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

[Johnseye]

9 minutes ago, sig8 said:

My question is that who does the clocking function when you have two clocks, one sCLK-EX, and one master clock with several taps which can be utilized for different components. In other words, if we have replaced the clocks on MoBo with taps from sCLK-EX, then what master clock does at that point? Thanks for helping me understand.

 

Google USB master clock.  Much simpler.

[rickca]

1 hour ago, ElviaCaprice said:

I think this would be a great opportunity to use two tXUSBexp pcie cards mounted in two lanes, one for input off a HDD, the other for ouput to the DAC.  One clock point could be utilized for both cards.

I don't understand how one clock point can suffice for two tX-USBexp cards.  How is this implemented?

[lmitche]

1 hour ago, rickca said:

I don't understand how one clock point can suffice for two tX-USBexp cards.  How is this implemented?

You can chain the clock source in parallel.

[rickca]

12 minutes ago, lmitche said:

You can chain the clock source in parallel.

Sorry to be dense, but what does that mean?  For a single tX-USBexp something from the sCLK-EX board gets transplanted onto the tX-USBexp, right?  So what gets put on the second tX-USBexp?

[Johnseye]

1 minute ago, rickca said:

Sorry to be dense, but what does that mean?  For a single tX-USBexp something from the sCLK-EX board gets transplanted onto the tX-USBexp, right?  So what gets put on the second tX-USBexp?

 

Having not physically seen this myself, I'm only going off of what I've read.  Nothing gets transplanted.  A cable connects the two boards so the sCLK-EX is used at the tap points.  The clock on the sCLK-EX board just needs to be the same frequency.

[lmitche]

1 hour ago, rickca said:

Sorry to be dense, but what does that mean?  For a single tX-USBexp something from the sCLK-EX board gets transplanted onto the tX-USBexp, right?  So what gets put on the second tX-USBexp?

Think y-cable from a single clock source going to both devices.

[rickca]

1 hour ago, Johnseye said:

Nothing gets transplanted.

OK, but something gets done to the device that's running off the sCLK-EX clock point that makes it dependent on the sCLK-EX board.  What happens to the original clock on the device that gets modified?

[Johnseye]

Just now, rickca said:

OK, but something gets done to the device that's running off the sCLK-EX clock point that makes it dependent on the sCLK-EX board.  What happens to the original clock on the device that gets modified?

 

I believe the original clock is bypassed.  Someone would need to confirm however.

[ElviaCaprice]

36 minutes ago, Johnseye said:

 

I believe the original clock is bypassed.  Someone would need to confirm however.

It's removed and replaced with a clock control input which SOtM will cleverly attach unhindered to the circuit board.  I'm very impressed with their work (they definitely know what they are doing) and would never dream of trying this DIY.

Having fried a mobo in the past with my own DIY clock replacement.

 

Only question I have is that on their own server builds, why only the one clock replacement of the main system of the  mobo?  This I think falls short in what should be replaced on the mobo and could be the reason for the low interest of their own servers success, besides the cost factor? 

Thus if you we're streaming from a HDD.  You would only benefit from 2 sCLK-EX clocks, which is probably why these servers fall short in performance?????  (highly speculative since I've never heard one)

 

I also think they try to do too much with these server builds and would be far more successful in different builds that focused on different streaming/data sources.  Something like PPA, offer different choices in mobos and there modifications.

[Forehaven]

Perhaps time to ?repost a diagram of the chained devices? Incl a mobo implementation?  

[ElviaCaprice]

36 minutes ago, Forehaven said:

Perhaps time to ?repost a diagram of the chained devices? Incl a mobo implementation?  

 

Sure,

 

SERVER =>  DAC => SPEAKERS

 

It doesn't get any simpler. 

See my bottom signature for details.  Don't know what you mean by mobo implementation?  You mean the clock mods?  Don't even know the details myself.  Other than 3 sCLK-EX points are used with a bunch of clock control points, some interconnected using the same point.  I think it's best just to allow SOtM to figure out what they can replace and go at it.  All you need to do is outline to them what your data/streaming configuration is going to be and have them pick those clocks that could benefit the most.  Optimally you want all 4 sCLK-EX points effecting your audio stream in one given event.

 

12V HDPlex to 2Qute (PH SR7 12V)

19V HDPlex to NUC (PH SR7 12V)

12V (2) LPS-1s to sCLK-EX (PH SR7 12V)

7V LPS-1 to tXUSBexp => USPCB => 2Qute => Omegas

5V LPS-1 to 2.5" HDD (both OS and DATA) => PPA sata cable to SATA II NUC

Coming soon sclk-ocx10 reference master clock (PH SR7 12V) => 50ohm to sCLK-EX

Coming soon Paul Hynes SR7 MR4

 

All powered by 1KVA Topaz.

Windows 10 PRO DIY optimized, JRiver 23 (64bit)

 

 

[rickca]

1 hour ago, ElviaCaprice said:

It's removed and replaced with a clock control input

OK I found it in the SOtM sCLK-EX operating instructions.  They remove the original oscillator and replace it with a  supplied clock connect PCB.   So I guess each of the two tX-USBexp would get one of these clock connect PCBs installed.

 

I'm not trying to figure out how to do this myself.  I just want to understand how it works.

 

[ElviaCaprice]

14 minutes ago, rickca said:

Interesting to learn that the sCLK-EX board input power is SATA.

 

If you mean by SATA, 12V, you are incorect.  Not all SATA HDD's operate off of 12V.  In fact my 2.5" HDD 5TB SATA III is 5V.

The sCLK-EX has jumpers and options for power input.

[rickca]

4 minutes ago, ElviaCaprice said:

 

If you mean by SATA, 12V, you are wrong.  Not all SATA HDD's operate off of 12V.  In fact my 2.5" HDD 5TB SATA III is 5V.

The sCLK-EX has jumpers and options for power input.

I edited my post.  I think we're going too far into the weeds so I'll stop.  

[ElviaCaprice]

36 minutes ago, rickca said:

I edited my post.  I think we're going too far into the weeds so I'll stop.  

My apologies, rickca, for my strong replies.  Yes, I think your correct.  I think I've outlined more than enough what is going on with my system and it's SOtM components.

[atxkyle]

Very interesting last couple days of posts.  Although I have heard the same things as everyone else, I'm still a newbie to this hobby and don't understand the 'why' ... why ethernet network streamers are sensitive to what's upstream.  Or why the server should be sensitive to the file it receives from Tidal via the router.  In my mind, the streamer (or server) should be loading the file data into RAM, and then outputting it to the DAC.  i.e. I'd think the starting point for a music player is always a 'clean' file, and the limiting factor is then how perfectly it can time the output of the 1s and 0s.

 

Also strikes me that at least a few companies have already invested in developing very minimalist, low power, low noise motherboards for network streamers.  Seems like there could be a good market for leveraging the tech from those devices to evolve them into Roon servers (with inputs for super-duper-master-clocks, and careful attention paid to isolating/filtering noise from storage, etc.).  I guess SOtM already has their version of this with the 1000SQ - so are people doing the custom mod servers cause they want something different than the 1000SQ offers, or cause it's less expensive to do the custom (or both)?

[austinpop]

6 hours ago, sig8 said:

My first basic question is that how a master clock plays in to this scheme when you already have a clock in the form of sCLK-EX. Pardon my ignorance in this topic please. Thanks.

 

Not sure if this got answered? I was away from the thread for a day, and a lot of interesting posts!

 

I think I addressed your question in this post here: https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?do=findComment&comment=722622

 

But in a nutshell:

1. the sCLK-EX board uses a frequency synthesizer to generate the 4 independent frequencies that appear on it clock outputs.
2. The sCLK-EX board uses an internal oscillator to generate the reference frequency for its frequency synthesizer
3. Additionally, it is designed to accept an external reference input, which can replace the output of its internal oscillator
4. Assuming the external reference has ultra low phase noise - like the Mutec, the Cybershaft, or the upcoming SOtM OCX-10, this low phase noise gets "inherited" by the sCLK-EX outputs.

 

[austinpop]

Over on the Lush thread, someone mentioned how they knew their system was sounding good when they could listen to Adele's 21 without wincing. I'm paraphrasing.

 

That got me thinking. After all,  the clock improvements in trifecta medley have so dramatically removed that digital haze and fatigue. It's been years since I listened to 21. Would my system be up to the Adele test? 

 

I'm happy to report - YES! I ended up listening to the whole album. On my system. With HD800 headphones, that do not suffer bright recordings gladly. The whole album.

 

If that doesn't epitomize what we mean by smoothness and lack of digital glare, I can't imagine what does.

[lateboomer]

I agree Adele's 21 is not easy to listen to, not to mention it has to finish the whole album.

[lateboomer]

16 hours ago, Forehaven said:

nice post Elva, but kind of a bummer for those who like to use HQP...

 

May be the recently launched JCAT Femto Net Card can fill in this gap? Up-sample to dsd256 or dsd512 in HQPlayer server and output direct via this femto card with audiophile CAT7 cable to the child femto card at NAA. The NAA would be your ultimate tricked out low-powered NUC with  SCLK-EXs, ISO-Regen, master clock or the extreme trifecta medley.

 

May be this would beat the direct usb method, who knows.

[lateboomer]

16 hours ago, Johnseye said:

 

 

Not sure if you guys read my post above.  My SR7 rail one has a 19v @ 190w and 12v @60w option.  My proc is an i7-7700T.  I'm upsampling PCM via HQPlayer and proc utilization is at 5%.  The actual wattage utilization for the entire server is around 20-25w.

 

You can't just add up the watt ratings for each device.  That's a worst case scenario.  Adding up all devices in my server my total watts are 70w but I'm using 25w max.  Power from the SR7 is not an issue.

 

Using a 95w proc and upsampling to DSD512 with HQPlayer would increase your requirements, but nothing the 19v rail can't handle.  You're not going to exceed that 190w utilization.

 

May I know is proc i7-7700T capable of upsampling to dsd256?

[tapatrick]

11 hours ago, austinpop said:

 

Not sure if this got answered? I was away from the thread for a day, and a lot of interesting posts!

 

I think I addressed your question in this post here: https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?do=findComment&comment=722622

 

But in a nutshell:

1. the sCLK-EX board uses a frequency synthesizer to generate the 4 independent frequencies that appear on it clock outputs.
2. The sCLK-EX board uses an internal oscillator to generate the reference frequency for its frequency synthesizer
3. Additionally, it is designed to accept an external reference input, which can replace the output of its internal oscillator
4. Assuming the external reference has ultra low phase noise - like the Mutec, the Cybershaft, or the upcoming SOtM OCX-10, this low phase noise gets "inherited" by the sCLK-EX outputs.

 

@austinpop thanks for clarifying even more how the clocking works. This is a compelling thread and I'm still trying to get my head around all the spaghetti to see how/when I may invest in it. What I'm puzzled by is buying a sCLK board and then reclocking it by another new master clock.  I presume down the line this could be purchased as one integrated item? I mean master clock and sCLK-ex...

[Johnseye]

4 hours ago, lateboomer said:

 

May I know is proc i7-7700T capable of upsampling to dsd256?

 

I haven't tried with this proc.  My current DAC doesn't support DSD256.  Perhaps someone else with this proc can answer.

[afrancois]

7 hours ago, austinpop said:

Over on the Lush thread, someone mentioned how they knew their system was sounding good when they could listen to Adele's 21 without wincing. I'm paraphrasing.

 

That got me thinking. After all,  the clock improvements in trifecta medley have so dramatically removed that digital haze and fatigue. It's been years since I listened to 21. Would my system be up to the Adele test? 

 

I'm happy to report - YES! I ended up listening to the whole album. On my system. With HD800 headphones, that do not suffer bright recordings gladly. The whole album.

 

If that doesn't epitomize what we mean by smoothness and lack of digital glare, I can't imagine what does.

What I have noticed as my system got better (less jitter) is that highly compressed music became listenable. Adele’s CD's are highly compressed and her voice has been digitally manipulated to emphasize the typical Adele sound and this at a point that it has become ridiculous on Adele 25.

[austinpop]

1 hour ago, tapatrick said:

@austinpop thanks for clarifying even more how the clocking works. This is a compelling thread and I'm still trying to get my head around all the spaghetti to see how/when I may invest in it. What I'm puzzled by is buying a sCLK board and then reclocking it by another new master clock.  I presume down the line this could be purchased as one integrated item? I mean master clock and sCLK-ex...

 

It's hard to say. It depends on demand. Remember - it's all about cost. Reference clocks are expensive - even the economical ones. What fraction of people who pay for sCLK-EX are willing to take the next step to a reference clock?

 

If it's 90%+, say, then a business case exists for vendors to offer integrated solutions. If, on the other hand, the number is more like 20%, then today's situation - the modular approach - is likely to endure.