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

[elan120]

With recent success from Alan and Mozes on sCLK-EX implementation, along with all previous successes from others, it appear that two or more re-clocking cycles through sCLK-EX is one of important keys to greatly improve sound qualities.  This had me thinking perhaps I can try one of two configurations listed below and hope to get your inputs on which likely be better choice or other suggestions.

 

1.  PC (sCLK-EX output-1 to replace PC system clock / sCLK-EX output-2 to replace stock txUSBexp oscillator) > Singxer SU-1 (sCLK-EX output-3 to replace USB input clock (24.000MHz) / output-4 to replace both 22.5762MHz/24.576MHz oscillators with SEL0 connected).

 

2.  PC (sCLK-EX output-1 to replace stock txUSBexp oscillator / sCLK-EX output-2 to replace stock ) > Singxer SU-1 (sCLK-EX output-3 to replace USB input clock (24.000MHz) / output-4 to replace both 22.5762MHz/24.576MHz oscillators with SEL0 connected).

 

My thought for choice 1 is having good clock starting from PC system clock all the way to DDC before DAC, and choice 2 will be a little easier to implement by not replacing the mobo system clock, but still having 4 re-clocking cycles done before DAC.  Both methods assume keeping re-clocking cycles from sCLK-EX as close to DAC as possible will yield better sound quality.

 

 

[LTG2010]

17 minutes ago, elan120 said:

My thought for choice 1 is having good clock starting from PC system clock all the way to DDC before DAC, and choice 2 will be a little easier to implement by not replacing the mobo system clock, but still having 4 re-clocking cycles done before DAC.  Both methods assume keeping re-clocking cycles from sCLK-EX as close to DAC as possible will yield better sound quality.

The clock on the motherboard is a source of jitter - timing error. If you can replace this then go for option 1. The sCLK -EX isnt just a clock- filters, regulators, all help - a cleaner signal.

[mozes]

36 minutes ago, elan120 said:

With recent success from Alan and Mozes on sCLK-EX implementation, along with all previous successes from others, it appear that two or more re-clocking cycles through sCLK-EX is one of important keys to greatly improve sound qualities.  This had me thinking perhaps I can try one of two configurations listed below and hope to get your inputs on which likely be better choice or other suggestions.

 

1.  PC (sCLK-EX output-1 to replace PC system clock / sCLK-EX output-2 to replace stock txUSBexp oscillator) > Singxer SU-1 (sCLK-EX output-3 to replace USB input clock (24.000MHz) / output-4 to replace both 22.5762MHz/24.576MHz oscillators with SEL0 connected).

 

2.  PC (sCLK-EX output-1 to replace stock txUSBexp oscillator / sCLK-EX output-2 to replace stock ) > Singxer SU-1 (sCLK-EX output-3 to replace USB input clock (24.000MHz) / output-4 to replace both 22.5762MHz/24.576MHz oscillators with SEL0 connected).

 

My thought for choice 1 is having good clock starting from PC system clock all the way to DDC before DAC, and choice 2 will be a little easier to implement by not replacing the mobo system clock, but still having 4 re-clocking cycles done before DAC.  Both methods assume keeping re-clocking cycles from sCLK-EX as close to DAC as possible will yield better sound quality.

 

 

Hi Elan

I recommend that you reach out to SOTM and share Your thoughts with them. They are helpful and will advice you on what is best. For example, they believe ive that it is important to mod the 2 LAN clocks  on the motherboard and I honestly don’t know why it matters. Having said that, when I listen to my server, I don’t question their expertise  

[elan120]

3 minutes ago, LTG2010 said:

The clock on the motherboard is a source of jitter - timing error. If you can replace this then go for option 1.

Thank you for the input, I have been thinking about this for a while also, but the trifecta implementation seem to suggest not replacing system clock but having re-clocking done later in the chain can offset the noise/jitter from system clock, and that is the reason why I wonder if option 2 would have the same effect just like option 1. 

[elan120]

6 minutes ago, mozes said:

Hi Elan

I recommend that you reach out to SOTM and share Your thoughts with them. They are helpful and will advice you on what is best. For example, they beleive that it is important to mod the 2 LAN clocks  on the motherboard and I honestly don’t know why it matters. Having said that, when I listen to my server, I don’t question their expertise  

Hi @mozes, thank you for the suggestion, and I did reach out to SOtM for their input last night, but still waiting for their response, likely be soon.  My recent experience with LAN clock is very positive, and was very surprised with the improvement by using high quality ethernet card, I am hopeful by implementing sCLK-EX clocks to other areas as the two options listed earlier will help improve sound quality further.

[LTG2010]

12 minutes ago, elan120 said:

Thank you for the input, I have been thinking about this for a while also, but the trifecta implementation seem to suggest not replacing system clock but having re-clocking done later in the chain can offset the noise/jitter from system clock, and that is the reason why I wonder if option 2 would have the same effect just like option 1. 

That's the million dollar question, depends how good your motherboard clock is I suspect the difference will not be huge, but as @ moses suggested contact May @ SOTM she is usually most helpful. Edit just seen your reply, yes they'll be fast asleep

[mozes]

7 minutes ago, elan120 said:

Hi @mozes, thank you for the suggestion, and I did reach out to SOtM for their input last night, but still waiting for their response, likely be soon.  My recent experience with LAN clock is very positive, and was very surprised with the improvement by using high quality ethernet card, I am hopeful by implementing sCLK-EX clocks to other areas as the two options listed earlier will help improve sound quality further.

Also based on my experience, modding the clock of the switch that feeds ethernet to your server is very impactful. This is only for streaming. I know that you only have 4 outputs the optimize, but thought you should know about this just in case.

[elan120]

10 minutes ago, LTG2010 said:

That's the million dollar question, depends how good your motherboard clock is I suspect the difference will not be huge, but as @ moses suggested contact May @ SOTM she is usually most helpful. Edit just seen your reply, yes they'll be fast asleep

As I am eagerly waiting for replies from May from SOtM, I hope option 2 will yield the same result as option 1 since that will be much easier for me to duplicate the result than having to change out the system clock, especially when I finalize all the components I want for my new computer build.

[elan120]

2 minutes ago, mozes said:

Also based on my experience, modding the clock of the switch that feeds ethernet to your server is very impactful. This is only for streaming. I know that you only have 4 outputs the optimize, but thought you should know about this just in case.

Thank you for the suggestion.  This is one thing I have also been thinking about, but you are right, with only 4 clock outputs from sCLK-EX, I will leave the switch mod alone for now until I get these 4 clock outputs finalized.

[Alan_J]

23 hours ago, Johnseye said:

 

I may be missing something in how this is connected.  Are you using a cable to come out the tX-USBexp then into the tX-USBhubIN then out of the tX-USBhubIN to your DAC?  That is how I would envision these devices working.  Merely having the tX-USBexp plugged into the PCI slot and connected to the sCLK-EX doesn't do anything.  Coming off the motherboard direct to the tX-USBhubIN bypasses the tX-USBexp.

 

You are right. An internal usb cable comes out of tX-USBexp then goes into tX-USBhubIN. For this, SOtM modded tX-USBexp to make an internal usb connector. In the May's photo I attached, the yellow cable comes out of the newly created connector on the back of tX-USBexp and goes into tX-USBhubIN.

[Johnseye]

25 minutes ago, Alan_J said:

 

You are right. An internal usb cable comes out of tX-USBexp then goes into tX-USBhubIN. For this, SOtM modded tX-USBexp to make an internal usb connector. In the May's photo I attached, the yellow cable comes out of the newly created connector on the back of tX-USBexp and goes into tX-USBhubIN.

 

Very smart move.  I like it and wish I had done the same thing.  Nice!

[Alan_J]

21 hours ago, rickca said:

Is this really equivalent to having an external tX-USBultra?  I thought there were differences between the tX-USBhubIN and the tX-USBultra with respect to filtering.

 

Could this chaining be done with two tX-USBexp cards, or can that card not accept a chained internal input?

 

I thought @romaz was using tX-USBhubIN for input and tX-USBexp for output to his DAC.  So you are using the tX-USBexp exclusively for reclocking/filtering, right?  Where are your music files?

 

What you're doing seems clever, but I'm not sure I understand it.

 

I'd also like to know some specs of your PC ... CPU and motherboard.  Thanks.

 

 

tX-USBultra is basically tX-USBhub(IN or EX) + sCLK-EX. tX-USBexp is quite different from tX-USBhubIN in design.

tX-USBexp is designed to receive data from PCI bus, so it will require more modification to chain two tX-USBexps even if it's possible. In addition, I thought the different designs of exp and hubIN might have some complementary effects.

exp and hubIN are used exclusively for usb output.

My music files are in a hard disk inside PC.

My 3-year-old PC is i3 common Asus microATX MoBo win 10 with no software optimization. Some cares were taken to absorb noise, and power supply is TDK Lamba (still SMPS). System clock was replaced with C-clock, which is a tiny board. It has some positive effect, but not like sCLK-EX. It costed me around $1000 USD.

[Alan_J]

I mean $1000 USD for the whole PC except 3 SOtM cards.

(PS is TDK Lambda.)

[j3294]

how much did you have to pay for the whole modification of the 3 SOtM cards?

 

[Alan_J]

19 hours ago, j3294 said:

how much did you have to pay for the whole modification of the 3 SOtM cards?

 

 

Mostly the price of the 3 SOtM cards. Total around $1500.

[Alan_J]

On 12/20/2017 at 5:30 PM, j3294 said:

how much did you have to pay for the whole modification of the 3 SOtM cards?

 

 

Mostly the price of the 3 SOtM cards. (Actually $1600 USD)

This is what May sent me:

sCLK-EX(2points) : USD750

tX-USBexp : USD350

tX-USBhubIN : USD350

Installation/modification cost : USD150(USD50/a point)

[firedog]

On 10/9/2017 at 10:30 PM, ElviaCaprice said:

LOL, it's a disease building high power PC's.  Did it for many years myself, thus I can see how the computer audiophile (especially beginners) find it compelling to go the HQP route with these kind of builds.   Just like big ass speakers and amps.  The bigger the better sound.  NOT

It was hard at first for me to scale back my PC mentality and go the low power route for audio.

 

I think it's a mistake to generalize. Take a maker like Innous - they use "off the shelf" MOBO's. But, they claim they painstakingly measure basically everything on the market (slight exaggeration) and find the boards that are low noise. Acc'd to them, some of these exist, and you don't need an "audiophile" board if you use them and then optimize OS, etc. 

[Johnseye]

On 10/9/2017 at 2:30 PM, ElviaCaprice said:

LOL, it's a disease building high power PC's.  Did it for many years myself, thus I can see how the computer audiophile (especially beginners) find it compelling to go the HQP route with these kind of builds.   Just like big ass speakers and amps.  The bigger the better sound.  NOT

It was hard at first for me to scale back my PC mentality and go the low power route for audio.

 

7 hours ago, firedog said:

 

I think it's a mistake to generalize. Take a maker like Innous - they use "off the shelf" MOBO's. But, they claim they painstakingly measure basically everything on the market (slight exaggeration) and find the boards that are low noise. Acc'd to them, some of these exist, and you don't need an "audiophile" board if you use them and then optimize OS, etc. 

 

Innuos is not using a high power PC.  One of their servers runs between 10-15w and their high end one runs at 12-25w.  They also advertise a low ripple LPSU.  They are on the same page, doing the same thing we are.  We are using low power, off the shelf boards with LPSUs and sharing our experiences with them here.

 

After using my latest board for a couple months, while it is good practice to keep it low power, I do miss some of the speed and pep my previous proc had.  My current board has a built in Celeron with a 1.6Ghz proc IIRC.  It's rated at 6w and while I haven't measured it with a Kill-a-watt yet I'm guessing it uses all 6.  I'll measure it this weekend.  My previous proc was an i7 rated at 35w.  That entire server never used more than about 25w.  Where the Celeron is lacking is its ability to quickly analyze files.  Anyone who uses Roon or JRiver know that when you import new music it analyzes the tracks.  It is painstakingly slow with the Celeron.  It also doesn't help that I was cleaning up metadata recently and have 55,000 tracks to analyze.  I'm churning through about 10k per day where in the past this would have just taken a few hours.

 

So there are trade offs.  Windows installation time is another, but fortunately that's usually one and done unless you're doing a lot of testing of various OS's.  Linux, and ROCK in particular is always a fast install.  Whether there's a difference in SQ between the 6w Celeron vs the 35w i7, on their own and not taking into consideration the other motherboard factors, is unknown to me at this point.  There was a difference in SQ between these two unmodified server builds, with the Celeron Jetway board sounding better.  I can only guess it's because of the lower power.

[firedog]

1 hour ago, Johnseye said:

There was a difference in SQ between these two unmodified server builds, with the Celeron Jetway board sounding better.  I can only guess it's because of the lower power.

I'm not an expert and not arguing with you, but I still think it is difficult to generalize. There are some makers that say they get better sound with high powered boards. I'm guessing there are some implementations that work better with low power and some with high power. 

[ElviaCaprice]

I think you need to look at it as a system wide approach, low power.  The synergy it offers and the simplicity of the results with the overall SQ.  Eliminating high power components and simplifying the component chain can have a dramatic effect not only on the pocket book but the over all SQ.  More power doesn't translate to better SQ in audio.  Especially on the nearfield side.

sCLK-EX server => Chord DAC => Hi sensitive speakers

[Johnseye]

39 minutes ago, firedog said:

I'm not an expert and not arguing with you, but I still think it is difficult to generalize. There are some makers that say they get better sound with high powered boards. I'm guessing there are some implementations that work better with low power and some with high power. 

 

I agree, it is difficult to generalize.  The theory behind low power is that high power introduces more noise and is more difficult to power with an LPSU.  Using a switching PSU again introduces more noise.  Are there benefits that components with higher power bring to the table, yes.  One I can think of is with upsampling to DSD512.  However this is the only situation I know of where more power can be used for better SQ.  Otherwise introducing more power means more noise.

[mozes]

3 minutes ago, Johnseye said:

 

I agree, it is difficult to generalize.  The theory behind low power is that high power introduces more noise and is more difficult to power with an LPSU.  Using a switching PSU again introduces more noise.  Are there benefits that components with higher power bring to the table, yes.  One I can think of is with upsampling to DSD512.  However this is the only situation I know of where more power can be used for better SQ.  Otherwise introducing more power means more noise.

After my experience with Dave and sPS-500, I no longer believe that linear is always superior to switching PSUs.

[Johnseye]

4 minutes ago, mozes said:

After my experience with Dave and sPS-500, I no longer believe that linear is always superior to switching PSUs.

 

I don't know about the Dave, but my understanding is that the sPS-500 has excellent filtering, as does most of SOtM's products.  Lee's a noise filter expert wherever that's applied.  I don't think that the type of filtering he does is applicable to many switching PSUs.  Those appear to be exceptions.  Linear power supplies do not generate high frequency noise in the first place.  These power supplies are bigger, heavier and more expensive.  You need to consider the effect the filter has.  It's a requirement to eliminate the noise of an switching supply and the impact of the filter needs to be considered wholisitically. 

[mozes]

9 minutes ago, Johnseye said:

 

I don't know about the Dave, but my understanding is that the sPS-500 has excellent filtering, as does most of SOtM's products.  Lee's a noise filter expert wherever that's applied.  I don't think that the type of filtering he does is applicable to many switching PSUs.  Those appear to be exceptions.  Linear power supplies do not generate high frequency noise in the first place.  These power supplies are bigger, heavier and more expensive.  You need to consider the effect the filter has.  It's a requirement to eliminate the noise of an switching supply and the impact of the filter needs to be considered wholisitically. 

The only concern is maybe the high frequency noise that these switching PSUs kick back to the AC mains. To deal with this concern, I have Dave on a dedicated line by itself since it is my DAC/PRE/Amp.

[AmusedToD]

26 minutes ago, Johnseye said:

 

I don't know about the Dave, but my understanding is that the sPS-500 has excellent filtering, as does most of SOtM's products.  Lee's a noise filter expert wherever that's applied.  I don't think that the type of filtering he does is applicable to many switching PSUs.  Those appear to be exceptions.  Linear power supplies do not generate high frequency noise in the first place.  These power supplies are bigger, heavier and more expensive.  You need to consider the effect the filter has.  It's a requirement to eliminate the noise of an switching supply and the impact of the filter needs to be considered wholisitically. 

 

Well, that’s not new, even in high end audio. For example Jeff Rowland is a firm believer in switching mode power supplies with filtering. All of his products in the current lineup (even the M925 monoblocks with a price of $50k or more) use high efficiency SPMS.