Jump to content
austinpop

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

Rate this topic

Recommended Posts

4 minutes ago, rickca said:

If we all pretend we're not interested, maybe it won't be too high

Good strategy :)

Also the fact that it needs an external PSU is a good thing, but it means we have to factor in the cost of a decent PSU. You can't buy a master clock and compromise its performance with a mediocre PSU.

Share this post


Link to post
Share on other sites
4 minutes ago, mozes said:

Also the fact that it needs an external PSU is a good thing, but it means we have to factor in the cost of a decent PSU. You can't buy a master clock and compromise its performance with a mediocre PSU.

Good point, the Mutec REF 10 has an integrated LPS.  The SOtM unit will need 12V/2A.  

Share this post


Link to post
Share on other sites

With all due respect, looking at Austinpop's "pasta medley" it occurs to me that SOtM would do well to come out with a product that includes their "trifecta" plus their new master clock all placed in a single chassis. A slot to integrate the switch mod unit could also be included. Optimal connecting wiring/cabling would be a nice feature as well.

 

The chassis then could be paired with a separate power supply unit that has the functionality of several isolated sPS-500s. People that don't want the power supply unit could purchase their own collections of LPS-1s and Paul Hynes, etc,, power supplies.

Share this post


Link to post
Share on other sites
On 9/22/2017 at 10:54 AM, Bruce Orr said:

Auricgoldfinger -  thank for the heads up. Glad I asked!  Is there another piece that would work as a facilitator to enable the click to work? Or would I have to swap out something?  

 

Bruce, see my post above. Getting ISO-R and UltraRendu to play in a reference clock ecosystem requires modification, and SOtM will not mod these components. You're welcome to ask, of course!

Share this post


Link to post
Share on other sites
5 minutes ago, austinpop said:

I know this clock stuff gets complicated. Here is my layman's summary of what we know so far. I claim no deep knowledge - I am a bit of a dotard when it comes to clocks! But we have the benefit of @JohnSwenson's posts, which I had the foresight to bookmark. I've included them in the reference section below.

 

I am going to assume you've looked at the index on the first post, and read through the sections on clock chains. It'll explain the context of how we got here.

 

We now have a lot of empirical evidence from contributors to this thread of SQ benefits from clock improvements:

  • "better" clocks in individual components matter. This applies not just to SOtM gear (the Ultra components), but also the ISO-Regen, the UltraRendu, AQvox switch etc. How to define "better" gets complicated. Certainly, low phase noise is one metric, but overall component design matters too.
  • Some real magic happens with clock "chains," where the chain of components from the DAC and upstream ALL have "better" clocks.

But how do you make the clocks better? One way to do this is to use a very high quality 10MHz reference clock - like the Mutec Ref 10, the Cybershaft OP-14, and hopefully soon, the upcoming SOtM OCX-10. But very few components are designed to accept a reference clock. Let's look at why:

  • To accept a reference clock, the component needs to be using a frequency synthesizer to generate the clock frequencies it needs, AND be designed to accept a reference input. See [1].
  • Such devices do exist:
    • DDCs like the Mutec MC-3+/USB
    • DACs like the Teac NT-503, various Esoteric and dCS
    • reclockers and endpoints like the tX-USBultra, and sMS-200ultra
  • However, most digital components use fixed frequency oscillators that are designed for the purpose of the component. This is especially true of non-audio gear like switches, routers, motherboards, NICs and USB adapters.
  • What frequencies do these components typically use? As JS explains in [2], 24MHz for USB and 25MHz for Ethernet is common. Similarly in [3], he explains why DACs often use frequencies of 22.5792 MHz and 24.576 MHz. The common theme is multiples (or submultiples) of the frequencies needed for the core function. For example 2.5792 MHz is 512x the Redbook 44.1 kHz sample rate.

So how do you build a chain where every element can be raised to the quality of a reference clock? Either buy from the limited set of devices that support this, OR do some mods. Enter the SOtM sLCK-EX board. The unique aspects of this device, that has made it a popular choice on this thread are:

  1. While there is no published phase noise data, its clocks have been empirically found to have much better SQ than the base clocks on the sMS-200 and other USB and Ethernet devices
  2. It has a frequency synthesizer that accepts a reference clock
  3. It has 4 independent clock taps, that can be programmed by SOtM to generate any of the frequencies discussed earlier
  4. It can be housed either in the sMS-200ultra, the tX-USBultra, the dX-USB HD Ultra, or in an enclosure of your own choosing
  5. And most importantly - SOtM will modify non-audiophile fixed clock components - like routers, switches, mobos, NICs etc - to replace the builtin fixed clock with the output of an sCLK-EX board. It does this in 2 parts:
    • The internal clock(s) of the component are modified to accept external clock(s) (don't ask me gory details), which are accepted via SMB 50 ohm coaxial input ports that are added
    • The relevant clock tap(s) from the sCLK-EX board are delivered as output SMB ports on the component housing the sCLK-EX board (for example, an sMS-200ultra).
  6. One final note: SOtM usually declines to modify components it considers audio, or audiophile. So they will decline to modify DACs, or devices like the ISO-Regen, UltraRendu, etc. This is a very understandable position!

If you think about it, what the sCLK-EX board is doing is:

  • replacing a fixed builtin clock with a better quality clock, that happens to be synthesized
  • providing a distribution mechanism for high-quality clocks across multiple devices
  • This distribution mechanism then provides the pathway to distribute the even higher quality of the reference clock to the whole chain.

This is how in my chain, I can reap the benefit of the single-output Cybershaft reference clock for a clock-chain of size 5. Here's how;

  • The reference clock drives the sCLK-EX in my tX-USBultra. Through the existing sCLK-EX distribution in my chain, I now have 5 clocks benefiting from the Cybershaft:
    • Switch: 25MHz clock for Ethernet
    • modded sMS-200:
      • 24 MHz clock for USB
      • 25 MHz clock for Ethernet
      • 25 MHz clock for system
    • tX-USBultra: 24 MHz clock for USB

I hope this clarifies things.

 

Again, this is meant to be simplistic, so I am sure there is a lot more complexity under the covers.

 

Reference JS Postings

[1] JS's Taxonomy of clock types

[2] JS explains rationale for 24 and 25 MHz for USB and Ethernet

[3] JS on typical DAC clock frequencies

 

 

I did not mention this explicitly, but implicit in the above is the other dimension of consideration: the PSU. To achieve and exploit the highest quality clock distribution, every device in the chain must have PSUs of outstanding quality. 

 

Hence the focus on the LPS-1, the VR Mini, the PH SR7, the sPS-500, etc.

Share this post


Link to post
Share on other sites
29 minutes ago, austinpop said:

I know this clock stuff gets complicated. Here is my layman's summary of what we know so far. I claim no deep knowledge - I am a bit of a dotard when it comes to clocks! But we have the benefit of @JohnSwenson's posts, which I had the foresight to bookmark. I've included them in the reference section below.

 

I am going to assume you've looked at the index on the first post, and read through the sections on clock chains. It'll explain the context of how we got here.

 

We now have a lot of empirical evidence from contributors to this thread of SQ benefits from clock improvements:

  • "better" clocks in individual components matter. This applies not just to SOtM gear (the Ultra components), but also the ISO-Regen, the UltraRendu, AQvox switch etc. How to define "better" gets complicated. Certainly, low phase noise is one metric, but overall component design matters too.
  • Some real magic happens with clock "chains," where the chain of components from the DAC and upstream ALL have "better" clocks.

But how do you make the clocks better? One way to do this is to use a very high quality 10MHz reference clock - like the Mutec Ref 10, the Cybershaft OP-14, and hopefully soon, the upcoming SOtM OCX-10. But very few components are designed to accept a reference clock. Let's look at why:

  • To accept a reference clock, the component needs to be using a frequency synthesizer to generate the clock frequencies it needs, AND be designed to accept a reference input. See [1].
  • Such devices do exist:
    • DDCs like the Mutec MC-3+/USB
    • DACs like the Teac NT-503, various Esoteric and dCS
    • reclockers and endpoints like the tX-USBultra, and sMS-200ultra
  • However, most digital components use fixed frequency oscillators that are designed for the purpose of the component. This is especially true of non-audio gear like switches, routers, motherboards, NICs and USB adapters.
  • What frequencies do these components typically use? As JS explains in [2], 24MHz for USB and 25MHz for Ethernet is common. Similarly in [3], he explains why DACs often use frequencies of 22.5792 MHz and 24.576 MHz. The common theme is multiples (or submultiples) of the frequencies needed for the core function. For example 2.5792 MHz is 512x the Redbook 44.1 kHz sample rate.

So how do you build a chain where every element can be raised to the quality of a reference clock? Either buy from the limited set of devices that support this, OR do some mods. Enter the SOtM sLCK-EX board. The unique aspects of this device, that has made it a popular choice on this thread are:

  1. While there is no published phase noise data, its clocks have been empirically found to have much better SQ than the base clocks on the sMS-200 and other USB and Ethernet devices
  2. It has a frequency synthesizer that accepts a reference clock
  3. It has 4 independent clock taps, that can be programmed by SOtM to generate any of the frequencies discussed earlier
  4. It can be housed either in the sMS-200ultra, the tX-USBultra, the dX-USB HD Ultra, or in an enclosure of your own choosing
  5. And most importantly - SOtM will modify non-audiophile fixed clock components - like routers, switches, mobos, NICs etc - to replace the builtin fixed clock with the output of an sCLK-EX board. It does this in 2 parts:
    • The internal clock(s) of the component are modified to accept external clock(s) (don't ask me gory details), which are accepted via SMB 50 ohm coaxial input ports that are added
    • The relevant clock tap(s) from the sCLK-EX board are delivered as output SMB ports on the component housing the sCLK-EX board (for example, an sMS-200ultra).
  6. One final note: SOtM usually declines to modify components it considers audio, or audiophile. So they will decline to modify DACs, or devices like the ISO-Regen, UltraRendu, etc. This is a very understandable position!

If you think about it, what the sCLK-EX board is doing is:

  • replacing a fixed builtin clock with a better quality clock, that happens to be synthesized
  • providing a distribution mechanism for high-quality clocks across multiple devices
  • This distribution mechanism then provides the pathway to distribute the even higher quality of the reference clock to the whole chain.

This is how in my chain, I can reap the benefit of the single-output Cybershaft reference clock for a clock-chain of size 5. Here's how;

  • The reference clock drives the sCLK-EX in my tX-USBultra. Through the existing sCLK-EX distribution in my chain, I now have 5 clocks benefiting from the Cybershaft:
    • Switch: 25MHz clock for Ethernet
    • modded sMS-200:
      • 24 MHz clock for USB
      • 25 MHz clock for Ethernet
      • 25 MHz clock for system
    • tX-USBultra: 24 MHz clock for USB

I hope this clarifies things.

 

Again, this is meant to be simplistic, so I am sure there is a lot more complexity under the covers.

 

Reference JS Postings

[1] JS's Taxonomy of clock types

[2] JS explains rationale for 24 and 25 MHz for USB and Ethernet

[3] JS on typical DAC clock frequencies

 

 

@austinpop, thank you for this wonderful summary, very nicely (and if I may say - scholarly) written! 

 

I have been reading a lot about the Ayre QX-5 Twenty these days. Apparently it looks like one of those highly sought after one-box solutions (streamer, DAC, preamp and very high quality clock in a single box). What is interesting about this device is that it uses multiple separate built-in power supplies for different digital inputs as well as a built in linear PSU for the analog output. It also apparently reclocks every digital signal (be it USB, ethernet, AES/EBU, Toslink) through the built in ultra high quality crystal clock oscillator with an extremely low phase noise. Charlie Hansen also developed special minimum phase filters for this DAC.

 

So, did anyone have the opportunity to compare the SOtM trifecta with the Ayre QX-5? I know @austinpop uses (and loves) the Ayre Codex DAC, but perhaps the QX-5 can be the component a lot of us are looking for, especially having in mind it’s a certified Roon Ready streaming DAC. It ain’t cheap though. :)

Share this post


Link to post
Share on other sites
38 minutes ago, AmusedToD said:

 

@austinpop, thank you for this wonderful summary, very nicely (and if I may say - scholarly) written! 

 

I have been reading a lot about the Ayre QX-5 Twenty these days. Apparently it looks like one of those highly sought after one-box solutions (streamer, DAC, preamp and very high quality clock in a single box). What is interesting about this device is that it uses multiple separate built-in power supplies for different digital inputs as well as a built in linear PSU for the analog output. It also apparently reclocks every digital signal (be it USB, ethernet, AES/EBU, Toslink) through the built in ultra high quality crystal clock oscillator with an extremely low phase noise. Charlie Hansen also developed special minimum phase filters for this DAC.

 

So, did anyone have the opportunity to compare the SOtM trifecta with the Ayre QX-5? I know @austinpop uses (and loves) the Ayre Codex DAC, but perhaps the QX-5 can be the component a lot of us are looking for, especially having in mind it’s a certified Roon Ready streaming DAC. It ain’t cheap though. :)

 

Thanks!

 

As you know, I have been conversing with Charles over on the QX-5 thread. I think the key question is - how well does the QX-5's:

 

38 minutes ago, AmusedToD said:

built in ultra high quality crystal clock oscillator with an extremely low phase noise.

 

compare with the reference clocks we are discussing here: the Ref 10, the OCX-10, and the OP-14?

 

One confounding factor would be that a built-in clock will not suffer the unavoidable losses due to clock cable lengths. So ultimately, it would boil down to an SQ comparison.

 

This gives me an idea. I know my local Ayre dealer well. I might just borrow a QX-5 for comparison with my pasta medley. 9_9

 

And of course, the QX-8 is coming!

Share this post


Link to post
Share on other sites
24 minutes ago, austinpop said:

This gives me an idea. I know my local Ayre dealer well. I might just borrow a QX-5 for comparison with my pasta medley. 9_9

 

And of course, the QX-8 is coming!

 

Yes, the QX-8 is coming, but not with the same quality clock, hence the lower price.

 

So if you have the chance to compare your SOtM trifecta (with the added Cybershaft) to the Ayre QX-5 - please do this as the QX-5 is a reference quality product and it should be a clear marker as to the real value of the SOtM “spagetti” solution :) In your case you might even get rid of the headphone amp if the one in the QX-5 is that good :)

Share this post


Link to post
Share on other sites

It seems like a couple of people have chosen the single-box solution of the Linn Klimax DSM over the partial SOtM trifecta.  It'd be great to hear from them.

Share this post


Link to post
Share on other sites

Well, predictably enough, Charles Hansen does not think any external clock could improve over the internal clock in the QX-5.

 

 

Now I am really interested to try the comparison with the QX-5!

 

1 hour ago, AmusedToD said:

So if you have the chance to compare your SOtM trifecta (with the added Cybershaft) to the Ayre QX-5 - please do this as the QX-5 is a reference quality product and it should be a clear marker as to the real value of the SOtM “spagetti” solution :) In your case you might even get rid of the headphone amp if the one in the QX-5 is that good :)

 

I do not expect the headphone amp to be better. The Cavalli amps are truly special. Ditto the QX-5 vs the pasta medley.

 

But that is the fun of comparisons! What you expect and what you find are often very different. We shall see.

Share this post


Link to post
Share on other sites
3 hours ago, austinpop said:

I might just borrow a QX-5 for comparison with my pasta medley.

Sure you can compare QX-5 vs your pasta medley.  But what if the trifecta + Cybershaft into the QX-5 USB input sounds the best?

Share this post


Link to post
Share on other sites
24 minutes ago, rickca said:

Sure you can compare QX-5 vs your pasta medley.  But what if the trifecta + Cybershaft into the QX-5 USB input sounds the best?

 

I'll just have to cross that bridge when I come to it!

Share this post


Link to post
Share on other sites
1 hour ago, austinpop said:

Well, predictably enough, Charles Hansen does not think any external clock could improve over the internal clock in the QX-5.

 

It cracked me up whenever these claims were popping up while nobody really bothered to back things up with any kinda evidence whatsoever

 

Quote

As far as external clocks go, the internal clock in the QX-5 is equal or better than any external clock on the market - even the $20,000 ones. Adding an external clock input to the QX-5 would actually degrade its performance.

 

They wouldn't even show any kinda numbers / data / measurements etc. but it's just "better" anyways. LOL

 

I don't mean to be disrespectful but sometimes we've gotta be realistic IMHO. Now I'm gonna be the bad guy since I'm spelling it out.

Share this post


Link to post
Share on other sites
31 minutes ago, seeteeyou said:

 

It cracked me up whenever these claims were popping up while nobody really bothered to back things up with any kinda evidence whatsoever

 

 

They wouldn't even show any kinda numbers / data / measurements etc. but it's just "better" anyways. LOL

 

I don't mean to be disrespectful but sometimes we've gotta be realistic IMHO. Now I'm gonna be the bad guy since I'm spelling it out.

+ 1

 

Typical comment from a manufacturer. 

Share this post


Link to post
Share on other sites

On another note, SOtM has solved the problem with my reclocked NUC.  Turns out they mislabeled the first two Sclk-EX clocks.  Once swapped/corrected, works great, video provided by them to show so.  I await it's arrival.  

Share this post


Link to post
Share on other sites
2 minutes ago, ElviaCaprice said:

On another note, SOtM has solved the problem with my reclocked NUC.  Turns out they mislabeled the first two Sclk-EX clocks.  Once swapped/corrected, works great, video provided by them to show so.  I await it's arrival.  

Hey, that's great!  So please remind me which clocks you replaced with sCLK-EX and which piece of equipment hosts your sCLK-EX board.

Share this post


Link to post
Share on other sites
16 minutes ago, ElviaCaprice said:

On another note, SOtM has solved the problem with my reclocked NUC.  Turns out they mislabeled the first two Sclk-EX clocks.  Once swapped/corrected, works great, video provided by them to show so.  I await it's arrival.  

 

Awesome! Look forward to your impressions.

Share this post


Link to post
Share on other sites
40 minutes ago, rickca said:

Hey, that's great!  So please remind me which clocks you replaced with sCLK-EX and which piece of equipment hosts your sCLK-EX board.

I gave SOtM liberty to use the remaining 3 clocks available on the NUC, 1 went to the txbexp pcie card.  So that is exactly what they did.  I did specify for them not to replace any USB or Ethernet port clocks, since I would be using the PCIE card to stream from.  I know they replaced the system clock but the other two??   Unfortunately I won't be able to use my new PH SR7 MR4 until next summer, stuck in Minnesota and just too expensive to ship down here to Costa Rica, never know what customs might charge.

The SCLK-EX board is free standing.

 

Here will be the chain I shall use for now, the PH SR7 will replace any temporary 12V inputs, there are 3.

JETWAY NUC JBC311U93-2930-B   (modified with 3 SCLK-EX clocks) (added mini pcie to PCIE 2X adapter) (txbexp pcie usb card modded with 1 SCLK-EX clock) => (USPCB) Chord 2 Qute => Omega Super 8XRS

 

For power, the NUC shall be powered via LIPO battery (I have two powerful 16AH ones here in CR for my fishfinder)

SCLK-EX and Chord 2Qute shall be powerd by 2 LPS-1's in a series 12V.

The Jetway NUC luckily has a SATA II port, not shown in specs.  So I shall power a 5V 2.5" HDD with an LPS-1 for data.  The txbexp card shall be powered by another LPS-1, 7V.

 

The addition of the SCLK-EX board and modifications wasn't cheap, I also added the master clock input.  All said and done it was just under $1300.  I only paid $100 for the NUC used and $150 for the txbexp used. 

 

I will test with and without the ISO Regen.  But from Roy's findings I expect the ISO Regen may be a burden to SQ.

Share this post


Link to post
Share on other sites
5 hours ago, austinpop said:

Now I am really interested to try the comparison with the QX-5!

 

And we are patiently waiting for you to do it! :):):) 

 

There is an interview on YouTube with @Charles Hansen discussing the abilities of the QX-5 and the long listening hours invested in its development. The story of the Russian company that makes the crystal clock oscillators for Ayre is particularly an interesting one :)

Share this post


Link to post
Share on other sites
3 hours ago, ElviaCaprice said:

+ 1

 

Typical comment from a manufacturer. 

 

Well, @Charles Hansen is not really any manufacturer, so his claims should be taken very seriously. I think I read somewhere (or perhaps heard in his YouTube interview) that his Russian produced clock has the lowest ever phase noise in an audio product and that the performance of this clock equals or exceeds some of the most expensive ones in the market (the Russian company also produces clock oscillators for satellites). He didn’t specifically mention the competition, but I think he was aiming at dCS and Esoteric, so I guess there were some A/B tests involving gear from the big boys. Just my guess, Charlie will shed more light on the topic.

Share this post


Link to post
Share on other sites
4 hours ago, seeteeyou said:

 

It cracked me up whenever these claims were popping up while nobody really bothered to back things up with any kinda evidence whatsoever

 

 

They wouldn't even show any kinda numbers / data / measurements etc. but it's just "better" anyways. LOL

 

According to Mr. Hensen, they used some very high end testing equipment for measuring the performance of the clock, so data and graphs would really be beneficial. :)

Share this post


Link to post
Share on other sites
4 hours ago, seeteeyou said:

 

It cracked me up whenever these claims were popping up while nobody really bothered to back things up with any kinda evidence whatsoever

 

 

They wouldn't even show any kinda numbers / data / measurements etc. but it's just "better" anyways. LOL

 

I don't mean to be disrespectful but sometimes we've gotta be realistic IMHO. Now I'm gonna be the bad guy since I'm spelling it out.

 

Most (>90 %) manufacturers doesn’t publish their clock measurements, probably because most people doesn’t know how to interpret them. Charles Hansen and his team is one of the most respected HIFI innovators. You don’t think they have actually tried if an external clock input to the QX-5 would degrade its performance or not?

Share this post


Link to post
Share on other sites

Not really what I think or anyone thinks, it's just a matter of apples-to-apples comparisons. Talk is cheap and that's all.

 

BTW, let's compare SoTM sCLK-OCX10 to one of their best MV336

 

https://www.morion-us.com/morion-inroduces-xtal-clear-ocxo/

Quote

Morion has released Xtal-Clear®, the lowest ADEV, lowest close-in phase noise and ultra-stable 10 MHz OCXO available in the world today! The MV336 has short-term stability (ADEV) of <1E-13 at 1 Sec. and <3E-13 to 100 Sec. Temperature stability is < ±2E-11 from -20 to +70 deg. C and phase noise is better than -92 dBc/Hz @ 0.1 Hz and -120 dBc/Hz @ 1 Hz.


https://www.morion-us.com/quartz_oscillators/790/mv336/

http://www.morion.com.ru/eng/news/open/95/

http://www.morion.com.ru/eng/oscillators/ocxo/

http://www.morion.com.ru/catalog_pdf/31-MV336_REV_9.pdf

http://www.morion.com.ru/catalog_pdf/32-MV336M_REV_5.pdf

https://www.morion-us.com/catalog_pdf/mv336m.pdf

 

Just stealing (MAN, I'm such a petty thief) something from Rajiv that's posted on page 145

 

https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?page=145&tab=comments#comment-722223

 Typical Phase noise(dBC/Hz)

                                                                                           My

                                       sCLK-OCX10                       Cybershaft    Mutec
     Frequency  Standard  Advanced  Reference       OP-14         Ref 10 (published)
         10Hz  :       -125         -135        -140                   -132.4              ≤ -142
       100Hz  :       -150         -160        -160                   -141.9              ≤ --155
         1KHz  :       -160        -163        -165                                             ≤ --160
       10KHz  :       -165        -165        -165
      100KHz  :      -165        -165        -165
      Noise floor:                                                                                        ≤ -166

 

https://www.morion-us.com/catalog_pdf/mv336.pdf#page=2

                                       Morion MV336
     Frequency         -              LN          ULN  
        0.1Hz  :       <-80          <-85        <-92
           1Hz  :       <-113      <-116        <-120
         10Hz  :       <-143      <-144        <-145
       100Hz  :       <-154      <-156        <-157
         1kHz  :       <-160      <-160        <-160
       10kHz  :       <-160      <-160        <-160
 

Maybe they're putting something even better inside QX-5 Twenty?

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×