Master Clock for your EtherREGEN

[TomJ]

Hello guys,

 

Regardless of what values a clock has in terms of phase noise, what is ultimately relevant is what is output from the Etherregen via Ethernet and arrives at the receiver.

 

Are there any measurements that can prove an improvement of the differential signal (e.g. Ethernet jitter) and common mode noise when using an external clock?

 

 

[TomJ]

15 minutes ago, Mihaylov said:

I doubt it. It's also like you can't measure SQ.

You can measure that very well. I have been doing this for a while with other switches.

I have published several measurements in a German Audiophile Forum:

https://www-open--end--music-com.translate.goog/forum/privatforen/thomas-michael-rudolph-tmr/651284-messungen-von-ethernet-infrastruktur-nur-lesen?_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=de&_x_tr_pto=wapp

 

I just have done measurements of the Etherregen with and without a 10MHz NeutronStar 3 10MHz Clock and will publish the results next.

[TomJ]

Hi guys,

 

my summary of the measurements of the EtherREGEN are currently published:
https://www.open-end-music.com/forum/pr ... post658415

 

I also measured the EtherREGEN with an external 10MHz clock to see if there is really any improvement on this.

 

Maybe this will interest one or the other.

 

I had also requested here in with the forum owner an article concerning the topic Ethernet sound with my measurements, on this unfortunately then no more answer received.
So you have to read the article with Google Translate and maybe accept small translation errors.

 

Best regards and happy reading,

 

Tom

 

 

[TomJ]

46 minutes ago, dylanesque said:

Thanks for sharing the results @TomJ

intriguing that ext. clock measures poor. 
Could you please shed light if the Neutronstar 3 is sq wave output? Would’ve been very interesting with Mutec Ref10120 or Adark. 

Its sq wave.
But the users of this clock report a similar sound improvement as with the afterdark giesemann, but not as strong.

 

And here is a statement of Ed Meitner regarding external clocks (for DACs, but this is the same for switches):

 

https://www.stereophile.com/content/meitner-emm-dv2-dsd-mqa-digital-audio

Quote:

JVS: You don't believe in using an external word clock. Why?

EM: Because I think this is the most stupid thing I've ever heard in the audio business. That means you have a precision clock that you have to connect to a wire to connect to a DAC, when the clock should be straight away where it belongs, inside the DAC, beside the DAC chip, if there is such a thing—not through a cable in a different box. This is so idiotic, it's not even funny. It's a money grab.

[TomJ]

30 minutes ago, Savolax said:

Did you find any improvement (or any difference) with external clock while listening?

As i am biased, i refer statements regarding sound improvements in this regard to the owners and users.

And as I said, they report an improvement in sound, which I do not question.

[TomJ]

9 hours ago, Superdad said:

Hi Tom (I thought you are Eric):

Hi Alex,

 

let me describe it in a little more detail:

 

I am Eric and had previously used the nickname Tom here on the forum. I will sign with Eric in the future to make it easier. My measurements and reports are posted in the open-ended music forum by Thomas-Michael Rudolph, as he is a friend in mind and has many years of experience in reducing noise in the power line. Like me, he is a wanderer between the worlds of science and audiophilia. He had offered to post the measurements in a read-only thread so that they would not be constantly fragmented by posts and trolls. Therefore, the forum member in the post itself is Thomas, but always names me as the author.

 

9 hours ago, Superdad said:

Thank you for your interesting work project with the measurements of switches--including the EtherREGEN--and the effect with clocks, power supplies, etc.

While we think you are facing in the right direction, there are considerable limitations with your methodology and test equipment.  I shared your report with @JohnSwenson and we had a long and interesting conversation about it.  

 

We may have a number of significant suggestions to guide you towards achieving more truthful and revealing measurements, both with respect to common-mode leakage, jitter, and ground-plane noise. Some of these may require a very modest investment of your time (and a little money) to build a more proper set-up that will reveal far more than what you are seeing. (Though there may be nothing we can do about the fact that your jitter measurements with that digital sampling scope are only ever going to be as good as the rather poor clock in the scope generating the eye-pattern. That is why all your numbers come out rather high.)

 

This is really great and really appreciated. I am grateful and open to any professional advice.

If he doesn't want to make the clues completely public in the first step, feel free to contact me via PM.

 

I have been on quite a journey with this topic. I started discussing with the hardware guys, but they declared me a nut because I hear things. Then I started discussing with the audio guys, but they treated me like a traitor just because I started measuring. In addition, which is always argued here only with half-knowledge, so that a discussion is difficult. (You will realize this if you read through the threads, in the German forum). This then brought me to Thomas in the open-end forum, because he understood my approach and showed interest in it.

Regarding the scope: The scope was currently used by 100base tx and also 1000base tx in the Ethernet area as standard for jitter measurements. It is sufficient to show differences between the switches (EtherREGEN is one of the best, which agrees with your statements).
It is not intended to measure phase noise of clocks - that is not my claim, but it is measured what comes out of the boxes.

 

9 hours ago, Superdad said:

We do see it as beneficial for us to engage with thoughtful and technically curious people like yourself. Because unlike closed-minded folks who already have their minds made up that there is nothing to see or hear, you come from a place where you have heard things and want to explore the technical reasons behind what is heard. 

 

Yes, you got it! I am not a copy of audio “science” review!
I don't want to prove that sound changes via ethernet are not possible, I'm just an incredibly curious technology and audio nerd who wants to know the reasons for this and no longer wants to be satisfied with myths and half-knowledge.

For a long time I was convinced that nothing can change on the way of the data. Until I heard sound differences in isolators and also LAN cables. I began to tinker with cables and insulators, but only design by ear, which was totally unsatisfactory in the long run. I then began to assemble my measurement equipment with the recommendation of Ethernet professionals. I also discussed with Galen Gareis and other pros.

9 hours ago, Superdad said:

Concerning the NewClassD Neutron Star 3 clock you tested with the EtherREGEN:

One the years I have yet to find ANY published phase-noise plot for their modules. And since they sell it as sort of a DIY board for users to implement, there seems to be (from the waveform graphs seem at the AH forum) a great variation in quality of even its basic output. Referring to our white paper on clocks (sine vs. square, impedance matching, cables, etc.) a poor squarewave clock will be worse than an average sine wave.

Regarding the NeutronSTAR 3 clock: I know there is no technical data, but all owners of the clock report a sound improvement with EtherREGEN. Even if the implementation is lousy (as seems to be the case with many clock upgrade switches). And the actual question is, how can a lousy implementation with worsen jitter and noise improve sound quality? One conclusion could be, that it is done by noise demodulation in the analog area.

By the way: Its a square-wave clock

 

9 hours ago, Superdad said:

P.S. I have been trying to register for user membership at both Aktives-Hoeren.de and at open-end-music.com where you and so many of our German clients and friends post. But I have so far been unsuccessful in completing the process (the registration confirmation e-mails never come). So I may ask you to share over at those sites some of our postings from here.  

 

This may have something to do with Google translate. I had provided my links as Google Translate links.
Please try again without Google Translate - otherwise I will inform the forum management. Jürgen from the Aktives-Hören Forum has apparently already contacted you personally outside the thread here. The guys have already discovered the communication here and are already curious what comes out.

 

I look forward to a constructive exchange with you and thank you for your feedback,

 

All the best,

 

Eric

[TomJ]

Hi Alex,

 

44 minutes ago, Superdad said:

I will mention that it would be much more interesting to look at jitter histograms than eye-patterns distilled to a single number—using an analyzer with a far lower jitter floor.  I get that he has said his measurements are meant to be valid only for comparative purposes between devices, but really we should be down in the tens of picosecond range. The specs on his scope—used for generating the eye-pattern—are up in the tens of nanosecond range. (And 1 nanosecond equals 1,000 picoseconds.)

 

meanwhile i have upgraded my scope with the TDS784D. This can also do histograms of jitter.

And here are the timebase system specs:

TDS784D Timebase System

Time bases	Main, delayed
Time base range	200 ps to 10 s/div
Time base accuracy	±25 ppm (over any interval ≥1 ms)
Delta time measurement accuracy	±(0.15/sample rate) + (25 ppm x [reading])
Trigger jitter	7 ps (typical)
Pre-trigger position	0% to 100% of any record
Delay between channels	≤50 ps (any 2 channels with equal V/div and coupling)

 

47 minutes ago, Superdad said:

Also, while Eric (TomJ) is looking at common-mode interference, the impedances of his probes are getting in the way, and also he ought to be looking FAR lower in the spectrum—where this stuff really matters.

 

Any help is warmly welcome!

As I mentioned in my thread, the entire setup including scope and also probes did not happend by accident.

It is exactly the setup described in the Intel Compliance Test for 10base tx, 100base tx and 1000base tx - including scope and probes.

https://community.intel.com/cipcp26785/attachments/cipcp26785/embedded-core-processors/2456/1/1000BASE_T 100BASE_TX 10BASE_T Physical Layer Compliance Tests Manual.pdf

Although the paper is from 2006, it should still be valid - it was valid in 2006 and is still currently provided.

Perhaps there are other requirements from an audio point of view, but as already mentioned, this should be better understood.

 

56 minutes ago, Superdad said:

I really must leave this all for John to explain, in large part because I simply do not have anywhere near the technical understanding or engineering insights into this.  He is the guru on such matters. I am merely the struggling pupil—with the memory of a goldfish. 

 

I'm really curious and eager to hear john's tips!

 

Eric

 

 

[TomJ]

P.S.: The jitter analysis and eye pattern is done, how explained also in the Intel Doc, with standard trigger rising edge, not with ethernet trigger and PLL. So to me it seems that the trigger jitter matters more, which is 7pS.

[TomJ]

Hi Sunny,

 

many thanks for your post.

 

 

But just to clarify: I am not performing a compliance test, but have orientated my measurement setup that way.
As far as jitter is concerned, this is sufficient to detect and evaluate differences. There is no "pass" from me.

 

 

And regarding my current thesis, I have summarized these here:

https://www.open-end-music.com/forum/privatforen/thomas-michael-rudolph-tmr/651284-messungen-von-ethernet-infrastruktur-nur-lesen?p=659024#post659024

But attention! Here is still the chapter missing regarding CM Noise, which is still in work.

 

Regarding jitter, we should really define what we mean by it before we discuss it further.

I only examine what goes over the cable into the end device - because that's what counts. There may be other effects that act via the power net in the background, but that's not my business.

And if we're talking about jitter in this area, then it's clearly defined, as it is for everyone in the industry, and so is the measurement method.

 

I'm not interested with my measurements in what is understood, interpreted or believed within the device about phase noise or jitter - if someone can explain how that effects across the cable, of course. What counts for me is what goes over the CAT cable.

 

Kind regards,

 

Eric

[TomJ]

5 hours ago, kennyb123 said:

EM is right about the clock belonging inside the box.  As far as his point about it being a money grab, I suspect that it’s aimed at companies who are direct competitors, like dCS. I can’t imagine that he’s considering gear that’s designed to hit an affordable price point.

 

As I reported earlier in this thread, I did some jitter measurements in comparison. External clock is jitter-wise not an upgrade, which I measured. 
I would be interested to know if there has ever been a measurement that proves that an external clock with all the problems with cables and plugs brings an advantage, except that an additional source of noise is created, which may produce a nicer sound for some.

 

Here my measurements with google translate: 

https://www-open--end--music-com.translate.goog/forum/privatforen/thomas-michael-rudolph-tmr/651284-messungen-von-ethernet-infrastruktur-switches-nur-lesen/page3?_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=de&_x_tr_pto=wapp#post658450

 

Regards,

Eric

[TomJ]

8 minutes ago, kennyb123 said:

All your measurements tell us is that with the test you constructed 'the EtherRegen with an external NeutronStar 3 shows worse values'.  The scientific method does not allow you to apply those results broadly and apply it to other clocks.  We also don't know if the measurements you elected to take would be able to explain why "many users of the EtherREGEN report that an external 10MHz clock improves the sound considerably".  We don't even know if the clock you used for your test would be one that would even falls into that category.

 

Please show me some measurements that shows, that an external clock improve Jitter of EtherRegen - otherwise the conversation is useless.

[TomJ]

It's clear, that I do not earn any credits in this forum for real investigation what really happens with Ethernet.

Just a little hint: There is something between your ears, which could be useful if used.

[TomJ]

11 hours ago, scolley said:

IMO this is a conversation better suited for the temple of measurement worship: Audio Science Review. In this forum making decisions based on what you hear generally has merit, and comments like “if you can’t measure it, it’s not real” is just trolling at worst, but off topic at best.

 

So, can we get back on topic please? Thanks.

Unfortunately, as always, I am misunderstood.
It is not about proving that external clocks are useless, but to find out why. Unfortunately I can't find anything in the net where the jitter value of Ethernet was measured except my own measurements. If now the jitter value is not improved at all - and with the realization by my measurements I am not alone (e.g. also Ed Meitner), the question is nevertheless where sound improvements come from. I would have thought that there is interest in this. But apparently not, because you immediately feel attacked again and would have to question his world view that you have built up over years. 

[TomJ]

@Clockmeister

You write "Measurements are only helpful to people who know how to draw the right conclusions from them." Unfortunately, this does not apply to you.

No matter what you measure at the clocks, in the end the Ethernet signal and its noise is what arrives at the receiver and can be responsible for the sound.

Why don't you measure this?

Either you don't know anything about Ethernet or you are afraid that it will turn out that all this external clock thing to be useless.

[TomJ]

Funny, there are disagrees to my last post, so apparently the laws of physics do not apply here in this thread.

The clocks can only work over ethernet and if you are not able to evaluate the signal here, everything that happens before is useless stuff.

 

It is always funny to see that guys in the field of Ethernet have no idea and try to apply the same approaches as in the field of DA conversion and think that the result is the same. this is nonsense.
I am sure that there are differences in sound over ethernet. Also by clocks, but not by better jitter value of a clock connected externally over cable. there the clock can have a still so good phase noise, that helps unfortunately nothing. 

[TomJ]

@kennyb123

It is always easy, if one does not want to or cannot engage in a factual discussion, to insinuate trolling.

 

You are discussing clocks here and this is an open forum where everyone can contribute with their view. There are measurements of clocks made here, but on the other hand said, the ears are the only measuring instrument. I had only remarked that if one starts to measure against the aversion to measurements, one measures nevertheless best there where it is relevant. I would like to be convinced of the opposite of my findings, but this is apparently no one in a position.

 

This is very sad because guys out there are investing thousands of dollars in clocks believing they will get better jitter value, which no one can seem to prove.

[TomJ]

If you are no scientists, why someone measure phase noise of a clock in this thread? Just let the ears judge - that's fine. But if you start measure, measure at the right point. And this is the Ethernet signal at the and of the day what is of interest - no matter the clock measures stand alone. You always have to look at the hole system, not the single puzzles.

You can have the best clock ever, but if implementation is done wrong, it's worse than a 50cent onboard clock and you add just extra noise to the chain - which can also influence sound.

 

 

[TomJ]

6 minutes ago, kennyb123 said:

I would rephrase this.  Measurements are only meaningful when one knows how to apply them.  

 

The measurement-chasers who troll on audio forums tend not to have the slightest clue about this.   They don't even care to know this because their goal isn't to try to understand what's really going on.  

 

I believe it comes from them resenting the fact that they are unable to hear this stuff themselves.  They pick measurements that allow them to believe that the problem isn't with them but with everyone else. The measurements are simply the weapon they use to try to devalue and criticize those who have what they lack.  There's a good reason why they steadfastly avoid letting their own ears decide and why they have zero regard for the findings from those who do.

Ones again, i here differences with different Ethernet stuff, also different clocks - but you are not able or not willing to look in the right corner where this comes from, because of lack of knowledge or lack of willingness.

[TomJ]

25 minutes ago, R1200CL said:

Mayby you find something here:

https://audiophilestyle.com/search/?&q=Phase noise&quick=1&author=JohnSwenson&search_and_or=and&sortby=relevancy

 

https://blog.bliley.com/ultimate-guide-to-phase-noise

 

 

 

I do understand phase noise and it's effect. The question is, is an external clock coupled with BNC cable and connectors capable to deliver better phase noise and so jitter than an internal optimal integrated clock and if not, where do sound improvements come from. But it seems that there is not much interest to understand the mechanism to get a clearer picture. So continue to play trial and error or follow the recommendations of dubious experts.

[TomJ]

15 minutes ago, Exocer said:

I have not exactly been following, but @TomJ did you use a clock with objectively better specs than the ER’s objectively high performance clock to test with? Without that assurance, it will be scientifically impossible to improve upon the ER’s internal clock.

 

If you were to use something like the AfterDark’s Queen for example or a mid tier CyberShaft like the one @Superdad uses we would then be able to start from a solid foundation of logic (just my .02).

There are no specifications available for the NeutronStar3 - this is very stupid. But the clock has a very good reputation in various audiophile forums.
I will soon get a Mutec Ref10 to measure with ER and have also improved my measurement setup with a newer 1GHz scope with better eye-pattern evaluation.
I would also like to measure an Afterdark, but so far nothing has been offered or lent to me.

I'm really not about trolling or bashing, I just want to know what's going on.

[TomJ]

29 minutes ago, kennyb123 said:

Those of us who have been able to rely on our own ears don't see this a bad thing.  It's part of what makes this hobby fun!  Sorry that you need measurements to tell you what you should expect to hear.

You still don't get it - I want to know why this is happening, not that it is happening .....
If your ears can explain it to me, then go ahead and explain.

[TomJ]

1 hour ago, kennyb123 said:

Clearly the reduction of the clock's phase noise is providing a sound quality improvement.

I know John's papers very well. 
Let's assume that the phase noise of a clock converts to voltage noise and thus enters the end device (although this has not yet been proven by measurement - but anyway, let's assume it).
Then the phase noise in the ER is relevant and thus the jitter value of the Ethernet signal, since this, as you surely know, is nothing else.
Your phase noise values should then be measured at the ER, i.e. at the outgoing Ethernet signal. Everything else is irrelevant. No matter how great your clock is regarding phase noise, what counts is what arrives at the ER. And it is simply not proven yet, no matter what your golden ears whisper to you, that an external clock with the suboptimal connection can do that at all. In addition, additional noise comes into the EtherRegen via the ground of your external clock, which migrates as common-mode interference into the EtherNet signal and also influences the sound.
 

[TomJ]

From my perspective and experience with all the gears I measured so far with external clocks or clock mods, PS is by far the better invest, as Sunny mentioned, If you don't want to do sound tuning by noise.

 

The things Sunny mentioned can all be seen in my measurements. And if there is a subjective sound improvement with this gears, I bet my a… that it comes from all the interferences these solutions inject.

[TomJ]

16 minutes ago, kennyb123 said:

If that’s your takeaway then you didn’t learn anything from John’s paper.  I’m going to drop now as it’s evident that you lack the humility needed to be able understand the limits of your own understanding and the limits of what you’ve been equipped to understand.  I’d have better luck explaining it to my parrots so I will focus instead on them.  
 

You won’t ever find the answer you claim to be seeking because all that matters to you is that you can explain it with the one measurement you think you know how to capture.

 

Why do you react so pissed off and have to discredit me all the time?
If you are so convinced that you pass John's paper and conversion of frequency noise into voltage noise as the basis of the migration of phase noise over the digital chain, then I am very surprised that you still can't follow me.

 

You should really talk better to your parrots, who parrot everything after you and don't ask unpleasant questions.

[TomJ]

Hi Alex,

 

many thanks for the detailed post!

I hope you were able to overcome the disaster in your garden in the meantime.

 

4 hours ago, Superdad said:

Running your tests (as limited as they are given that they only give a single jitter figure, and as I will attempt to explain below the mechanism that makes this stuff matter is with low-frequency-offset phase-noise) should yield different result with a Mutec REF10 than with the long-in-the-tooth Neutron Star3, for which nobody has ever run a proper phase-noise plot on. I suspect that its 10Hz offset performance into an EtherREGEN is worse than the EtherREGEN's Crystek CCHD-575 (one needs to get to about -120dBc/Hz @10Hz with a 10MHz clock just to equal the internal clock).

 

I must admit that it has bothered me that for nearly a year you have taken a stand that an external clock with EtherREGEN is worse--based solely on the use of the Neutron Star3. Such does not seem very objective to me...

 

I agree that the results with the NeutronStar are not necessarily representative. However, I am surprised that there are no other measurements with all the technical know-how. Thus, these findings of the measurement with the NeutronStar are the only serious clue for the discussion other than speculation.

 

However, the measurement of the Ethernet 100base TX  jitter is not that complicated. Regarding my equipment a few more words. The scope has a trigger jitter of 7pS, which should be sufficient to make a serious evaluation of Ethernet jitter - it was also apparently sufficient that chip manufacturers recommended this model for their Ethernet signal evaluation.

 

Meanwhile, I also have a limited histogram function for evaluation. I measured it again before I had to return the loaner (sorry for the poor image quality):

 

ER with internal Clock
PK-PK: 640pS
Standard deviation: 80.3pS



ER with Neutron Star 3
PK-PK: 660pS
Standard deviation:81,5pS

 

As already written, I will soon receive the Mutec Ref10 and am curious about the measurements.

 

But with all the discussion about how jitter affects the sound, it would certainly be helpful to first check whether the external clock offers a real advantage in this regard or other circumstances are responsible for sound changes. I thought that these investigations must interest all, which concern themselves hereby. Only for this one must measure in my opinion at the correct place and not only at the output of the clock, but at the output of the Etherregen.

 

Best Regards,

Eric