[allhifi]

COMPARISON OF 5 HIGH-END MUSIC SERVERS - Aurender N10, CAD CAT server, TotalDac d1-Server, Auralic Aries and Audiophile Vortex Box





romaz: SUPERB, write-up. Simply brilliant. And refreshingly honest ! Guys like yourself (tech-minded) are often first to dismiss "sound quality improvements/differences" not associated with one's treasured education/understanding in said field. You referenced this in your excellent post.

As a result (of it all) I vote you "Review(er) of the Decade" (perhaps even history -thus far as I plan to succeed you (lol). )

Very honorable mention must go to your fine writing skills -and obviously interpretation/listening skills as well. (I find it astonishing that many in the business (Hi-Fi gear) in capacities ranging from entry-level positions to principals (running/owning multi-million dollar enterprises) are astonishingly tone/pitch deaf. So sad, it's funny actually. I speak from personal observations/interactions from said folk. I'd be happy to share some time.)

In any case, if you have not compared/controlled for AC power variations (and correcting), your ceiling shall extended further skyward still -much more, beyond our atmosphere actually, perhaps even touching the outer-boundaries of heaven itself !

However, I believe it NOT necessary to pay heavenly amounts, mercifully; consider two techniques:

1) Balanced/Symmetrical Power. (i.e. Equi=tech) for analog gear such as preamp and power amplifier.
2) AC-Regeneration: (PS Audio P-3 ) for CD, digital; servers, DAC's)

The dance (to the top) shall become even more delicate, nuanced, satisfying, revelatory.

peter jasz

 

[Marcin_gps]

romaz, too bad you don't have one more PCIe slot available in your music server...

 

[leeg]

Following with interest guys

 

[romaz]

Great review as usual. Can you please let us know how it compares against SOtM 200 ultra?

The SOtM sMS-200ultra is an excellent piece and a good place to start if you decide to do nothing to improve your server as Ethernet provides good (but not perfect) isolation against a noisy server. When well powered (especially by something like a Paul Hynes SR7), I believe it gets you competitive with either the dCS or the Zenith SE and unless you decide to do a direct A/B, it's doubtful you will find yourself wanting. Against the dCS Network Bridge, if you decide to power the sMS-200ultra with something as good as an SR7, from the standpoint of dynamics, you will easily exceed the dCS. I believe this is where the dCS falls short. It sounds flatter and less dimensional in comparison. I suspect the dCS wins, however, with regards to detail resolution although you potentially come very close should you decide to pair the sMS-200ultra with either SOtM's new Reference grade external master clock. Should you go for the SOtM trifecta which includes a reclocked switch, sMS-200ultra and tX-USBultra along with an external master clock and an SR7, my money is on the trifecta but now this setup costs considerably more than the dCS and then you have to contend with a chain of components. Against the Zenith SE, my opinion is the same with regards to SQ although the Zenith SE has a much more robust feature set and can stand on its own.

 

[romaz]

Hi Roy,

Great report, as usual. It's funny that both you and I, having assembled our sCLK-EX based chains, are now curious to see if there are one-box solutions that can match up, or at least come close. You've gone the server route, while I have stayed with the endpoint based trifecta based approach.

Since my approach is still endpoint based, the 2 devices about which I am very curious are the dCS Network Bridge, and the Bricasti M5. While I have had no luck getting my hands on a loaner M5, I was able to audition a dCS NB from my local dealer.

As you know from my post on the CA thread, I had a chance to compare the dCS NB to my trifecta medley. Like you, I found a lot to like in the NB, but due to limited time, and an unfamiliar system, I reserved judgement until I could try out the NB in my system, and @Limniscate 's system. I will be doing that once the USB output firmware is out.

Obviously, the key question for many is - can the one-box NB achieve the SQ of the fiddly "spaghetti" trifecta chain? We shall see!

One area of uncertainty in my mind with the NB (and the M5 too) is regarding the Ethernet and the USB clocks. With the SOtM clock chain, the big revelation, and indeed the raison d'être for the chain, is that clock quality for system things (USB, Ethernet, and mobo) matters just as profoundly as the clocks for the data (the area dCS has been renowned for decades). I have tried to ascertain exactly what clocks dCS uses for their system things, but have not got a good answer. I am hopeful that the clock used here is of a quality approaching, or indeed surpassing the SOtM sCLK-EX, but we don't know.

Your experiment with the Mutec MC-3+ USB will be interesting, but there again, you will only be able to get the Ref 10 "goodness" in through the word clock input of the NB, and it is unclear whether that will lift the quality of the system clocks.

I'll be reporting my results in Nov, because I will only borrow the NB from my dealer after the USB firmware update is in place.

Hi Rajiv,

As you know, as you bridge your LAN ports, you gain the benefits of increased transparency both to the recording (which is a good thing) but also to your upstream components (which can be either a good thing or a bad thing depending on the quality of your upstream components). Once it became clear that the benefits of Ethernet isolation were somehow no longer there to the same extent with LAN bridging, then the usefulness of these Ethernet endpoints were now much less for me. At this point, why not go straight USB and incorporate some of this messy spaghetti chain internally, at least that's how I saw it. For sure, a single box solution has always been more attractive to me from a space and aesthetic standpoint but since SQ is the ultimate goal, well, I had to go with what my ears told me sounded better and thus far, each single box solution I have encountered including both the dCS Network Bridge and the Innuos Zenith SE have come with compromises.

With the dCS, you get this wonderful sense of detail resolution and timing precision but at the expense of dynamics. With the Zenith SE, it's the opposite. With what I have put together, I feel confident in saying that you get a greater measure of both and while I have moved away from your particular spaghetti situation, I still have a spaghetti situation, nonetheless, when you factor in my REF10 external master clock, a separate tX-USBultra and several Paul Hynes SR7s (I have now moved away from the Iso Regen).

Here's something that you and others may find interesting. In the past 24 hours, there have been new and surprising observations. As I transferred these servers yesterday evening to my near-field Voxativ setup in my home office for the very first time, I was surprised to find the Zenith SE preferable to the dCS NB by a large margin. For the past few weeks, with Blu2 and DAVE presently in my large listening room with the Martin Logans, I have been using my Hugo2 to directly drive my Voxativs (without a separate speaker amp) and while it does a very good job in this capacity, compared against DAVE, it had been sounding thinner. I was able to compensate for this to a large degree with a cable change and adjustment of my subwoofer but with the Zenith SE now feeding the Hugo2, I was shocked at how the Zenith SE transformed the Hugo2. This is now a different DAC and it feels like it's been turbocharged. The grip and authority of this DAC as it directly powers my speakers now seem Hulk-like. Transitions carry more weight and bass is so much stronger that I have had to literally turn down my subwoofer. Every track I've played has been several fold more enjoyable compared to when Hugo2 was being fed by my modified Mac Mini powered by my HDPlex and even without Blu Mk2 in this setup, I've found myself hardly missing it as I was simply in awe at what I was hearing. When I swapped the Zenith SE out for the dCS NB, there was this superior technical correctness on display akin to Yo-Yo Ma playing the cello but when compared against the Zenith SE, it was like Yo-Yo Ma was playing bored and uninspired. Lacking was the energy and emotion brought forth by the Zenith. Obviously, these servers are system dependent and so I don't feel that either of these servers/streamers is a lock in every system.

As I returned these servers back to my large listening room this morning where my Martin Logans and my Blu Mk2 and DAVE are presently kept, I was surprised by how much more dynamic the Zenith SE was sounding. In direct A/B testing, the dCS was now sounding flat as a pancake and this simply was not the case during the first 3 days I had these servers. I corresponded with the Zenith distributor this morning and to my surprise, I found out that the unit I received is virtually brand new and only had about 20 hours on it when I first received it. He told me this unit isn't even broken in yet. I know from my own experience that my SR7 didn't start to sound really good until it had at least 100 hours on it. Same with my REF10 Master Clock, my SOtM components and especially my SOtM cables (USB and CAT7). WIth the Zenith continuously playing back music all day, I gave it another good listen this evening and the Zenith has continued to improve with respect to dynamics, however, if I am to be honest, detail resolution is still a tad better with the dCS. At this point, however, if I had to choose one of these units for either my Voxativ room or my Martin Logan room, I would now go with the Zenith SE.

I decided to experiment some more. Since the Zenith SE has an "Ethernet out" port designed specifically for direct connection to a streamer, I decided to connect the dCS NB to this port. I suspect this is essentially a bridged LAN port (the first I have seen thus far in a mass produced Linux-based server) and sure enough, SQ from the dCS NB improved with regards to better clarity as if a thin veil had been removed. I decided to bridge the LAN ports on my custom server and connected the dCS NB to my custom server this way and this same SQ improvement was evident. Configured this way, the gap between the dCS and the Zenith SE has narrowed but given the choice, I think I would still go with the Zenith SE.

From there, I decided to connect my tX-USBultra to the Zenith SE (which is not yet possible to do with the dCS) and so now, this tX-USBultra (which is a USB regenerator) sits between the Zenith SE and my Blu Mk2. This tX-USBultra remains connected to my REF10 master clock and is powered by my SR7. All I can say is that this tX-USBultra transforms the Zenith SE in a very magical way. The wonderful dynamics are still all there but gone is the slight HF harshness I was detecting with the Zenith SE by itself. More importantly, the Zenith SE sounds even more dimensional in both depth and width and detail resolution is now actually better than the dCS NB. Of course, there is the option of pairing the dCS NB with an external master clock and I'm sure it will improve but guess what, with either of these scenarios, you will now find yourself with a spaghetti setup.

 

[romaz]

romaz, too bad you don't have one more PCIe slot available in your music server...

The nice thing about DIY is that you can cherry pick your components. Turnkey servers have their place but then you're forced to fit into someone else's box.

You probably are the only one currently marketing a true audiophile grade LAN card that can be independently powered. If I had one more PCIe slot, I would have loved to have used your card, Marcin.

 

[romaz]

Thank you. I don't disagree with your approach at all. My situation, however, is a bit different. First of all, I live in a rural area with a population of about 2,000 people. Sitting right outside of my house is my local utility company's large isolation transformer and so not surprisingly, I have found my power to be very clean, no matter what time of day. I used to own a Son of Q balanced power supply and I have detailed my experience with it on this thread. Essentially, I didn't find it did much for me that was good and once I removed it and replaced it with a 20A dedicated line using a 50 foot run of inexpensive 10g Romex, my dynamics improved. Earlier this year, I brought in a P5 and a P10 because a dealer insisted I would see improvement and I experienced nothing that was good. Noise floor in my system didn't change at all and I found that while the P10 was less harmful than the P5, even the P10 robbed my system of subtle dynamics. I soon realized that the output impedance of these AC regenerators is not that low and plugged straight into the wall, I found better dynamics, even with small digital devices like my Chord DAVE DAC that consumes barely 20 watts. Regarding my dedicated lines, that is where I found my money best spent. Some people install dedicated circuits to isolate from noise. I chose to install my lines based on Vince Galbo's premise of lowering line resistance:

http://www.msbtechnology.com/faq/how-to-wire-your-house-for-good-power/

Over the past few years, I have tried various conditioners from Audience, Synergistic Research, Nordost, PI Audio Group, Akiko, AudioQuest, Bybee, Furutech, and Shunyata and I found that the "in-series" conditioners seemed to rob dynamics while the "in-parallel" conditioners did not although neither really did anything for my components' noise floor. Part of this is due to the very high PSRR of the power supplies I use. The double regulated Paul Hynes SR7, for example, has a line rejection of more than -150dB which is better line rejection than even the latest Shunyata Denali 6000T. I did find one conditioner, however, that resulted in benefit and that is the Shunyata 2000T. This conditioner doesn't offer much conditioning at all (only -15dB of noise reduction) but what it offers is a feature called QR/BB, a patent pending technology that dramatically reduces any sense of dynamic compression without the use of capacitors. They claim that dynamics are actually improved when amplifiers are directly connected to the Denali 2000T compared to when these amplifiers are directly plugged into the wall. I was a skeptic and while I still don't understand how it works, the improvement is very easily hard as soon as you plug in a component, even a DAC. Combined with the patented magnetic conduction technology used in High Fidelity Cables' line of cabling, I have yet to hear a better power solution. Unfortunately, this setup is not inexpensive and so I have spent more on my power infrastructure than any other aspect of my audio system.

 

[allhifi]

romaz: Another excellent, informative post. Thanks for sharing.

I must return to further AC power evaluations; my experience consistently finds success with Balanced/Symmetrical (i.e Equi=tech) for analog/power gear while AC Power Re-gen (i.e PS P-3) for source -particularly digital components.

pj

 

[allhifi]

COMPARISON OF 5 HIGH-END MUSIC SERVERS - Aurender N10, CAD CAT server, TotalDac d1-Server, Auralic Aries and Audiophile Vortex Box





In the typical digital headphone chain, the chain generally begins with the digital source followed by the DAC, pre-amplifier, amplifier and finally the headphone. It has long been my contention that in this chain, the headphone is the most important piece. In my recent article for Inner Fidelity as a participant of Tyll Hertsens’ Big Sound 2015 (http://www.innerfidelity.com/content/big-sound-2015-participant-report-roy-romaz#B1XGewqDPFt3TyjO.97), one of my concluding statements was “Find the headphone that you love first, everything else comes second.” From here, logic would suggest that the headphone amplifier would be the second most important piece of the chain and then you simply work your way backwards leaving the digital source as the least significant piece. After all, bits are bits, aren't they?

Recent revelations have now caused me to reconsider my position. For the beginner head-fier looking to put together his or her first headphone system, this recommendation of “headphone first” remains the best words of wisdom I can impart. For those of us further along in our head-fi journey, however, I now believe the opposite may be true. For many of us who have been at this hobby for awhile, I find it interesting that most that I speak with own or have owned at least 2 headphones already and seem to always be in search of the next one because for some reason, their collection is always somehow inadequate or no longer engaging. Before long, many of us have spent 3- and even 4-figures amassing more and more headphones as if the portal to audio nirvana lies within that next headphone. Moreover, many of us subject our headphones to tweaks and mods and while these things have their place, many, including myself, have failed to consider that perhaps the headphone is not the problem nor is the headphone amp or DAC. Perhaps we’ve been channeling our resources disproportionately to all the wrong areas and just maybe those bits aren’t just bits.

Some of you will argue my claim but as one of my good friends and fellow head-fier is so keen to say, “if it’s not in the source, then it’s not in the headphone.” Those of you coming from a vinyl background probably have a better sense of this already. If you think about it, an excellent headphone improves nothing before it but an excellent source improves everything after it and in a digital system, most DACs, amplifiers and headphones are resolving enough today to reveal as good a source as you can place before them suggesting that the source is often the limiting factor.

Having built several dozen computers dating back to the days of DOS and as the head of IT in my own business, I’ve taken a crack at building several music servers for myself and others. While I have not gone as far as building a fully tricked out dual box PC, I have benefited like many others from the guidelines posted by Chris Connaker in Computer Audiophile. For years, Chris has championed the idea that a DIY CAPS server could provide a superior listening experience and so for some time, I had assumed that some of the CAPS machines I had built were as good as it got. This belief was supported by my experience with other purpose-built devices including a modified Mac Mini, Moon MiND, W4S Music Server MS-2, Aurender X100, Auralic Aries and an Audiophile VortexBox as these devices at their best were only on par with what I had built. In other words, there was not much that separated these devices.

Well, on April 1, 2014, Chris Connaker, the man responsible for the CAPS server as we know it, proclaimed in his review of the Aurender W20 that the W20 sounded better than any server he had ever used. In his words, “the Aurender W20 makes CAPS servers look like children’s toys with inferior sound” (http://www.computeraudiophile.com/content/573-aurender-w20-review/). On July 19, 2015, DMelby wrote a nice review of the Aurender N10 on Computer Audiophile that essentially said the same thing. In his words, "I found, simply, that the N10 sounds better, by every measure, than my CAPS" (http://www.computeraudiophile.com/f10-music-servers/aurender-n10-review-25192/). By this point, it had become clear there was something to both the Aurender W20 and N10 that were superior to any of the digital sources I had thus far listened to and I knew I had to investigate the N10/W20 further. After scouring other forums and speaking with other head-fiers who have similar tastes, two other music servers piqued my interest, the TotalDac d1-Server and the CAD CAT server.

As I evaluate these three servers, I will disclose that at the present time, I do not own these units or at least I have the option of returning them. I do still presently own an Auralic Aries and Audiophile Vortex Box and so I have decided to include them for comparison. While all reviewers hope to be as unbiased as possible, it has been my observation that most are partial to something they already paid good money for and indeed this is true for me. The burden of proof is always on the challenger and less on the incumbent.

As for equipment used to evaluate these digital sources, here is my equipment chain: digital source -> TotalDac USB cable -> USB Regen with linear PSU -> TotalDac d1-monobloc DAC -> HE-1000 headphone with Silver Spore4 cable by DHC. To better assess soundstage and imaging, listening was also performed through my 2-channel system comprised of my Omega Super 3i monitors powered by a Bantam Gold amplifier. You will notice that a headphone amplifier is not mentioned in my chain and that is because the TotalDac d1-monobloc DAC is capable of driving headphones directly through its balanced XLR outputs and I found the transparency of this direct connection superior to that of my Eddie Current Balancing Act with PX-4 tubes. For evaluating any digital front end, I can think of no better way than this direction connection. I’ve had at least a couple of weeks with each digital source to the extent that I feel comfortable with the sonic character of each. Files used for evaluation include a mix of DSD and PCM as uncompressed FLAC/WAV/AIFF and Tidal streaming.

Ok, so after perhaps the longest introduction in history, here is my take on these digital sources:

Audiophile Vortex Box by Small Green Computer with 1 TB SSD, SOtM USB card and HD Plex linear PSU ($1,954)


This is a well-implemented device that works as advertised and has proven to be reliable and trouble-free over the 18 months that I have owned it. It runs silently, consumes barely 20 watts and runs cool to the touch so I run it 24/7. In addition to being capable of playing all formats up to dual DSD as well as Tidal and Spotify, it is also an excellent CD-ripping device. When not used as a music server, it also very competently fills in as a high-level NAS and so this unit also performed NAS duties for the other units in this review. It runs Linux with a minimal number of process running and so it has a very quiet software footprint. As far as SQ, it does everything well and plays everything without fault. Presentation is balanced and compared to the other devices I have owned in the past including a customized Mac Mini and several Windows boxes I have built, it was at least as good (often better) and proved to be more reliable and convenient and so those other servers were either sold or repurposed.

Auralic Aries with linear PSU ($1,599)




I found a used one for an excellent price earlier this year and so I bought this unit based on so many excellent reviews. This is a very good streamer that is also based on Linux and relies on a NAS or USB hard drive to play music that you own although streaming from NAS has easily been the superior experience for me. I have tried streaming via wi-fi and ethernet and my experience with wi-fi was initially superior although this difference went away when I moved my router to my rack and upgraded my Blue Jeans CAT 6 cable to an audiophile CAT 6 cable made by SOtM. It took me a long time to buy into the benefits of audiophile USB cables and it took even longer to believe there were differences among CAT 6 cables but the proof is in the listening and with blind testing, I was able to pick out the SOtM cable consistently because with the SOtM cable, volume increased by at least a decibel and the soundstage appeared slightly more material. I would rate the sound quality of the Auralic Aries as equivalent to my Audiophile Vortex Box meaning it does everything well. In isolation, you don’t get the sense you are missing anything although I have to say that the music I heard never resulted in the same engagement I have heard with the vinyl systems that certain friends have. Where the Aries surpasses my Vortex Box is in its software interface. It’s Lightning DS app for the iPad is beautifully implemented, intuitive and reliable. I cannot overstate the importance of the software interface for me. From the standpoint of overall user experience, the Aries is superior because of its software interface.


Aurender N10 ($7,999)


The Aurender N10 has Linux underpinnings as well. From what I have been able to gather, aside from the two 2 TB Western Digital Green hard drives, Samsung SSD and RAM, the internals of the N10 including the motherboard is an Aurender design. I had also targeted the W20 although with an MSRP of $16,800, you’re forced to wonder if that player is twice as good. I spent a day with a W20 during my time with Tyll Hertsens at Big Sound 2015 and the W20 played superbly. My initial comments after hearing everything Tyll had on hand was that everything sounded really good. There wasn’t a lemon in the bunch and differences came down more to personal preference than night and day superiority among the various headphones and amplifiers. Initially, I attributed this to the cumulative impact of excellent AC power and TOTL cables but in hindsight, I wonder if the W20 had more to do with that experience. After speaking with my Aurender dealer who had both units on hand and had good personal experience with each, his opinion was that unless you can take advantage of the dual AES/EBU output and external clock option of the W20, the N10 and W20 are sonically equivalent and so we mutually decided he would send me his N10 evaluation unit which was fortunately already well broken in. Within the first 30 seconds of listening to the N10 in my own system, I knew exactly what Chris Connaker and DMelby were saying. It didn’t take a few hours and I didn’t need to run through a long list of music to understand that what I was hearing was superior in EVERY way. I went back to my Aries and Vortex Box and this confirmed what I was hearing. There was no need to even consider blind testing because the differences were not subtle. It was like watching Blu-ray vs DVD or listening to FM vs AM radio. I decided to downgrade my TotalDac d1-monobloc DAC into a d1-dual and then proceeded to substitute it with my Bricasti M1 and with each DAC, the differences were consistent and real. The music I was hearing was just more “material” and felt more dynamic and full-bodied. The timbre and decay of every difficult instrument to reproduce — pianos, violins, cymbals, etc — were excellent. Detail was exquisite, especially the layering of detail. The soundstage was expansive and the sense of space and air was intoxicating. Whether it was a solo vocal in an intimate venue or a grand orchestral performance, the level of engagement to what I was hearing had been multiplied several fold.

TotalDac d1-Server ($4,925)


Compared to the Aurender N10 and CAD CAT server, this is the value proposition of the three. If you are targeting the TotalDac d1-monobloc DAC, since it already incorporates a relocker, the Server option actually adds only another $1,000 and so this option for that DAC is a no brainer. When combined with any TotalDac DAC, there is an obvious synergy present as one was made for the other but the d1-Server plays superbly with even my Bricasti M1. Like the Aurender N10, this server plays at another level altogether against my VortexBox and Aries and again the differences are quite stark. Like the N10, the sound you get is powerful, dynamic and full bodied. It embodies the TotalDac “house sound” in terms of vivid and rich tone with a bloom superior to the N10. The N10 presents a larger soundstage and details are a bit better finessed; however, it is a matter of splitting hairs between these two sources. In direct comparison, these two units are sonically more similar than different meaning that both are superb. The overall user experience with the N10 is better, however. While both are Linux-based, the TotalDac d1-Server is much more like my Vortex Box than the N10. It utilizes MPD on an iPad which I find to be second rate software compared to Aurender’s Conductor app. While SQ from MPD is as good as Aurender’s Conductor, usability and reliability is better with Conductor. MPD is buggy and clunky in comparison and you cannot stream Tidal from MPD, you have to step down to another application called iPeng which unfortunately also results in a step down in SQ. All of these complaints instantly go away if somehow ROON with its superior library skinning abilities and excellent Tidal integration can be ported to and optimized for the TotalDac d1-Server although for now, according to Vincent Brient, TotalDac’s creator, there are no plans.

CAD CAT server with 1TB SSD internal storage ($7,277)


This is a purpose-built PC that runs Windows 8.1 and on face value, I would normally have no interest in such a machine. Having plenty of experience building my own PCs, I was convinced I had already experienced the pinnacle of Windows-based servers and that they did not compete on the same plane as the N10 or TotalDac d1-Server or even Linux-based CAPs devices. It was, in fact, with some reluctance that I agreed to evaluate this unit especially given its asking price but after detailed discussions with @isquirrel who owns this server and after even more detailed discussions with Scott Berry, the creator of the CAT, I decided to give it a go, especially as Scott offers a 30 day money back guarantee.

It turns out this CAT is easily more than the sum of its parts and with this machine, you may actually get more of what you pay for. Scott was quick to point out that the parts used for his CAT are far from off the shelf. He spent nearly 4 years in its design and development. What Scott has found to be one of the biggest issues in digital audio is high frequency noise that results in digital harshness and so he has obsessed about finding ways to eliminate or minimize it using a variety of methods and technologies that he did not feel he could share with me. He did say that the motherboard is heavily modified and incorporates a custom BIOS that has taken years to refine. The motherboard uses 2 oscillators/clocks that receive a highly filtered DC voltage from an outboard linear PSU. There are no motherboard switch mode regulators between the oscillators and external power supply. There is also additional filtering applied as close as physically possible to both oscillators. His USB ports are of his own specification and are fed by a dedicated rail from his external linear PSU. The OS is installed onto a small dedicated SSD with data stored on separate SSDs that have been custom manufactured to his specifications and optimized for music playback. His SSDs cannot be user installed in the same way that his team installs them as they are individually encased, shielded and carefully mounted to guard against resonances. All of his cabling, both data and power, are of his own design after considerable experimentation with different metallurgy and his CAT is handwired to keep lengths to a minimum. The unused ports on his CAT have been deactivated. When run headless, even the integrated video card is deactivated. While he uses a low power quad-core Intel CPU, it has intentionally been dethrottled to run at a specific frequency that he found optimum and just enough to drive necessary processes. As I used FireFox to download my DAC’s ASIO driver, I noticed how sluggish it was and so without question, this CPU has been detuned. The proof of this is the CAT barely consumes 20 watts and runs cool to the touch at all times.

Scott spent nearly 3 years optimizing Windows 8 for audio playback and much of these optimizations are highlighted in the thread he started on Computer Audiophile (http://www.computeraudiophile.com/f11-software/windows-8-audio-optimization-script-16353/). Scott is quick to point out, however, that not all of his optimizations are revealed on this thread and have been reserved for his servers. To keep his OS running reliably and optimally, it has been locked down. No firewall is active and antivirus software and Windows updates are strongly discouraged as they are not felt to be necessary for a music server. In my time with the CAT, I have run it 24/7 and I can attest it has run reliably. Scott further defended his use of Windows 8.1 given its robust networking capabilities and very broad compatibility including the option of running ROON headless. Using an iPad, Android tablet, PC or MAC, I can remotely run ROON on the CAT and the experience has indeed been excellent and superior even to Aurender's Conductor or Auralic's Lightning DS. I have found no better way to organize and play my collection of over 1,000 CDs as well as discover new music as ROON’s Tidal interface is better than even Tidal’s own interface.

What about SQ? I saved this for last. As good as the TotalDac d1-Server and Aurender N10 are, the CAT is better. Not night and day better but better in important ways. The presentation is both smooth and effortless yet incisive at the same time. It embodies the tonal richness of the TotalDac d1-Server but outdoes the TotalDac in terms of tonal clarity. There is a more vivid sense of presence on vocals. It matches the soundstage of the N10 and improves on the N10s ability to render delicate details and subtle nuances. The bass on the CAT is fuller, tighter and better defined. The treble is well extended, maybe even a little more airy than the N10 but both excel here. It also lacks the slight digital edge present in the treble of the N10 that was not obvious except under direct comparison. Tracks that I previously considered bright now seem more tolerable. My gut tells me that the CAT’s well-designed outboard linear PSU with its four separate rails has something to do with this. Having read the recent review by @dan.gheorghe on how a more robustly built power supply transformed his already excellent MSB Analog DAC, without question, the quality of the power supply matters (http://headmania.org/2015/11/01/volent-audio-custom-lps-for-msb-analog-dac/). An even more robust linear PSU is being designed for the CAT by Paul Hynes and it is scary to think how much better this CAT can sound because as of now, it is the best I have heard.

Something else I never considered before, the better the source, the more illuminated everything else sounds after it and the better components seem to receive the greatest illumination. While the Bricasti M1 and TotalDac d1-dual each benefit greatly from these fine sources, the largest beneficiary is the d1-monobloc. It’s as if this was where the biggest choke hold was occurring and set free from confinement of lesser sources, the monoblocs were showing how much higher it could scale compared to the other two. Similarly, with the HE-1000 and a highly resolving cable like the DHC Silver Spore4, I have never heard this headphone sound better. I used to believe that gear like a $25,000 DAC was overkill for headphones because headphones were incapable of scaling to equipment like this and in truth, one of the reasons I decided to take part in this exercise was to see just how high the ceiling is. Well, I have now found how high that ceiling is and it is as high as your source will allow because I now firmly believe for most high-end digital headphone systems, the digital source is the limiting factor. As for the value proposition of this kind of setup, you will have to ponder this for yourselves. For a 2-channel setup, somehow, this is all reasonable but for headphones, many will suggest this is insanity although as many of you can attest, this audiophile hobby of ours is not a sane one. As to what I will be buying for myself, because negotiations are ongoing, I will keep that private. What I will say in closing is this, given the choice of a Mac Mini, TotalDac d1-monobloc and HE-1000 versus an alternate setup consisting of a CAD CAT, Schiit Gungnir Multibit and Sennheiser HD 600, I would choose the latter.

As this comparison was done two years back, any newer updates/rec's you could suggest (ore re-direct me to from earlier replies?)

I've just begun upon my "Streaming" journey and find it exciting (and clearly capable of excellent sound), but there is a plethora of concepts and gear to wade through.

pj

 

[Whazzzup]

Antipodes audio, innous, among others

 

[allhifi]

Thanks Whazzy, but I was hoping for 'romaz' to chime in, or perhaps model designations of your rec's (and if you have any personal experience with any of them)

pj

 

[austinpop]

Hi Rajiv,

As you know, as you bridge your LAN ports, you gain the benefits of increased transparency both to the recording (which is a good thing) but also to your upstream components (which can be either a good thing or a bad thing depending on the quality of your upstream components). Once it became clear that the benefits of Ethernet isolation were somehow no longer there to the same extent with LAN bridging, then the usefulness of these Ethernet endpoints were now much less for me. At this point, why not go straight USB and incorporate some of this messy spaghetti chain internally, at least that's how I saw it. For sure, a single box solution has always been more attractive to me from a space and aesthetic standpoint but since SQ is the ultimate goal, well, I had to go with what my ears told me sounded better and thus far, each single box solution I have encountered including both the dCS Network Bridge and the Innuos Zenith SE have come with compromises.

With the dCS, you get this wonderful sense of detail resolution and timing precision but at the expense of dynamics. With the Zenith SE, it's the opposite. With what I have put together, I feel confident in saying that you get a greater measure of both and while I have moved away from your particular spaghetti situation, I still have a spaghetti situation, nonetheless, when you factor in my REF10 external master clock, a separate tX-USBultra and several Paul Hynes SR7s (I have now moved away from the Iso Regen).

Here's something that you and others may find interesting. In the past 24 hours, there have been new and surprising observations. As I transferred these servers yesterday evening to my near-field Voxativ setup in my home office for the very first time, I was surprised to find the Zenith SE preferable to the dCS NB by a large margin. For the past few weeks, with Blu2 and DAVE presently in my large listening room with the Martin Logans, I have been using my Hugo2 to directly drive my Voxativs (without a separate speaker amp) and while it does a very good job in this capacity, compared against DAVE, it had been sounding thinner. I was able to compensate for this to a large degree with a cable change and adjustment of my subwoofer but with the Zenith SE now feeding the Hugo2, I was shocked at how the Zenith SE transformed the Hugo2. This is now a different DAC and it feels like it's been turbocharged. The grip and authority of this DAC as it directly powers my speakers now seem Hulk-like. Transitions carry more weight and bass is so much stronger that I have had to literally turn down my subwoofer. Every track I've played has been several fold more enjoyable compared to when Hugo2 was being fed by my modified Mac Mini powered by my HDPlex and even without Blu Mk2 in this setup, I've found myself hardly missing it as I was simply in awe at what I was hearing. When I swapped the Zenith SE out for the dCS NB, there was this superior technical correctness on display akin to Yo-Yo Ma playing the cello but when compared against the Zenith SE, it was like Yo-Yo Ma was playing bored and uninspired. Lacking was the energy and emotion brought forth by the Zenith. Obviously, these servers are system dependent and so I don't feel that either of these servers/streamers is a lock in every system.

As I returned these servers back to my large listening room this morning where my Martin Logans and my Blu Mk2 and DAVE are presently kept, I was surprised by how much more dynamic the Zenith SE was sounding. In direct A/B testing, the dCS was now sounding flat as a pancake and this simply was not the case during the first 3 days I had these servers. I corresponded with the Zenith distributor this morning and to my surprise, I found out that the unit I received is virtually brand new and only had about 20 hours on it when I first received it. He told me this unit isn't even broken in yet. I know from my own experience that my SR7 didn't start to sound really good until it had at least 100 hours on it. Same with my REF10 Master Clock, my SOtM components and especially my SOtM cables (USB and CAT7). WIth the Zenith continuously playing back music all day, I gave it another good listen this evening and the Zenith has continued to improve with respect to dynamics, however, if I am to be honest, detail resolution is still a tad better with the dCS. At this point, however, if I had to choose one of these units for either my Voxativ room or my Martin Logan room, I would now go with the Zenith SE.

I decided to experiment some more. Since the Zenith SE has an "Ethernet out" port designed specifically for direct connection to a streamer, I decided to connect the dCS NB to this port. I suspect this is essentially a bridged LAN port (the first I have seen thus far in a mass produced Linux-based server) and sure enough, SQ from the dCS NB improved with regards to better clarity as if a thin veil had been removed. I decided to bridge the LAN ports on my custom server and connected the dCS NB to my custom server this way and this same SQ improvement was evident. Configured this way, the gap between the dCS and the Zenith SE has narrowed but given the choice, I think I would still go with the Zenith SE.

From there, I decided to connect my tX-USBultra to the Zenith SE (which is not yet possible to do with the dCS) and so now, this tX-USBultra (which is a USB regenerator) sits between the Zenith SE and my Blu Mk2. This tX-USBultra remains connected to my REF10 master clock and is powered by my SR7. All I can say is that this tX-USBultra transforms the Zenith SE in a very magical way. The wonderful dynamics are still all there but gone is the slight HF harshness I was detecting with the Zenith SE by itself. More importantly, the Zenith SE sounds even more dimensional in both depth and width and detail resolution is now actually better than the dCS NB. Of course, there is the option of pairing the dCS NB with an external master clock and I'm sure it will improve but guess what, with either of these scenarios, you will now find yourself with a spaghetti setup.

Hi Roy,

Fascinating findings! But just to be clear, after all the dust has settled: your custom server, with end to end clocking and SR7 PSUs, is still sounding superior to the Zenith SE, correct?

Also - what have you been able to discern about the internals of the Zenith SE, especially when it comes to the things that matter most - PSUs and clocks?

I would like to reiterate my deep appreciation, that I'm sure is shared by many here and over on CA, for your tireless efforts in seeking the optimal SQ in this confusing space. Your insights are valuable for so many reasons:

1. you don't indulge in the dogma about direct-connect vs. network-streaming; instead, you've explored both sides, using your ears as your guide, and clearly articulated your rationales and findings. That is pure gold for those seeking unbiased audio advice and knowledge.
2. So many - myself included - have tried your recommendations, and raised the SQ of our systems tremendously. Ultimately that is the true test for whether your findings can be trusted. You are one of the most trusted audio reviewers on the Internet today.
3. conventional reviewers are often beholden to their advertisers, whereas you are using your own resources to evaluate these components. In my estimation - that makes you far more trustworthy.
4. Having met you recently at RMAF, I can speak to a side of you that most people do not see. Roy is the quintessential gentleman - polite, charming, and incredibly generous when it comes to sharing his vast knowledge.

Thank you for your contributions.

 

[STR-1]

Antipodes audio, innous, among others

I would add Melco to that list, even though its servers don't do Roon. I hope soon to have both the Innuos Zenith SE and Melco N1ZS/2 on loan for home shoot-out. You'll have seen from recent posts that the Zenith SE is highly thought off, but there are some who would say that even the (2nd generation) base Melco is better than the standard Zenith. That's enough for me to give both flagships a listen.

 

[austinpop]

Once it became clear that the benefits of Ethernet isolation were somehow no longer there to the same extent with LAN bridging, then the usefulness of these Ethernet endpoints were now much less for me.

On this point: I encourage folks to look at the recent postings by John Swenson on CA - across several threads - on his findings with regard to SMPS grounding, and Ethernet switches. With regard to the latter, it appears that the degree to which they let noise on the data line has to do with the particular "magnetics" used. At this point, he just has empirical findings that Netgear FS105/108 switches appear to block leakage far better than any other switches he's tried.

I would expect that over time, his efforts will lead to a better understanding of how to effectively isolate Ethernet better.

It's an exciting time to be a computer audiophile.

 

[greenmountain]

I could compare the complete series of the Innuos range at this year's Highend in Munich and the Zenith SE was very obviously the best (sadly also much more expense). If one wants to save some bucks the Zen mini II was pretty good.
Has anyone any opinion about Nativ Vita https://www.nativsound.com/en/music-server? I really like its design with integrated touchscreen, wall mountable and one can get an accessary CD ripper. It has quite a variety of output options except no ethernet out (which I hope will be my best option on my Devialet 250Pro after the promised core board gets installed).