Audio Grade Router and Switches in RMAF with SOtM

[plissken]

26 minutes ago, ismewor said:

most of the audio folks saying power cord won't make a different. And look at it now.

 

And now it still doesn't... 

[plissken]

 

 

 

4 hours ago, d_elm said:

If a data stream stays in the digital domain then clock jitter is handled by the circuit design.  It is when the data stream is processed by a DAC the upstream handling can become important for the generated analog signal.  In the thread I linked to many people describe their findings.  One finding is that good upstream clocks make a difference but only if not followed by a bad clock.

 

All I see is a thread full of people that don't understand how buffered systems, clock domain boundaries, FiFo buffering works.

 

So if we use your assertion that it's cumulative clocks then what about all the clocks on the routers between here and www.tidal.com

 

Hop  RTT    Lost/Sent = Pct  Lost/Sent = Pct  Address
  0                                           ########.cinci.rr.com [192.168.0.117]
                                0/ 100 =  0%   |
  1    1ms     0/ 100 =  0%     0/ 100 =  0%  192.168.0.1
                                0/ 100 =  0%   |
  2    8ms     0/ 100 =  0%     0/ 100 =  0%  142.254.147.213
                                0/ 100 =  0%   |
  3    9ms     0/ 100 =  0%     0/ 100 =  0%  be63.lsvnkyfb01h.midwest.rr.com [74.128.8.17]
                                0/ 100 =  0%   |
  4   13ms     0/ 100 =  0%     0/ 100 =  0%  be20.lsvqkydb01r.midwest.rr.com [65.29.31.24]
                                0/ 100 =  0%   |
  5   25ms     0/ 100 =  0%     0/ 100 =  0%  be24.clevohek01r.midwest.rr.com [65.189.140.166]
                                0/ 100 =  0%   |
  6   32ms     0/ 100 =  0%     0/ 100 =  0%  bu-ether17.vinnva0510w-bcr00.tbone.rr.com [66.109.6.70]
                                0/ 100 =  0%   |
  7   30ms     0/ 100 =  0%     0/ 100 =  0%  0.ae1.pr1.dca20.tbone.rr.com [107.14.17.210]
                                0/ 100 =  0%   |
  8   29ms     0/ 100 =  0%     0/ 100 =  0%  24.27.236.44
 

 

But yet I've read posts about how Tidal sounds just as good as playing a local file. 

 

I'll come out to anyone's setup and take a .wav file and make one copy local storage, another over a network and using foobar ABX plug in show you that this is 100% bullshit. 

 

I'll give you $2000 if you can hit 18/20  and if you can't just pay my travel expenses. 

[Charente]

I think I get what @plissken and others are saying ... I'm no expert, but makes sense to me. There is some debate on other threads that lead to the notion that effects of leakage current could manage to 'jump' ethernet transformers and potentially could end up as 'noise' downstream. So, no effect of the music signal, as demonstrated, but possible effects on say, the DAC. I am sceptical about high priced mods to switches, but is this what it is they are trying to address? Apologies, if my point is not very clear.

[plissken]

The T/O XCO argument being forwarded is one of Jitter. Not leakage currents. 

[marce]

56 minutes ago, ismewor said:

Yes, it sure make a huge different if it done right, that is why we are now tracing to replace the clock on the mobo. 

perhaps one of the main contributors to simultaneous switching noise (the main background noise of any microprocessor based system) is DDR memory, that's why DDR3 has spread spectrum ability. The clocks on PC's are critical, the routing of the clocks is critical, changing one is not just plugging another clock in on the end of a piece of wire... The chances are the length of wire and possible return path discontinuity will add as many problems as it is supposed to solve. Why do we put clocks as close to the input pins as possible, shortest routing (except DDR3+ where the clock is the longest signal, but routed as a diff pair with minimum skew, 0.25mm), minimal vias, minimal impedance changes in the signal path... 

Unless the system is designed to be operated by a master clock, but that requires a whole system exercise in clock distribution... With clock buffers etc.

[ismewor]

1 minute ago, marce said:

Why do we put clocks as close to the input pins as possible, shortest routing (except DDR3+ where the clock is the longest signal, but routed as a diff pair with minimum skew, 0.25mm), minimal vias, minimal impedance changes in the signal path... 

I agree and that is very true. 

[plissken]

7 minutes ago, marce said:

perhaps one of the main contributors to simultaneous switching noise (the main background noise of any microprocessor based system) is DDR memory, that's why DDR3 has spread spectrum ability.

 

I went into perfmon on a Windows 10 machine with almost everything that I could disable process wise, well disabled, I showed all the caching operations that was still going on.

 

There is just a ton of system activity.

 

Quote

The clocks on PC's are critical, the routing of the clocks is critical, changing one is not just plugging another clock in on the end of a piece of wire... The chances are the length of wire and possible return path discontinuity will add as many problems as it is supposed to solve. Why do we put clocks as close to the input pins as possible, shortest routing (except DDR3+ where the clock is the longest signal, but routed as a diff pair with minimum skew, 0.25mm), minimal vias, minimal impedance changes in the signal path... 

 

I pointed this out in another thread when they showed modified switches with a TXCO on a separate assembly with what looked to be about 16 inches of round trip lead wire. 

 

 

[marce]

10 minutes ago, Charente said:

I think I get what @plissken and others are saying ... I'm no expert, but makes sense to me. There is some debate on other threads that lead to the notion that effects of leakage current could manage to 'jump' ethernet transformers and potentially could end up as 'noise' downstream. So, no effect of the music signal, as demonstrated, but possible effects on say, the DAC. I am sceptical about high priced mods to switches, but is this what it is they are trying to address? Apologies, if my point is not very clear.

Ethernet, more precisely base band Ethernet we use around our homes (and other places) is designed to work with long cables, long cable make effective antennas, as well as lots of parallel cables to introduce crosstalk and due to the distances the relative "grounds" of two systems could be at quite different voltages. So the interface is designed to be robust in rejecting both CM and DM noise hence the transformer and common mode choke interface at every connection point.

[marce]

15 minutes ago, plissken said:

 

 

 

I pointed this out in another thread when they showed modified switches with a TXCO on a separate assembly with what looked to be about 16 inches of round trip lead wire. 

 

 

The 1ns rule...

Some rule of thumb:

http://www.hottconsultants.com/techtips/rulesofthumb.html

So an 8" run even if it was stripline (cables will differ, but will still be between 0.1-0.3ns)) would add about 0.2ns propagation delay, are we going to get pico second jitter figures with that length, and how closely coupled was the return path and finally what about the cable capacitance effecting the clock, longer cable higher C, add the extra inductance and you get filtering of the high frequency content of any wave, add the receiver and transmitters capabilities into the equation and you have fun...

[Speed Racer]

.

1 hour ago, plissken said:

 

 

 

 

All I see is a thread full of people that don't understand how buffered systems, clock domain boundaries, FiFo buffering works.

 

So if we use your assertion that it's cumulative clocks then what about all the clocks on the routers between here and www.tidal.com

 

Hop  RTT    Lost/Sent = Pct  Lost/Sent = Pct  Address
  0                                           ########.cinci.rr.com [192.168.0.117]
                                0/ 100 =  0%   |
  1    1ms     0/ 100 =  0%     0/ 100 =  0%  192.168.0.1
                                0/ 100 =  0%   |
  2    8ms     0/ 100 =  0%     0/ 100 =  0%  142.254.147.213
                                0/ 100 =  0%   |
  3    9ms     0/ 100 =  0%     0/ 100 =  0%  be63.lsvnkyfb01h.midwest.rr.com [74.128.8.17]
                                0/ 100 =  0%   |
  4   13ms     0/ 100 =  0%     0/ 100 =  0%  be20.lsvqkydb01r.midwest.rr.com [65.29.31.24]
                                0/ 100 =  0%   |
  5   25ms     0/ 100 =  0%     0/ 100 =  0%  be24.clevohek01r.midwest.rr.com [65.189.140.166]
                                0/ 100 =  0%   |
  6   32ms     0/ 100 =  0%     0/ 100 =  0%  bu-ether17.vinnva0510w-bcr00.tbone.rr.com [66.109.6.70]
                                0/ 100 =  0%   |
  7   30ms     0/ 100 =  0%     0/ 100 =  0%  0.ae1.pr1.dca20.tbone.rr.com [107.14.17.210]
                                0/ 100 =  0%   |
  8   29ms     0/ 100 =  0%     0/ 100 =  0%  24.27.236.44
 

 

But yet I've read posts about how Tidal sounds just as good as playing a local file. 

 

I'll come out to anyone's setup and take a .wav file and make one copy local storage, another over a network and using foobar ABX plug in show you that this is 100% bullshit. 

 

I'll give you $2000 if you can hit 18/20  and if you can't just pay my travel expenses. 

 

 

...and what happens when packets arrive out of order? Changing out switch clocks and caps is crazy! The power supply used by a switch might make a difference....but this is getting out of hand.

[d_elm]

17 minutes ago, Speed Racer said:

.

 

 

...and what happens when packets arrive out of order? Changing out switch clocks and caps is crazy! The power supply used by a switch might make a difference....but this is getting out of hand.

Just a question, what happens to noise generated by the switch (or downstream FMC), either from power or clock ?  I hear it downstream in my DAC.

[Speed Racer]

21 minutes ago, d_elm said:

Just a question, what happens to noise generated by the switch (or downstream FMC), either from power or clock ?  I hear it downstream in my DAC.

 

Of course you do....because you expect to.....

[plissken]

24 minutes ago, d_elm said:

Just a question, what happens to noise generated by the switch (or downstream FMC), either from power or clock ?  I hear it downstream in my DAC.

 

Years ago I took a TRS to XLR cable. On the XLR side I floated Ground (pin 1). You could hear mouse movement, HD access, Cache fill (using a RAM Disk). 

 

It was a system that was purposefully designed incompetently.  You should do the same thing with your DAC that I did with my cable: I fixed it. In my case it was to solder pin 1 backup. In your case it would most likely be a new DAC. 

[d_elm]

@speedracer and @plissken, you are not helping with your quick shots.  Bye.

[plissken]

Ok. Bye.

[Ralf11]

Let's forget about clocks for a minute.

 

Can you please describe the methodology for your blind listening tests?

[marce]

9 hours ago, d_elm said:

Just a question, what happens to noise generated by the switch (or downstream FMC), either from power or clock ?  I hear it downstream in my DAC.

You need a galvanicly isolated switch... Oh wait they all are...

[plissken]

Bottom line is the OP asked a question and has gotten into a huff because the answers don't fit the preconceived answer they already had when they asked the question. 

[MikeJazz]

On 12/10/2017 at 9:11 PM, ismewor said:

Just came back from RMAF this pass weekend. That is one thing i want to share. I was at the SOtM room and beside all the clock display and setup they have. I also notice the new audio switch inside their room.

Have anyone saw that too. They have TCXO and OCXO clock switches and the good thing is all the clock module is build in. Even the OCXO is build in. Wow that is awesome. And it also come with Blue and Red color as i remember. Plus they also include their linear power supply as well. That is a plus since i don't have to shop for linear power supply.  I will dig through my photo and try to post the look of it. 

 

Sorry if I sound ceptic.

But what on earth could be the impact of these better clocks in an asyncronous packaget transmission of data? 

[marce]

Well...

[ismewor]

This is what we call Audiophile world 

[marce]

Is that a song title...

[Ralf11]

7 hours ago, plissken said:

Bottom line is the OP asked a question and has gotten into a huff because the answers don't fit the preconceived answer they already had when they asked the question. 

 

 

He should have posted in a "safe space" like the tweaker thread

[ismewor]

On 10/24/2017 at 2:37 PM, d_elm said:

Just a question, what happens to noise generated by the switch (or downstream FMC), either from power or clock ?  I hear it downstream in my DAC.

Well, I would say not many notice of that. and you should be graceful for it. If you don't mind point out your setup will see if there something it can be done to isolate it. 

[ismewor]

Just got my TCXO router from the linear solution tonight. Wow, I was surprise by the packaging and craftsmanship of this company. Been busy and I’ll try to compare with my current router with LPS along with SOtM Trifecta. Hopefully I can get some  critical listening done over the weekend.