Master Clock for your EtherREGEN

[Superdad]

39 minutes ago, R1200CL said:

Any plans of adding these:

https://www.circuitstransistors.com/circuitstransistors?search=BLP-10.7-75

Fake site or ?

 

Yes, that is fake site with fake stock and prices.  Just like the place you e-mailed me about yesterday that showed 100,000 pieces of Linear Tech LT3045 regulators for just $1.25 each, and $90 Finisar SFP+ modules for $30.  Didn't your parents teach you that if something seems too good to be true it usually isn't? 

[Superdad]

1 hour ago, Rasputin said:

Then I introduced bg7 clock with stock dc cable no name psu and Canare bnc over square waive output. The SQ degraded quite a bit loosing life in the music. Changing the bnc cable for Acrolink did increase the SQ, but it was not better then ER alone.

 

Sorry, but even your $1,000 clock cable is not going to fix the square wave output of the BG7TBL clock enough to either justify that cable or to benefit the EtherREGEN as you could if you just use the sine wave output and this filter (right at the input of the EtherREGEN):

https://www.minicircuits.com/WebStore/dashboard.html?model=BLP-10.7%2B

 

Read our new paper, ignore that the linked filter is 50-Ohm, use any cheap coax cable for the clock, and put the money from selling your expensive clock cable into better power supples (sorry, I have one of those Studer900 and the performance is poor; does not even come close to meeting its output current rating and the output impedance is rather high).

 

I am not sure I follow those who think they have to spend big $$ in add-ons to the EtherREGEN. The vast majority of our clients--including some with 200,000+ systems--enjoy the EtherREGEN just as it is.  Sure you can enhance it, but that does not change the fact that we back more technical innovation (and expensive parts!) into it and its modest case at $640 than was we see in any of the $2,000+ switches popping up.  

 

Cheers,

--Alex C.

[Superdad]

34 minutes ago, PYP said:

Since Alex is busy with a thriving business and mentioned the mini-circuits low pass filter might not be his first priority, I ordered one from mini-circuits, specifically BLP-10.7-75+ Lumped LC Low Pass Filter, DC - 11 MHz, 75Ω to use with my Cybershaft.  

 

If you are in the US, the price is $36.45 + $12.31 ground Fedex = less than $50.

 

Puzzled about how you managed that since:

 

a) The 75-Ohm version BLP-10.7-75+ is currently a non-catalog item on their web site with no AddToCart button; But the 50-Ohm BLP-10.7+ will work just as well, even with 75-Ohm clock, EtherREGEN, and cable--or all mismatched. Referencing our paper, that's the whole point of putting this at the EtherREGEN's clock input, end of a noisy sine wave. The impedance and cable stuff cease to matter. 

 

b) I believe Mini-Circuits has a $50 minimum order (before shipping or sales tax).

[Superdad]

17 minutes ago, PYP said:

Since Alex has asked that thread participants, talk about cables elsewhere, I'll send a pm, if you don't mind, to ask about the costly cable you would recommend.

To clarify: I don’t care if you chat about clock cables (or whatever) in this thread. It’s the other thread where I posted John’s paper that I am trying to keep on-topic. 

 

And David @Clockmeister:

Please look to John’s paper and feel free to comment there. Part of the thrust of it is that a filtered-at-the-target (endpoint) sine wave clock dispenses with the need to worry much (if at all) about clock cable quality and impedance matching. 

Whereas with square wave clocks those things are very important to get right.

There are other points to the paper as well.

Best,

—Alex C.

[Superdad]

16 hours ago, Rasputin said:

I have found a huge difference among the bnc cables when used between DCS Rossini DAC and DCS clock, so the notion of a cable “cease  to matter” with a use of a filter goes against my  hearing experience, however I will try as advised in the John’s paper. 

 

That's because:

a) the dCS clock is a square wave--and cable will always matter for that;

b) what we said about cable not mattering much is only for sine wave and only for if you use a good low-pass filter at the target device input.

 

The above are not minor details. They are the essence of what we put forth. Would really hate to think that folks are going around thinking that John Swenson said clock cable don't matter at all. It is all there in the paper...

 

Cheers,

--Alex C.

 

[Superdad]

20 hours ago, Clockmeister said:

Genuine REAL world ultra low noise phase clocks that are really -120 @ 1Hz and have long term stability and those that do will have an oven or the very top oscillators will have a double oven possibly.

 

Aren't those double-ovens more for long term frequency stability and not so much for the benefit of phase-noise?

 

By the way, while that Holzworth looks like a nice machine (about $40K), it seems the less than half the price Jackson Labs PhaseStation (by John Miles of TimePod fame) which John bought last year has a much lower phase-noise measurement floor at low offsets 1Hz~10Hz):

 

http://www.jackson-labs.com/assets/uploads/main/53100A_introduction_1pg_revF.pdf 

https://www.holzworth.com/Portals/0/HA7062C_Web_Datasheet.pdf

 

Good chance your machine runs the cross-correlations faster though. 

 

Cheers to you David,

--Alex C.

[Superdad]

3 minutes ago, R1200CL said:

Well, I really doesn’t hope the same applies to the AfterDark clock you have. 

No it does not.

[Superdad]

1 hour ago, GMG said:

I don’t want to quantify and characterize the noise. Just want to see if there is anything severe on the line. Will I be able to see it on the wave form? I mean if the AC is perfect then I expect to see a perfect sine. If there is considerable noise I expect to see a distorted sine - wiggily, choppy, peak is not round but flat. Stuff like that.

or will such a cheap device only show the general shape of the wave form and frequency?

 

Even if you buy that inexpensive portable scope, PLEASE, PLEASE, PLEASE heed @Clockmeister's warning: NEVER CONNECT REGULAR SCOPE PROBES OR TEST LEADS DIRECTLY TO THE AC MAINS!!  Doing so will cause serious damage to both you and the scope!

Safest way to put a put a regular scope to the AC mains--other than buying an expensive high-voltage differential probe--is to use a small 24V step-down transformer.

 

But honestly, if you are not comfortable, familiar, or qualified working with test gear and high voltages, we advise that you let the idea of looking at your mains go for a while.  You probably won't like what you see anyway, and digging deeper into the waveforms to understand what is going on the potential sources of distortion takes some expertise. 

[Superdad]

4 hours ago, MartinT said:

Sadly, my oscilloscope is old and only 20MHz bandwidth

 

Wow, just how old?  Even the all vacuum tube Tektronix 543 I owned decades ago (built in 1966, just a few years after I was born) was a 30MHz scope. 

[Superdad]

33 minutes ago, MasterWarzombie said:

We buy a basic etherregen, which is actually not that good basic for audiophile experts because you have to add 4 times the price of etherregen to get a real result .....

 

That seems a bit cynical. 

 

a) Show us another Ethernet switch with anywhere near the technical innovation and parts cost relative to sale price.

 

b) As of today 2,200 people are happily enjoying the EtherREGEN in their music systems (including many with $100K+ invested), and only a small fraction of those feel compelled to add an LPS or an external clock.

 

c) If we were to get fancy and situate an EtherREGEN (with further functional and performance enhancements) in a large case with power supplies and top-end clock, it would have to retail for about $3K (with a build cost that would have other firms charge $5K for).

 

John and I would much rather have thousands of people able to enter and benefit at a low price than to only offer something extreme that far fewer can afford. 

 

We might go there someday—as we know we are leaving money on the table since we see there are those anxious to spend big bucks. But presently we are hard at work on two other projects using techniques and technology never before applied to USB. (I promise they will make some heads spin; but we don’t yet know the exact timeline so please don’t ask.)

 

It is more fun to change the game than to play by others rules and expectations! 

[Superdad]

3 hours ago, MasterWarzombie said:

I tease but I wonder about the feelings of each one.

 

Hi:

Rereading my earlier response to you I want to be sure that I did not hurt your feelings. 

I know and appreciate that you enjoy our products, and that you take seriously the never ending quest for musical perfection. 

 

My post was to address a complaint which we have seen before—and to dispel the notion that one must spend more money in order to realize the benefits of the EtherREGEN’s 100% unique design. 

 

We see see many other switches coming to market—some at reasonable cost, some at astronomical prices—yet when we look inside at the boards most all we see are conventional switch circuits with either/both power supply or unknown OCXO.

 

UpTone is more about innovation and addressing issues at the root cause.

Proof of that is the fact that we seem alone in attempting to explain the underlying electrical effects and why they are heard—as well as how our products work to cure the problems. 

(See our ‘white paper’ from a year ago:)

 

[Above is just the first page; full paper is here:

https://cdn.shopify.com/s/files/1/0660/6121/files/UpTone-J.Swenson_EtherREGEN_white_paper.pdf?v=1583429386 ]

 

(Am on my phone relaxing at campsite by a lake...)

 

[Superdad]

4 hours ago, R1200CL said:

What would it take ?

Any of these ?

Ha ha ha, not even close. Scopes and spectrum analyzers are not going to give a good look at phase noise (though with an expensive differential probe and a highly custom ultra-low-noise amp you might get a modest look at ground-plane noise).

 

The best price/exceptional performance device for measuring phase noise presently is this one (which John owns): http://www.jackson-labs.com/index.php/products/phasestation_signal_source_analyzer

About $19,000—and you still need to pair it with an OCXO that is better than what you are trying to measure. 

 

By the way, the above Jackson Labs unit was designed by the famous (well in timing circles anyway) John Miles. Mr. Miles designed the original TimePod (and the software to use it). When he could not keep up with demand at the bargain price of $5K, Symmetricom took it over, and the prices for various versions (with and without display) went to $19-32K, with Microsemi buying Symmetricom along the way—and of course Microchip swallowed them as they have with dozens of other firms. (BTW, the Symmetricom is what AfterDark and many other clock makers use.) John Miles was unhappy that his designs had become so expensive and out of reach. It seems that while the small Jackson Labs promised to make it affordable, it started at about $10-11K, but they must have either realized it was too costly to build and sell at that price or that they were not going to sell enough units to sustain and support it—so now the price is creeping towards $20K. Still a bargain for its capabilities.

 

What is cool is that Mr. Miles is there at Jackson Labs now and providing personal advanced support. @JohnSwenson has received replies directly from Mr. Miles!

[Superdad]

5 hours ago, LowMidHigh said:

You're endorsing their 75 ohm cable, but do you happen to know its attenuation? According to John, that's the most important measurement for a good ref clock cable. The fact their website lacks that information makes me wonder. 

 

Just a recap regarding clock cables:

—Really low attenuation across a broadband is most important only for square wave clocks; does not matter for sine wave clocks. And impedance matching matters.

 

—Really great shielding—with a semi-rigid tube being best—is most all that matters for a sine wave clock; and not even very much if the filter is used at the target device end.

And impedance matching does not matter with sine wave clocks. 

 

[Superdad]

Hey gents:

As we have explained several times, the impedance of the Mini-Circuits filter does not matter for this application with sine wave clock. Buy and use whichever one you can get! 

I we were not so busy here I’d order 100 pieces and resell them as a service to UpTone clients since I can ship around the world for just $13 (First Class Airmail) or with some faster and slightly more expensive FedEx or Priority Mail options.

[Superdad]

40 minutes ago, sgr said:

I have two 75 ohm ERs. It seems if adding an external clock the 50 ohm version is better as sine wave is much less quirky when it comes to cables.

As we keep explaining, impedance matching (of devices, cables, filter) does not matter with sine wave clocks!

[Superdad]

3 hours ago, owldiscourse said:

So just to confirm, the filter is recommended for sine wave, and NOT square wave?

That's correct--because the filter will turn the square wave into a sine wave!

But if the square wave is crappy (say from a clock with poor sine>square converter) and your cable is just average, then using the filter on that square wave may yield better final results there as well.

[Superdad]

26 minutes ago, R1200CL said:

However the EtherRegen stops working after a few minutes in this setup. So I need to pull power cable quite often. Setting it back to internal clock is OK. Or removing the splitter I think.

Must be something about either the splitter, interaction with the SU-2, or just a slight cable problem.  When run from an external clock EtherREGEN does not tolerate even the most brief hiccup in the clock.  That is what is happening and that is why you need to repower.  

Work on your set up a bit and you will likely cure the issue.

[Superdad]

46 minutes ago, LewinskiH01 said:

Does anybody know/measured if the Ghent E07 Belden 1694A BNC 75Ω Coax Cable is a genuine 75 Ohm, good cable for BG7BTL?

 

If you follow our guidelines--especially when it comes to the BG7TBL clock--then you really don't need to worry about the impedance of that cable.

a) Do not use the square wave output of the BG7TBL--it is lousy. Use the sine wave output.

 

b) With sine wave clock you do not need to be concerned with clock/cable/device impedance matching.

 

c) Get the Mini-Circuits 11MHz low-pass filter we recommend and put that right at the BNC input of the EtherREGEN.

(https://www.minicircuits.com/WebStore/dashboard.html?model=BLP-10.7%2B or

https://www.adark.co/collections/project-giesemann-ocxo/products/mini-circuits-blp-10-7-low-pass-filter-for-giesemann-clock-group-buy-for-audiophilestyle-com)

 

Read:

 

[Superdad]

13 hours ago, richard_crl032 said:

Thanks again and as usual, I enjoy your teaching.

Yes, John is a total time-nut!  He loves this stuff and has built some fancy GPS-disciplined clocks for his house—to tell time perfectly (me, I’m happy if my watch is within +/- 5 minutes ). In fact, John is a member of a mail-list-forum of scientists and hobbyists called Time-Nuts. http://leapsecond.com/time-nuts.htm

[Superdad]

4 hours ago, Mihaylov said:

The manufacture specifies larger phase noise values for this oscillator.

We order directly from Crystek (because 24.0MHz and 25.0MHz are custom order frequencies and pricing is better direct, with 14 week lead time) and so we have had numerous conversations with both their sales engineer and their chief engineer, Ramon Cerda (who has written a textbook on quartz crystals and oscillators: https://www.amazon.com/Understanding-Crystals-Oscillators-Microwave-Library/dp/1608071189/ref=sr_1_1?dchild=1&qid=1627346729&refinements=p_27%3ARamon+M.+Cerda&s=books&sr=1-1&text=Ramon+M.+Cerda).

 

I have asked them many times why they choose to publish the above early plot of CCHD-575 phase noise instead of publishing the plot of the far better production that they had refined their process to for the 575.  (As mentioned, our three original 2015 24.0Mhz samples came with plots showing -108, -110, -112, and Crystek swears those were random samples.) They are a very conservative company and although I do not understand why, they prefer not to publish the real and better performance graphs.  But anyone with a proper phase-noise analysis station ($20K+) can--as we have--confirm that indeed production CCHD-575 perform at average of -110dBc/Hz (25MHz carrier, 10Hz offset).

 

And as a reminder: -110dBc/Hz with a 25.0MHz clock is about equivalent to -119dBc/Hz for a 10.0MHz clock (for a given oscillator circuit about 6dB per octave difference is typical; 25MHz is 2.5 octave higher so 9dBc/Hz difference to compare).

So if anyone can show me a better performing production XO available for $10 (in 250 piece quantity) I would be pleased to sample and confirm.

 

And to obtain a new (not reclaimed surplus or new-old-stock spot market) OCXO in a metal can with at least 6~9dBc better performance (about where folks begin to be able to hear further benefit) would cost us (at 250 piece quantity) somewhere between $375 to $450. And that's just where they start. If you want a 10MHz clock with -140dBc/Hz or better the cost doubles.  "Free lunch" will not be served... 

 

Of course the raw clock itself means nothing if the circuit it is put into is not very carefully designed.  Do a poor job with clock distribution (distance, trace techniques, etc.) and all the $$ spent on a nice clock are flushed away.  That's why in EtherREGEN the XO is 3mm from the jitter attenuating clock synthesizer--whose 4 outputs are all differential clock lines.  Show us a DAC with differential clocking.  That's how it should be done. 

[Superdad]

31 minutes ago, R1200CL said:

@lxgreen

But you shipped everything back already without an accept from Adrian ?

Please stay out of this Andreas.

[Superdad]

59 minutes ago, mfaoro said:

Have you seen directions or know of anyone who will modify the REF 10 to use an LPS?

Mutec REF10 already has a fully integrated linear power supply (not an SMPS as in MC-3+ USB). 

[Superdad]

21 hours ago, R1200CL said:

I’m planning to verify my clocks. I don’t have access a Symmetricom 5125A. I know the Symmetricom 5125A can be rented. Not an option for me. 
 

My question to the community is if the HP4395A is sufficient ?

https://www.digchip.com/datasheets/download_datasheet.php?id=2844459&part-number=4395A

 

I also have access to HP3563A and HP4195A, but I think the HP / Agilent 4395A is my best option.

You'll not get phase-noise plots out of those general-purpose spectrum analyzers.

 

Besides, each of your AfterDark clocks came with a individual phase-noise plot already. What more do you need?

[Superdad]

7 hours ago, Mihaylov said:

Do you need a clock with the best features for ER if the limit link still turns out to be Si5340? 

 

This is not correct.  First off, the Si5340 special sheets are just showing RMS jitter output with some XO (or crystal in your case--you cut off the top of the matrix showing with XO) of unknown performance.  So that's total jitter, including that of the unspecified clock or crystal. Have no idea why they do that as it makes the spec meaningless.  

John will chime in when he has a moment, but he tells me that the Si5430 tracks the phase-noise of the input reference extremely closely at low offsets from carrier (above 10KHz offset it gets a bit looser but for audio purpose that does not matter).

Somewhere (just not on Silicon Labs spec sheets) there are graphs showing actual performance tracking of these jitter-attenuating clock synthesizers.  They are VERY good.  Hence the use of a great external 10MHz reference (performing with at least-125dBc/Hz or better at 10Hz offset) can be heard and appreciated. 

[Superdad]

1 hour ago, LewinskiH01 said:

Upon turning on the LPS the led goes from red to orange and when it wants to turn green the power and alarm leds on the BG7BTL blink briefly and go off and the led on LPS blinks red, and this cycle continues for as long as keep the LPS on.

 

The UltraCap LPS-1.2 is simply reporting that the connect load (your clock) is exceeding the LPS-1.2’s 1.14A limit.

The BG7TBL clocks vary in their circuitry and OCXO used, so that explains why some people have been able to power theirs with an LPS1.2.

By the way, it I s during initial power-on—when the clock oven is heating up—that the most current is drawn. The load settles down quite a bit after a while.