A novel way to massively improve the SQ of computer audio streaming

[Johnseye]

7 minutes ago, KingRex said:

The Topaz did a very good job with my Conrad Johnson Premier 140.  It got rid of a lot of mechanical noise.  It made the front end equipment shift in a way I did not want to go so I only plugged the amp into the receptacle fed by it.  I sold that amp and use a set of Altec Lansing 1570B modded by Tom Tutay with Transition Audio Design.  The Topaz shifts the sound for those amps in an undesirable direction.   I wrote in another post how I view power conditioning as hit and miss.  You have to listen and see what it does for you.  With the CJ, it took Emmanuel Go with First Sound Audio to teach me how to hear the slight veil it added.  That veil was so small compared to the benefits, I would still be using it if it were not for my purchasing different amps.  The new amps show a pronounced suppression of way too much ??????  It just does not sound right.  The Altec responded very well to the Akiko Corelli.  So did all my front end gear.  I had the Tripp Lite first.  It worked well with the CJ but I did not feel it had the headroom so I went bigger with the 2.4kva.  Definitely freed up the air and ease of play.  Lot more headroom.  Its a very good transformer and will provide a great benefit in some circumstances.  

 

It's interesting that a device that is supposed to clean the line of noise has a negative impact on some devices.  It makes me wonder why.  I'm going to have to read some more about isolation transformers.

[Johnseye]

17 minutes ago, Speedskater said:

If you are daisy chaining dual receptacles that would be true. But if you are using a pig-tail system (from a large feeder), then each dual receptacle gets it's own pig-tails.

 

Considering the emphasis on wire gauge is there any degradation by using pigtails, stepping down the gauge with an adapter, using spades, or going through a jumpered isolation transformer? 

[Johnseye]

5 minutes ago, WuNgUn said:

For a short run like a pigtail (6 or 8"), I'd say no... As long as you're still using 14 or 12awg.

Voltage drop would be negligible on short runs as long as you calculate your loading with respect to awg. 

 

Which brings up another question...

If youre running 8awg was it...? to the breaker panel, how are you protecting that conductor? I don't even know if that would fit in a 15 or 20a breaker. 

 

I'm running 10 awg with 20amp breakers.  No issue there.  My main concern was when I'm coming off the wire with a spade whether I'm choking the current.  Also the spades aren't copper, or maybe they're tin plated copper.  A twisted pigtail gets better contact than a crimped spade.  It's probably negligible but when you go to the trouble of rewiring you don't want to do anything which would negatively impact the work. 

 

[Johnseye]

29 minutes ago, KingRex said:

Why are you using a spade.  Get rid of them. They are not helping the sound. Turn of the breaker, cut off the spade, strip the wire back about 3/4 inch and put them under the clamp. Tighten them pretty firm. Once you have them set strait and ready to screw back on the box, check the torque.   Didn't you say you have Porter Ports or Furutech.  They have a clamp, easy to do. You may have to put a twist on the ground. Not sure if you have a screw only or clamp there too.

 

I will try and post a couple pix on how to wire receptacles with pig tails or using large gauge wire. 

 

Thanks, will do.  The electrician used spades because the 10 gauge is solid and difficult to bend around the screws.  I'll work on them myself.

[Johnseye]

1 hour ago, KingRex said:

Johnseye, what has a spade connection?   The receptacle or the termination point in what appears to be a fused disconnect from what One And A Half is showing. I thought you were messing with a receptacle.  I would not advise getting into your panel. Way to dangerous. 

 

That wasn't my picture One and a Half was showing.  I'm referring to the receptacle and at the isolation transformer.

[Johnseye]

1 hour ago, KingRex said:

Then cut the spades off.  They will impact performance.  Do you have (2) duplex receptacles in a wall box or just one.  See photo.  I want to send a picture demonstrating how I would wire them.

 

Can you pull, the face off your transformer and send a picture.  Need to see that too.  This one is hard wired with wire nuts. My other had a terminal block. I want to look at the screws.  See photo. 

 

 

53 minutes ago, KingRex said:

This is how I wire a quad.  I am only showing the white and ground for the example. Notice the jumper on the neutral. Notice the ground is open to the clockwise rotation of the screw so it will clamp tight when you torque the screw. Also notice on the ground I pinched the wire together before tightening it. Most of the better receptacles have clamps on the side of the ground just like on the neutral and hot. If I were working with a number six I would pigtail off with number tens. This configuration is called a pigtail.  The tan wirenut is good for 3x#10 wires. Red is easier for a novice. 

 

Here's how I twisted the hot, neutral and ground.  I've got about 190 ft between the two circuits.  I'm looking at about $3k worth of work once the drywall repairs are done, plus the transformer which was another $350.  In the end, it was worth it but I can think of a lot of other things I'd rather spend $3k on.

 

 

Some outlets are duplex, some are quads.  Here's an example of a duplex (cryo'd Hubble for what it's worth).  This was done by the electrician.  I don't like the size of the nuts, they're too big and take up more space inside the box than needed.  Notice the spade on the ground.  The outlet looks close to the floor because I built a riser for a second tier of theater seats.

 

 

 

Here's the H side of the transformer.  It is in immaculate condition.  The X side is a bit more complicated and where I struggled with grounding the neutral.  I actually did the wiring on this.  There's a slight hum when sitting on the wood.  Much less on carpet.  I've got it sitting in a closet in the back end of the room.  Out of sight, sound and mind.  This sits between the breaker and 2 outlets, one hosting my amp and sub in the front of the room and the other hosting all my two channel gear off a single Shunyata PDU.

 

I introduced it at the same time as all the wiring and some other changes.  I plan to connect all my gear using an extension cord to the other dedicated circuit I had installed.  Then I can compare with and without.  What I believe I am hearing, with this is dead, black quiet where it should be, revealing sounds, or more of the music I couldn't hear before.  Recorded sounds like breath and guitar amp hum is much more noticeable.

 

[Johnseye]

8 hours ago, KingRex said:

I don't get it.  The sCLK-EX appears to be a clock.  What does it have to do with connecting Ethernet to my server.  Yes my internal data source is outstanding.  I am absolutely astounded at the SQ with Win Server 16, AO and HQ player.  Incredibly musical and engaging.

  And yes.  Simple is the key.  My wife needs to turn the  monitor on and select an artist to listen too.  There can be no selecting files or pathways to get or send a feed.  Right now its all there.  Very easy to navigate. 

When you say router based internet stream, do you mean th JCAT card is the router?  Thanks

 

8 hours ago, ElviaCaprice said:

One sCLK-EX point would be applied to your LAN/ethernet (both LAN if two) port for clocking of that ethernet stream being received, if not using LAN, skip this clock.  Then you would have 3 points left to apply to other clocks.  I would suggest the system mobo clock, add a tXUSBexp PCIe card and replace clock, mobo usb or external component.  

 

No changes to OS or software necessary.  Keep what you have.  JCAT is not needed.

 

It's my understanding that the sCLK-EX also provides noise filtering.  I don't know the details, only a brief comment from May at SOtM.

 

[Johnseye]

28 minutes ago, Lebouwsky said:

Sorry for this perhaps silly question, but how does one power an ssd externally and which cable is needed?

 

It's not a silly question.  I obtained my sata power adapter from SOtM.  Note the black and red cable between the sata filter and the external device PC slot cover in the bottom right of this picture.

 

 

 

[Johnseye]

27 minutes ago, Lebouwsky said:

Thank you for sharing this picture, very valuable, it is the route I want to take, mostly because of you and Roy’s observations. Correct me if I’m wrong, but this is what I see:

* 1 dc cabling lay-out for tx-usbexp, tx-usbultra and sclk

* 4 points coming out of sclk(ethernet, motherboard, tx-usbultra and tx-usbexp

* txusbexp directly in (mini) PCI lane.

* sata data cable going from mb to sata filter

* modified sata power cable going to sata filter

 

Note sure where the dc in goes for the tx-usbexp, tx-usbultra and sclk

 

I guess an usb cable will connect the tx-usbexp to the tx-usb ultra and from there to the dac, but not shown in picture and all goes into 1 case right? (except ultra ofcourse)

 

 

 

My current power application differs slightly from this picture.  It was the best pic I had to show the sata power.  This layout is completely viable, I just eliminated the internal power source from the motherboard.

 

9v Power comes in from an input on the tX-USBexp.  From that I feed the sCLK-EX which then feeds the tX-USBultra via the blue and white cable.

 

What's different from the picture is imagine that blue and white cable running to the left plugs directly into the tX-USBexp.

 

Some other things I changed recently were shorter RF leads once I knew how much length I needed inside the mobo, and a single direct connection via RF lead from the sCLK-EX to the input on the tX-USBultra.  Total length is 6" I think.  That requires the tX-USBultra to be back to back with the PC, but it works for me.  I like short cables.

 

There is a USB cbale from the tX-USBexp to the tX-USBultra, then another to the DAC.  All goes into one case.  I could probably pull out what's in the tX-USBultra and put that in the case but it

would need some customization and not worth it to me....yet.

 

Here's a picture from before I made my most recent changes with the shorter RF leads.

 

 

[Johnseye]

1 hour ago, tims said:

Hi @Johnseye

 

From your description of your power supply chain: tX-USBexp > sCLK-EX > txUSBultra, I'm curious if you tried powering the txUSBultra with it's own (good quality PS) and not from the sCLK-EX output.  If so, was there any difference SQ wise?

 

The reason I ask is that I've had my sotm sMS-1000U player upgraded to an SQ version (with a tx-USBexp and sCLK-EX) and I also got sotm to use one of the sCLK-EX clock points to feed an external tx-USBhubEX that I hope makes it essentially a tx-USBultra.

If I can somehow power the tx-USBhubEX with the output of the sCLK-EX then this will save me a substantial amount buying a LPS-1 or something similar to power the tx-USBhubEX.

 

Cheers.

 

The tX-USBexp, sCLK-EX and tX-USBultra all get powered by my Paul Hynes SR7 this way.  It's not going to get better.  Additionally, the sCLK-EX has some of its own filtering so powering the tX-USBultra through the sCLK-EX is optimal and what you get when you buy a standalone version and not broken out like I have.

 

That said, I could power it from an LPS-1 or now LPS-1.2 but I'd have to buy a new one because it's being used to power my SSD right now.

 

You should be able to power the USBhub from the sCLK-EX.  You can either go the route I did or probably connect the blue and white cable directly to the USBhub's board.  I think there's a power adapter on it for the cable to connect.

 

57 minutes ago, rickca said:

OK that assembly is exactly what I need.  Where can I order this online so I can power an internal SSD with external 5V LPS?  I need a DC jack for an LPS-1.  I haven't been able to find something like this googling.

 

Lots of people seem to do this.  I don't have DIY skills to make it myself.  Thanks.

 

I would either contact May at SOtM or Crux Audio who are SOtM's US distributor.  You may want to source a cheaper version though.  Mine was $60 and they used stranded copper.  They also sell a different copper version and a silver version for more.

[Johnseye]

5 minutes ago, Lebouwsky said:

Might this be an option:

https://www.moddiy.com/products/SATA-to-Gridseed-Power-Plug-(2.1mm-DC-Barrel-Plug).html

 

You'd need a female adapter but much more affordable than what I did.

[Johnseye]

4 hours ago, tims said:

@JohnseyeThanks for this information.

Regarding powering the tx-USBexp the options are: (from an earlier post of yours); internally via the 4 pin IDE Molex at 12v or externally at 9v maximum.  The only thing I don't get is that reading the tx-USBexp operating instructions it states:

"If the tX-USBexp get powered by the external power input, make sure not to cut off the external power

while PC is powered on. If the external power is suddenly cut off, the internal parts of tX-USBexp or PC
will be damaged.....If you connect the IDE power cable inside of PC to the IDE power input connector of tX-USBexp while the tX-USBexp get powered by the external power input, it can prevent the damage of the internal parts of tX-USBexp or PC when the external power suddenly cut off."

(tx-USBexp Operating Instructions)

 

Is this your understanding?  I don't get how you can power the tx-USBexp externally at 9v and also internally at 12v via the 4 pin Molex connector at the same time?

 

Cheers

 

 

It's one or the other.  Fortunately my tX-USBultra is powered from the same source as the motherboard, which is the 2 rail SR7.  I won't power the card off without the PC powering off.

[Johnseye]

16 hours ago, austinpop said:

Review of the Innuos Zenith MkII SE

 

Here - finally! - is my promised review of the Zenith MkII SE, referred to in the rest of this review simply as the SE. I know it feels like I've strung you guys along. I would have written this much earlier, had it not been for the untimely failure of my tX-USBultra, just when I was getting ready to do comparative listening at the 200 hour burn-in mark on my SE. Well, I have received my tX-USBultra back (a saga I posted about separately), which finally enabled me to complete this review. A side benefit of this delay is that my SE is truly burned-in now, with well over 300 hours of play time.

 

The SE came to my attention upon reading Roy's review on head-fi. After his public posts on the matter, he and I stayed in touch and he conveyed how impressed he was with it in his system. Finally, when a group of friends and I decided to do a private group buy, the die was cast. In due course, the SE arrived, and assumed a place in my system.

 

I'm not going to waste time and space here describing what the SE is and its feature set. If you haven't already, go read Roy's review on head-fi and Darko's review on his site. The SE is a standalone music player, that can store music locally, and serve them to a DAC directly via USB. For Roon users like me, the SE can function both as a Roon Core, but also as a Roon Bridge. If used as a bridge, the SE functions as a rather expensive streamer/renderer. In Roon Core mode, the box also provides a bridged streamer output, so you can drive an existing streamer. So it's a very flexible box! It also supports other modes like UPnP, Sonos, etc. I did not experiment with these.

 

As you know, my journey to SQ optimization has led me to my current spaghetti solution based on the SOtM trifecta. As a result, my perspective in evaluating the SE is uniquely mine. It's based on the question - how can the SE improve the overall SQ of my audio chain? Can it do so while simplifying my chain? And does it? This review is all about comparative performance. There are enough reports out there that suggest that on its own merit, the SE sounds wonderful. So my focus here is comparison with my trifecta chain.

 

Also note - we are at a fairly hefty price point here, and definitely in the diminishing returns region of the price-performance curve. I have no illusions on this score, nor am I even suggesting that this kind of expenditure is necessary for good SQ. 

 

Disclaimer

 

I've had people contact me wanting to know if I was affiliated with, or paid by, one company of the other. Let me set the record straight, once and for all. I only review, and plan to continue to only review, gear that interests me for my own system. All the gear is either purchased by me, or lent to me by friends and CA'ers - NOT by manufacturers. Like all of us, I hate paying retail, so will often buy used or discounted products, but not for review considerations. To be honest, I'm pretty small fry, and I prefer it that way! The main reason I write here is because a) I enjoy writing, and b) I feel too few reviews - especially from mainstream publications - take an unflinching, comparative approach, which is what I, and most of the audiophiles I know, really want to know.

 

OK - back to the interesting parts.

 

My Current Chain

 

Let's review the chain I have been using prior to the SE's arrival. The notation I'll use is to indicate the PSU in parentheses, including whether its output is shunted to ground via a JSGT (John Swenson Ground Tweak).  For all the LPS-1s in my chain, the energizing Meanwell SMPSes are all JSGT shunted. Also, unless otherwise indicated, upstream of the chains below is:

• Netgear CM600i modem (HDPlex LPS) > TP-Link Archer C7 router (HDPlex LPS) > Netgear GS105 switch (Breeze LPS, JSGT)
• TP-Link Archer C7 router (HDPlex LPS) > Synology DS916+ NAS (SMPS) 

The modem, router, and NAS, are in a different room, and on a different AC circuit. The Netgear switch is in my listening room, and immediately precedes the chains below. All my audio gear is on a dedicated AC circuit, using the 6 AWG wire and outlet sourced from Jim Weil, that I described in a post a few days ago.

 

Here is my current chain:

• sCLK-EX-modded switch (LPS-1) > sCLK-EX-modded sMS-200 (LPS-1) > ISO-Regen (LPS-1) > tX-USBultra (Hynes SR-4) > Codex
• Cybershaft OP-14 10 MHz reference clock feeding the sCLK-EX board in the tX-USBultra
• This clock topology essentially delivers OP-14 quality to my switch, sMS, and tX, and has really been the jewel in my audio crown, as it were.

SE Functional Quirks

 

As I previously posted, the SE's USB controller does not play nice with the USB controller on my Ayre Codex. So I was never able to run the simple chain:

• Zenith SE > Codex

In trouble shooting, I had several emails back and forth with Nuno of Innuos and Ariel and Ryan at Ayre. While this issue is still not resolved for the Codex, we did establish that it seems to be related to the fact that the Codex USB stage is bus-powered. Thanks to another loan of the QX-5 Twenty by my local dealer, which uses the exact same USB controller chipset as the Codex, but is internally powered, I established that this chain works fine:

• Zenith SE > QX-5 Twenty

Nuno did mention that another workaround is to use the iFi USB iPurifier2 between the SE and the Codex. Mark Sossa, the US Innuos distributor kindly lent me one to try. Sure enough, this does function correctly:

• Zenith SE > iPurifier2 > Codex

Usability

 

While the SE's UI is not geared to power users and tweakers like me, I have to appreciate its elegance and simplicity. For computer- or technology-phobes, this thing is a dream. One of the cool features is the CD ripping tool. This isn't something I expect to use much, as the bulk of my CDs were ripped more than a decade ago. Upon Roy's recommendation, I did try re-ripping a favorite CD, and the experience could not be nicer. Roy felt that the rips from the SE actually sounded a little better to him, but I didn't find that to be the case. Still, for those needing this feature, it's nifty.

 

Sonic Signatures

 

Before I get into comparisons, let me talk about the sonic signatures of both my trifecta and the SE. In the rest of the review, I'll refer to these signatures rather than enumerate them again and again. 

• Current chain based on SoTM Trifecta: As I've described before, the net effect of clock and PSU improvements - starting with the sCLK-EX, the use of LPS-1 and SR-4, and culminating in the addition of the Cybershaft OP-14 - has been in the following areas:
  • Tonal Smoothness, lack of digital glare
  • Sweet midrange
  • Huge and precise image
  • Air - excellent separation of instruments
  • Texture and dimensionality - instruments sound real, and have a palpable dimension to them
  • Resolving - fine details are exquisitely rendered. Brushes and high hat strokes come across very distinctly. A string section sounds like distinct violins (for example) playing together, rather than one homogenous violin, if that makes any sense.
  • As such, this configuration sets a very high bar to surpass!
     
• Zenith SE: the addition of the SE to a chain yields substantial SQ benefits. These are completely consistent with Roy's and Darko's observations, but I'll describe them in my words:
  • Dynamism: it feels like the music has gained some dynamic range, but it's more that dynamic range is is more realistically rendered. This is actually more apparent if you turn down the volume when making comparisons. 
  • Calmness and ease: related to the above, there is a sense of relaxation and ease to the sound, that I attribute to a lower noise floor
  • Richer and smoother tonality, especially the treble
  • Density, solidity: music had more weight and body to it. Instruments seemed more palpable and real.
  • Improved bass

Findings

 

Let me first dispense with the most obvious and financially most attractive comparison: could the Zenith SE standalone replace my spaghetti chain? If so, I could sell my entire trifecta, and recoup most or all of what I paid for the SE. Sadly - the answer is an emphatic NO. While the SE bettered the trifecta in terms of the traits described above, especially dynamism and weight, there was also a profound sense of loss for the air, dimensionality, and resolution that is so characteristic of all the clock improvements. This was not surprising, nor unexpected, as it's what Roy found in his evaluation too.

 

This would be a good time to take a detour to the topic of axes (or dimensions) of optimization. We have ample evidence now that SQ improvements can be gained by making extreme optimization on both the clocking axis AND the PSU axis. There are other axes too: for example, RF/EMI shielding, and mechanical/vibration control. If you've read about the SE, you know that one of the key things that distinguishes the SE from its lower-priced brethren is the use of very high-quality, bespoke PSUs designed by Sean Jacobs, another British boutique electronics designer. First Paul Hynes, now Sean Jacobs. Must be something in the water there! Or the whisky.  

 

The SE and the trifecta chains are perfect examples of systems that have been optimized along these two axes. My trifecta chain is highly clock-optimized, but while I am using excellent PSUs, it could probably sound way better if I used 4+ rails of an SR-7 instead. In contrast, the SE is highly PSU-optimized, but by Nuno's admission in discussion with Roy and me, not particularly optimized on the clock dimension.

 

With this backdrop, I went on a similar path as Roy described in his head-fi posts: could I combine the SE with elements of my trifecta to achieve a new high-water mark for my system's SQ? And what is the cost/benefit tradeoff of the resulting chain?

 

Note: this post is long enough already, so I will not lengthen it by enumerating all the interesting experiments I tried. At different points in the last 3 weeks, I've had loaners of the Ayre QX-5 and Eric's sMS-200ultra and dX-USB HD Ultra, that have allowed me to try all kinds of permutations and combinations. I'll cut to the chase and list only the most interesting findings.

 

Disclaimer: Keep in mind that all the adjectives and qualifiers I use below are relative. Take it as given that each configuration by itself sounds incredible, and without the other to compare, I would be very pleased with the SQ I was hearing. But for comparisons, it becomes necessary to use positive or negative terms to convey relative merits.

 

Finding 1: The best Zenith SE chain outperforms my current reference chain

 

This is the SE-based chain that sounded best in my system:

• sCLK-EX-modded switch (LPS-1) > Zenith SE > tX-USBultra (Hynes SR-4) > Codex
• Cybershaft OP-14 10 MHz reference clock feeding the sCLK-EX board in the tX-USBultra

You'll notice that this is still "spaghetti" but not as complex as my current chain. Why the switch, you ask, since the SE can store music locally? See the next finding.

 

This is the same combination that Roy has settled on in his system too. As good as the SE sounds by itself, the addition of the tX-USBultra, reference clocked by the Cybershaft, elevates SQ to an absolutely magical level. My system has never sounded this good! This combination takes the dynamism, the density, and the rich and smooth tonality of the SE, and adds in the resolution, micro-details, and texture of the clocking enhancements. 

 

If I'm honest, I still find my current reference chain to be a tad more detailed and resolving. This is where system synergy comes in. As I've mentioned in my past posts, I have always struggled with a slight thinness to the sound of my chain. This starts with my headphone setup, which is extremely resolving. Part of this is just inherent in headphone listening compared to speakers. Add to that my HD800's which, while they've been modded to mitigate the infamous 6kHz peak, are still unforgiving of any treble harshness or imbalance.

 

Compared to the SE chain, my current chain sounds a little constricted, and seems to have more treble energy. Switching to the SE chain delivers that "aahh" - a sense of calm and relaxation that one gets from reduced treble glare and hash. The sound is more dynamic, weighty, and richer. More importantly, voices sound more natural.

 

Having said all that, it isn't by any means a night and day difference. On some well-recorded and smooth-sounding tracks, it's harder to tell the 2 chains apart. The more inherently brighter the tonal palette of the recording, the more the difference is noticeable. To illustrate this - to myself, during listening - I fired up Kate Bush's Wuthering Heights. I know - I have dated musical tastes, but this is a song that will have you ripping the 'phones off your head on a sub-par system.  In this comparison, the constriction and brightness of the trifecta relative to the SE was immediately noticeable.

 

Ultimately though, the SE-based chain extends the SQ of my system to a new high-watermark. Is it worth the price? That is highly subjective. For my situation and my ears, the answer is yes, although I still have a few days left in the 30-day return window. I'll talk about this more in the summary.

 

One final note: Knowing the lust and envy that would result, I did also listen to the SE chain driving the QX-5. OMFG! This was a sublime chain. I know in the past I've said that I find DAC SQ differences to be rather small, compared to what I have heard in the digital chain upstream of the DAC. With the SE in the chain, it felt like the DAC difference between the Codex and the QX-5 was more pronounced. I'll talk about this more in finding 4.

 

Finding 2: The reclocked switch minimizes the gap between music on local SSD vs. NAS

 

Since I already have an sCLK-EX-modded switch available, along with an available clock output on my tX-USBultra, it seemed a no-brainer to try this in the chain. Without the reclocked switch, the difference between music stored locally on the SE sounded significantly better than when retrieved from my NAS on the network. Adding in the reclocked switch, the gap narrowed considerably. I have to listen very carefully now to register the slight improvement provided by local files. And of course, the reclocked switch makes a huge improvement for Tidal and other streams.

 

This is really important to me for flexibility. My music lives on my NAS for multiple reasons - it is accessible from multiple systems and devices, and I have a backup strategy in place. Putting music files locally on the SE would then require me to keep my SE copy in sync with my NAS copy. It's not an insurmountable problem, but it is a bit messy. Nuno has indicated they are looking into proving a sync tool within the SE to address this use case. I hope they do.

 

Finding 3: The Zenith SE internal player outperforms Roon

 

This is one of those ironic findings. On the one hand, kudos to the Innuos team for optimizing their internal SqueezeServer player. But as a committed Roon user, this finding actually annoyed me. I have no interest in switching back from Roon to a Squeezebox ecosystem. I still have iPeng on my iDevices from when I used to own a Squeezebox Touch, but I've moved on.

 

So yes - the internal player, controlled by iPeng, does sound better than Roon Core on the SE. From what I understand, much of the improvement comes from allocating a generous memory cache. To be fair to Innuos, they were able to make these improvements since the codebase is open source (I think), while the Roon code is opaque to them.

 

Still, as a Roon user, I would like to see this gap eliminated, and I hope Innuos devotes some energy to working with Roon to achieve that.

 

Finding 4: Improving music player (Zenith SE) has bigger impact than improving DAC (QX-5)

 

One point Darko made in his review about the "Portuguese fella" (the SE - I love his peculiar phraseology sometimes!) is this:

 

For the more seasoned digital audiophile caught between a streaming rock and a D/A converting hard place, know this: starting over with ten grand to spend, I’d take this Innuos and drop whatever cash I had left on a lesser DAC instead of putting the larger percentage of my dollars into a better D/A converter only to feed it ones and zeroes by a standard PC or Mac.

 

It occurred to me that with the loaner QX-5 sitting in my room, I actually had the makings of a very interesting experiment - i.e, compare these two configurations:

• Upgrade the DAC: replace the Codex with the QX-5 in my current chain
• Upgrade the music player: replace my current chain with the SE chain in finding 1, keeping my Codex DAC.

Cost wise, these are in the same ballpark, depending on whether you factor in retail vs. used cost, and the resale value of superseded gear. Not perfect, but an interesting comparison.

 

Well, I agree with Darko. Upgrading to the SE chain yielded more bang for the buck - to my ears - than upgrading the DAC. Now of course, upgrading both was to be transported to a divine plane, where you recline on a cloud, caressed by sunbeams, while angels sing.  

 

Finding 5: The tX-USBultra sounds better than the dX-USB HD Ultra

 

While my tX was en route to and from Korea, Eric kindly lent me his trifecta chain, comprising the sMS-200ultra and dX-USB HD Ultra. Since I had the QX-5 handy, I was able to drive this chain:

• Zenith SE > dX-USB HD Ultra (Hynes SR-4) > (AES) > QX-5
• Cybershaft OP-14 10 MHz reference clock feeding the sCLK-EX board in the dX-USB HD Ultra

The whole time I had it in my system I felt like something was missing, but with so much going on, I didn't really delve into why. When I got my tX-USBultra back, I immediately noticed how much better my system was sounding - and at a level of refinement that matched my recollection. Now curious, I compared the above to this chain:

• Zenith SE > tX-USBultra (Hynes SR-4) > (USB) > QX-5
• Cybershaft OP-14 10 MHz reference clock feeding the sCLK-EX board in the tX-USBultra.

Based on this experiment, the tX sounded significantly better. Now I'll caveat that I don't know per se what the relative merits of the USB and the AES input of the QX-5 are. Still, it is a data point, and suggests that the older dX may not be as refined as the tX.

 

Finding 6: The SR-4 has pulled ahead of the LPS-1 after more burn-in

 

I know I mentioned a couple weeks ago that I found these PSUs essentially head-to-head. Since then, the SR-4 has continued to burn in, and I am now able to compare on a chain of overall higher SQ with the SE. Whatever the reasons, I now find that switching back and forth powering the tX-USBultra, the SR-4 is the more refined PSU. Of course, this finding will be revisited when my LPS-1.2 comes in, and we shall see if the tables turn again with the 1.2. 

 

Summary

 

The Innuos Zenith MkII SE is a pretty remarkable piece of gear, and I recommend it heartily. Whether it's right for you depends on many things, but most importantly where you're coming from. If you're running your DAC direct from your stock MAC or PC, prepare to have your mind blown. As you go up the spectrum of existing system refinement, the SE's value proposition becomes more and more nuanced.

 

For my system, the SE does three things:

1. It allows me to achieve a level of SQ beyond what I previously had
2. It allows me to simplify my digital audio chain, although it does not completely eliminate the spaghetti
3. It allows me to sell parts of my current chain to partially defray the cost. I will soon be putting my modded sMS-200, my ISO-Regen, and 2-3 LPS-1s up for sale

There are some cons too:

1. For Roon users, the fact that Roon Core SQ lags the internal player is annoying
2. The fussiness of the USB controller with certain DACs requires care with system matching

Final Thoughts

 

Some people will look at this review and ask if I now believe "straight USB" is superior to "endpoint-based." I have never been tied to one approach or the other. For me, it was always about SQ and simplicity. Roy said something similar to me in one of his emails:

 

If you look back at my posts on CA, while I did find SQ to be a bit better with my "straight USB setup," a key reason for switching was because I preferred the simpler setup.  It is interesting just how many people have misinterpreted my rationale for leaving the trifecta, it was never for SQ reasons.

 

It did occur to me that an alternate path to achieve the same improvement as the SE provided in my system, was by the purchase of 4 rails of SR-7. Of course, I'd have to wait a year to get one! I discussed this with Roy, and his take was while the SR-7 still seemed like it had better specs:

 

With the Zenith SE more broken in, it has become a monster.  If I am to be honest, dynamics are a bit better than my server driven by the SR7 but I suspect this is due to the long DC leads I am using (1.5m) and the high impedance barrel connectors used (Paul Hynes hates these).

 

What I have concluded from my experience with the SE is that extremely high-end PSUs do make a palpable difference. This goes beyond the table stakes of low ripple noise, linear supplies, low output impedance, and isolation from high- and low-impedance noise. What these extreme PSUs also provide is a deep reservoir of instantaneous transient current delivery. We know this matters in analog amplification, but who knew it was so important in digital transmission!

 

What I would love to see is an SE-like product that married the concepts of extreme system and PSU optimization with clock optimizations. As Roy's experience shows, his (and others' on this thread) custom sCLK-EX modded servers have the potential for doing this. The trick is to incorporate several rails of an SR-7-like supply into the box, while keeping DC power leads as short and as low impedance as possible. There's the rub.

 

I sure hope Innuos's (or SOtM's or Antipodes') future systems give us the best of both worlds.

 

Great review.

Let's put the SE head to head with an sCLK-EX modded PC.

 

[Johnseye]

On 1/27/2018 at 12:45 PM, austinpop said:

 

 

My point about chains earlier bears some explanation. I think Roy has already said that by itself, the SE equals, and even slightly betters his custom server in terms of dynamism. But it cannot compete in terms of all the clock-related SQ improvements. The downside for his custom server is (quoting an email snippet):  ...that it requires 5 PSU rails (1 for the mobo, sCLK-EX, tX-USBexp, tX-USBhubIN, SSD) and to devote this many PSU rails to it is a challenge.

 

So no matter how you slice it, you have spaghetti! In one case:

• SE + tX-USBultra (1 PSU rail)
• custom server (5 PSU rails)

So it's not a competition to determine "the winner." Some - like me - don't want to mess with building a custom server, and like the simplicity of the SE.

 

Others really enjoy delving into all the myriad decisions in assembling a custom server:  mobo,  SATA,  drives, USB,  Ethernet, case, connectors, ... the list goes on. Once you start adding in the cost of SR7 rails (or perhaps Sean Jacobs rails), I wonder if there is even a cost advantage. 

 

What you've determined, with your own particular gear, is that the straight USB solution sounds better than the endpoint.  There could be a myriad of different reasons for this, but similar to what Roy, ElviaCaprice, myself and a few others discovered was that it does sound better.  I'm not discounting the endpoint, because it has a purpose.  If you don't want your server in the vicinity of your other equipment there's a huge benefit.  Likewise, I can put endpoints anywhere in my house.  However, when you're listening for best sound quality, the direct USB approach has an edge.

 

With regards to using 5 rails to power a server, I think that's a personal preference.  I'm using one rail to power the board, one to power my SOtM cards and another to power the hard drive.  Would I benefit from adding 2 more, maybe but because SOtM allows for the chaining of power through their cards it's the same clean power source.  In fact, the sCLK-EX card filters so there may be an additional gain.  I could get away with powering the SSD through the mobo and bring it down to 2 rails.  At some point I'll run a comparison.  Again, same clean power source, but coming through the mobo to power the SSD which has an SOtM filter on it.

 

Custom server or SE is another level of comparison.  Both are similar in design and serve the same function, just different hardware.  Just as we discover in all our testing, one will sound better.  No, it's not a competition.  I don't think any of us are competing, only trying to find out if there's something which can improve the sound quality of our music.

[Johnseye]

10 hours ago, romaz said:

As easy as it is to discern the sound quality differences between a standard Mac Mini and a Zenith SE, the SQ differences among these 4 speakers make the gap between a Mac Mini and the SE sound rather minuscule in comparison.  While I am not in the market to buy speakers in the range of the Alexias, if someone were to offer me a pair of Alexias with the stipulation that I could only use a standard Mac Mini as a source, I would take that deal in a heartbeat and never look back.  In the end, it is still the transducer that has the potential to make the biggest difference.

 

 

What I quoted is something to always keep in mind as we dump time and money down other rabbit holes.

 

I suspect Sean will have his hands full after this exposure.  I'm very tempted to request an ATX PSU.

 

[Johnseye]

1 hour ago, rickca said:

Don't you already have an SR7 or are you still waiting for yours?

 

I have an SR7.  The SR7 won't work with an ATX interface unless I use the Pico or HDPlex adapters.  That's something I was trying to avoid.  You also need 3 rails for an ATX interface without the adapter 12v, 5v & 3.3v as well as the ground.

 

It's give and take.  There may not be a difference between Paul's SR7 with one of the adapters and Sean's.  Like I said, I'm tempted but there's probably better use for the money.

 

1 hour ago, rickca said:

Great.  My SE just arrived an hour ago.  I may need a doctor.

 

That's the issue with prebuilt equipment.  Unless it's modular, allowing for updates, it will be outdated within a year.  It's an endless cycle in the audiophile world.  That's the main reason I took the build-your-own approach.  I can upgrade any component anytime.

 

[Johnseye]

4 minutes ago, KingRex said:

Even mine which is modular could be scrapped at 2 years.  The case has improved as well as Incorporating a new mother board that only uses 6 Watts.  But, I still have jaw hits the floor amazing  digital playback.  Yes It is getting better but hey, my digital blows my $7000 analog out of the water.  No contest. 

 

I still listen to vinyl all the time.  Different sound.  I prefer listening to jazz on vinyl and classical on digital.  Classical music gets so quiet and well recorded and pressed vinyl that gets that quiet is rare.  I like pre 80's rock on vinyl, post digital with exceptions.  Often times it depends on the recording or pressing itself and not the year or genre.

 

20 minutes ago, rickca said:

I agree and I bought the SE with my eyes open.  Whenever you buy a ready-made product, a prototype of an even more advanced model is already in development in the lab.  Even a newly launched product reflects component selection and design decisions that were made well before it was even announced.  

 

I have two systems, and I want at least one that is a well designed turnkey solution that is supported by the vendor.  I can always have a second server that I build myself for experimentation with the latest technology.  I don't have the DIY skills to keep pace with the discoveries that users here so generously share.  I find it exciting to learn what I can by following along and selectively implementing things I know I can manage.  

 

Nothing wrong with either approach.  I just like the flexibility when I can manage it.  There are some guys out there who build everything themselves.  I can't do that.  Computers are one thing, speakers, DACs, etc are another.

[Johnseye]

12 minutes ago, austinpop said:

 

I'm sure our friends from the frozen Northern climes got a good laugh out of this!

 

 

 

 

[Johnseye]

1 hour ago, vortecjr said:

We use to make something like this. It was a full blown Sonore Music Server running Vortexbox with an optional output card (USB, SPDIF/Analog, AES/EBU, or i2s) They had a miniBox ATX or straight power, or a combination of the two. They were fanless and the drives had vibration isolation with filters. We had a single external power supply brick or external upgrade power supply made from a Daitron module. Then towards the end of the series we made a dual output power supply called Duex which had a linear supply for the output card and a Daitron module for the mainboard. They sounded great, but the Rendu series is better sounding, less parts, smaller, and much less expense. This was circa 2014 so no USB circuit design from John S or on board linear regulation feeding all the circuits and oscillators 

 

Here is a picture of the stack:  

 

While the endpoint is a great idea, you still need a host server. If you guys could make something simple to run Roon Core I'd be interested. 

[Johnseye]

3 hours ago, vortecjr said:

John, we have that already;) It's called sonicTransporter and it's made by Small Green Computer. We are hardware and software partners with them, but I don't make any money on the servers. sonicTransporter runs a version of Sonicorbiter that is setup for server duties. It can be a Roon server, DLNA server, Logitech Media Server, etc. Its very similar in look and feel to Sonicorbiter running on the micro/ultraRendu and meant to be complimentary. The are reasonably priced, include support, and are turnkey. 

 

Also, I would say that endpoints need "server software" running on any computer or NAS. The point is you don't need a dedicated music server. There are also people just streaming via Tidal or Quboz and in that case neither computer or NAS is needed.        

 

How can I say this...the sonicTransporter isn't as good as it could be?  I think the hardware side of it is too easily duplicated and not as well designed as the Rendu.

[Johnseye]

2 hours ago, rickca said:

OK but that doesn't answer my question.  Sure, there may be alternatives that sound as good as the Supermicro.  The question is why are some motherboards less noisy than others?

 

Is it the VRM?  Is it the layout?  Is it the PDN?  

 

I'm going to take a guess in that it's a combination of things with the layout, caps & regs having the most impact.  Just a guess though.  A communication with DFI, Supermicro or Jetway may answer that.

 

1 hour ago, vortecjr said:

They should sound the same. Also, the controllers does is tell the renderer what to play. The renderer is actually playing the content.

 

What does "playing the content" mean exactly?  At that point it's still a digital stream, not analog.  I always wondered exactly what the renderer does.

 

[Johnseye]

32 minutes ago, vortecjr said:

It’s the player in the digital loop. For example...It takes content in FLAC format and outputs PCM. The controller is the director telling the server what file to offer up to the renderer to play.

 

But it's the Roon Core/server/controller that applies the DSP or does the upsampling.  Same with HQPlayer.  So it does that while still in FLAC format?  I don't see how that's possible.  I thought it was the server that converts it from FLAC to PCM or DSD.

[Johnseye]

2 hours ago, vortecjr said:

people reading this should know that if they use a Supermicro board for a Roon server that it will be short on performance. Roon recommends:

• Intel Core i3, Ivy Bridge+
• 4GB RAM
• SSD boot drive
• 1440 x 900 Resolution

I can tell you from speaking with Andrew that he recommends an i5 at minimum and prefers an i7. The difference being that you want an i7 with a large music collection and with DSP. His i7 is only 11 watts so nothing to obsess over regarding noise. Professional use could mean better parts or longer runs, or any number of things.

 

As for the noisy DC-DC converter there is no free lunch you now have those right on the board. And if you use the 4 pin power source input and an ATX power source you will have 2x the number of noisy DC-DC converter:) 

 

Also, you said that the tX-USBexp does not draw current form the motherboard. You might want to double check this with SOtM. I asked them about this a long time ago, but I can't remember with certainty what they said. 

 

I've used an i7 dumbed down to 800mhz with no problem. I've used a Celeron with no problem. Upsampling and DSP have no issues. 

 

I'm sure the 1440 x 900 resolution is critical however. 

[Johnseye]

13 minutes ago, vortecjr said:

DSP aside it takes a ton of time/HP to sort out large collections. During this process we get reports of strange things happening. Andrew at SGC knows more about than I do and has had extensive chats with Roon on the subject. 

 

Specifically when doing an "Audio Analysis" in Roon on a lesser processor it takes a very very long time.  Otherwise it's all equal.

[Johnseye]

1 minute ago, lmitche said:

There is a control to slow or speed up the audio analysis speed in Roon. It has immediate impact on performance.

 

Yes thanks, I've used it.  It does speed things up, but analysis is still much longer with a Celeron vs i7.  This is really not an issue though as analysis only takes place when adding content or rebuilding a PC and restoring a library backup.