with el-cheapo clocks (powered by those dastardly switch-mode power supplies) that we add to our streaming network is adding more clock phase noise, leakage current, and noise on the ground planes of our endpoints and DACs, thereby increasing threshold jitter and, concomitantly, timing errors. The reason for this is explained in John Swenson’s White Paper for EtherREGEN referenced above:
“What this clock phase-noise overlay looks like depends upon the clocks and circuits involved. Most ‘residential’ network devices have very cheap oscillators with very high phase noise, so effects from this in the combined clock can be very strong. Therefore, if your endpoint has a very good local clock, the overlay from upstream can be significantly greater than the local clock.”
When I understood this, it became clear why I couldn’t bear to listen to the system after I added the TP-Link Ethernet switch into the configuration: The digital nasties from it were overlaid onto those already present from the other upstream networking devices (router, FMC, and music server). This additional overlay pushed things over the top for me, creating too much of a bad thing. And this is where EtherREGEN rushed in like the cavalry and saved the day, changing the audio experience from one that previously drove me from the room to one that became completely engaging, compelling, and beguiling.
In summary, I consider the UpTone Audio EtherREGEN to be both a breakthrough and a landmark audio product. If you have, or are looking to put together, a music-server-based digital streaming front end, and you need an Ethernet switch in your network, the EtherREGEN deserves serious consideration. Mine has provided me with the convenience, access, and functionality of Roon with curation of content of Qobuz. And, in doing so, it provides a musical experience that’s every bit as engaging and compelling to listen to as my analog system. At only $640, it’s got to be of the biggest value propositions in all of high-end audio, and has my highest possible recommendation.
Specs & Pricing
Circuitry: Fully differential clock circuitry, isolators, and flip-flops
Clock system: Ultra-low jitter/phase noise Crystek CCHD-575 oscillator
Power regulation: Linear Technology LT3045 and LT3042 voltage regulators
Operating Temperature: 48-52° C (118-126° F
Dimensions: 112mm x 110mm x 30mm
Weight: 298 grams (10.5 oz)
Digital sources: Schiit Gungnir Gen 5 USB Multibit DAC, SOtM SMS-200 UltraNeo network bridge, Mac Mini Roon Core Server, Sonore OpticalModule fiber media convertor, UpTone Audio EtherREGEN Ethernet switch, UpTone Audio LPS-1.2 and Keces P3 linear power supplies.
Analog source: Michell Gyro SE turntable, SME V tonearm, Koetsu Urushi Vermilion cartridge, Bob’s Devices Cinemag step-up transformer, Uni-Pro protractor
Phonostage: E.A.R. 324
Preamplifier: First Sound Presence Deluxe 4.0Si Mk-III dual mono preamplifier with dual mono external power supplies.
Power Amplifier: Conrad-Johnson LP70S
Loudspeakers: Harbeth 40th Anniversary 30.2, Dynaudio Contour S3.4 with Esotar 2 tweeters, REL R-305 sub
Cables: Shunyata Research Sigma XC (Everest), Sigma NR V2 (power amp), Alpha NR V2 (preamp), Venom NR-V10 (DAC), Venom V14D Digital (digital components), Black Mamba CX (phonostage) power cables, Shunyata Research Delta and Venom interconnects, Shunyata Research Sigma Ethernet & Alpha USB digital cables and Shunyata Research Delta V2 VTX-Ag speaker cables
A/C Power: Shunyata Research Everest power distributor and SR-Z1 wall outlet
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 ReviewSee all
Further Thoughts: Aesthetix Mimas Integrated Amplifier
The Aesthetix Mimas is one of the finest integrated amps […]
- by Neil Gader
- Apr 15th, 2021