Master Clock for your EtherREGEN

[Clockmeister]

I would suggest a fully gimbled Tx platform furthur isolated from the chassis, aloong with underboard suspension pieces and damping features on the circuit board and on all interfaces would be a basic starting point for any quality design of any sensitive electronic circuits.

With clocks its a simple but effective solution, I do have  model of a triple oven prototype sat on a solid block of copper some 15Kg in density using an electron micrco isolation platform plus many other vibration control devices that was used in University dissiation, did it result in such a large leap in total reduction of phase noise, NO but used in conjuction with other electrical and RF control mathods produced far more desirable results.

 

I do have a couple of the Cybershaft units and not to shabby they are either.

[Clockmeister]

1 hour ago, Superdad said:

Miniaturized Euler springs is what @JohnSwenson has always wanted to implement for a PCB or chassis isolation system. Their properties for ultra low frequency vibrations are outstanding. 
 

 

Hi Alex

 

In some of our non audio sensor feedback distribution circuits I designed a cruder but still effective model if you remember Thackery washers that in the good old days you used to torque up between a pair of DCOE/DHLA carbs and the fancy inlet manifoldwith a feeler gauge although I used a a very specific material for doing this, It works really well in many applications. We tried around a couple of dozen materials before settling on the one we use now. It did take a while to get there!

 

Although this was a few years ago prototype undergoing pre compliance testing those board isolation devices were trailed on the Tx's platforms we were working on back then before finialising the design.

 

IMHO a worth while excerise.

 

 

[Clockmeister]

I would agree and I use two of them in the electroncs lab to slave my analysers including a UXA signal analyser as well in the audio systems which it delivers vary natural and articuate music.

 

Would also suggest the the Cyber Shaft & Abendrot Audio are also very good

[Clockmeister]

This would be an FDN frequency dividing network.  As John has pointed out. However every time you use one of these to change / split the frequency you will incur noise and artefacts you really do not wish to have imho, some of these can be mitigated however not all.

Many of the chifi dac's use a 100Mhz master clock generator (some synthesised) which is then broken down (internally via FDN etc) to the required sample frequency for the incoming audio serial data rates.

 

The Mutec ref10-SE-120 uses a single frequency 10Mhz OXCO clock (Not Axtal) to generate its ultra-pure low phase noise reference 10Mhz signal. I recently designed an FDN/Synthetisers one for a Merging dac which uses 625Khz Master clock input frequency.

 

If you are referring to the individual sample rates such as 44.1-768Khz such as dcs et al, then usually the SE-120 or similar is used to slave the companies proprietary word clock device.

 

I did some serial data measurements not so long ago on the SE-120 see below. I know Leo Bodnar is in the process of purchasing either a R&S FSWP or Keysight UXA for ultra precise phase noise measurements. I have something a wee bit special coming for this every purpose towards December that will sort the wheat from the chaff as far as genuine clock measurements go.

 

John, I know has a phase station which is very fine tool for this also and he is well versed in PN measurements no question.

 

[Clockmeister]

On 9/24/2023 at 7:31 AM, justubes said:

Thanks Clockmeister, takes a fair bit to digest.

"you are referring to the individual sample rates such as 44.1-768Khz such as dcs et al, then usually the SE-120 or similar is used to slave the companies proprietary word clock device"

 

I was wondering as opposed to a DDS Synthesized  frequencies.

This is two totally different methodologies, DDC's relocking devices take the incoming serial data stream, extract the embedded word clock using an improved PLL array to reduce the jitter in that data transmission (AES/SPDIF/BNC/TOSLINK) which in turn improves the timing between the audio serial data and bit clock. I2s via RJ45/HDMI has a different transmission format the clock has a dedicated separate pathway. Remember at this point you have a maximum sample frequency of 192Khz. Those of you looking to obtain 1.5Mhz then look elsewhere lol

 

Now you can if you set up allows use both a DDC, slaved via a reference 10Mhz reference device, this will improve both the audio serial data stream via the DDC as well as delivering a improved world clock direct to the DDC and the streaming device as well.

 

I run this style of set up in my own personal systems.

 

On 9/24/2023 at 7:31 AM, justubes said:

if the master 10M's phase noise was 120db.

The proprietary wordclock 100db. 

 

How would one rate in order of overall importance /phase noise if:

A)10M clock at 120db  slaving a word clock of 100db phase noise

B)10M clock at 115db slaving  a word clock of 105db

 

How would these compare?

In short, which clock would be of greater significance ?

10M or proprietary word clock in your experience?

 

First you need to understand the difference between phase noise & jitter, phase noise is measured in the frequency domain, jitter in the time domain, while they are are the same essentially there are subtle differences purely depending on which application you are looking to use them with. 5/6G communications/High speed serial data movement (25Gb/s+) or audio (lol).  With test equipment particularly spectrum/phase/noise figure analysis then one of the main figures of merit is the phase noise at a specific frequency such as the following < –136 dBc (1 Hz) (f = 1 GHz, 10 kHz offset). The fundamental carrier here is 1Ghz something that audio doesn't have to worry about lol

 

My main analyser is in this region of accuracy, our main oscilloscope has a figure of merit of 60pS jitter (without the upgraded SE-120 attached)

 

These figures are a logarithmical scale not a linear one, therefore the difference between 1dBm on a plot is very different from say a linear plot from 115 to 116 and are sort of meaningless in this context.

 

For example, a speaker specification of say 40Khz +0/-3db has a true figure of just over 28Khz

 

Back to the question you posed, there is more to a systems clock performance than just the numbers, for example there are clocks out there that would normally be dismissed due to say having a lowish PN figure of merit at say 1hz however the amount of harmonic spurri on the phase noise plot is significantly less than quite a few of the so-called reference staples of audio.

 

Also as John S has pointed out some clocks are sine wave outputs as opposed to TTL square wave devices (Square waves do not exist in nature they are manufactured from the multiple various harmonic components from the main fundamental in odd multiples 1/3/5/7 etc)

 

Again, which is better? purely depends on the recieving equipment and how that has been design to accept that particular signal.

Also remember that the true figure of a clock performances needs to measure at the correct place on the DUT (device under test) at the point it is actually working in the circuit, which is very different to the point which is measured in laboratory conditions.

 

However, all things being equal and you have a well thought out & designed system then a quality clocking system can be a wonderful addition

 

Interestingly my Aurender prefers a matched audio file sample rate back from the DDC than it does a 10Mhz reference direct using identical cabling. Other systems we have used this on, and again clients prefer a 10Mhz input. Different systems different results.

 

Hope that helps

[Clockmeister]

As with everything in audio currently they will be good deals to gained all round I feel. Patience may help here.

The Ref 10 Nano, our first allocation we could have sold at least twice over!

[Clockmeister]

17 hours ago, Superdad said:

 

Hi David:

Happy to hear of your early success with the new REF10 Nano. 

We appreciate your technical contributions but if you are to continue posting in our forum I do request that you reveal the name of the retailer you are affiliated with. 
In fact, it ought to appear in the signature of your posts. 
Thanks,

—Alex C.

 

All done Alex

[Clockmeister]

You chaps may wish to see how quality phase noise measurments are made in real time Holzworth have released this informative video.

 

Phase noise measurements

 

We have one of these Holzworth, a Keysight UXA with PN application including the cross correlation method which can take a wee while lol Also a new AAronia V6 special 3Thz sweep speed and upto 2Ghz of real time bandwidth and 122Million FFT's its pretty special.

 

 

[Clockmeister]

Not sure why you don't source a dedicated superior correct primary voltage transformer?

 

The cybershaft clocks can be improved as can pretty much all of the Audio word clocks

 

There will be a new reference clock along in the not distant future by a long-established UK electronics company. No ourselves for this one, but its due very soon. All measurements using the R&S FSWP.

 

Personally I've finished my new standard for the main lab its -129dbc @ 1Hz with absolutely ultra minimal spurri on the phase noise plot without having to use smoothing techniques on the measurement instrument 

 

When I have finalised the psu configuration I will be heading to listening room!

[Clockmeister]

I would like to suggest the Nano is NOT a replacement for the Ref10 it an affordable option for those that can't strech toa ref 10. With a quality LPS it may well exceed a Ref 10. However, as I suggested a few pages before you can have a low phase noise, HOWEVER is the spuri on the phase noise plot are many not just in quantity but amplitude as well it does negate a lot of the so-called figure show off's. This directly effects the performance of the phase noise, which is after all Jitter measured in the frequency domain,

In the lab we have many clocks that only just reach -100dBc @ 1Hz yet are superior in sound quality especially spatiality, overall coherence and linearity than one of two mentioned here but more than a small margin. And are not simply off the shelf evolution clock boards with a couple of LT3045 (yawn) in parallel on some average ali express LPS.

 

Far too much BS on this subject by many that really are shall we say 'looking for audiophile pinnacle' with the wrong information that’s been fed to them by third parties with an agenda.

 

designing a clock circuit is about many different aspects of electronics, impedance pathways, circuit board stack up, connectors, supply rails, various layers of regulation/ quality of components/ and the supporting cast. before we approach the subject of high-speed serial data alignment. this is purely about Signal integrity & how to preserve it not how may DIY knock's up I can sell in a month. Leaves a nasty taste in the mouth and does the whole industry zero favours.

 

[Clockmeister]

On 11/25/2023 at 5:31 PM, Markus8 said:

Can you describe this simpler?

 

Very easy Markus

 

Like trying to reach the moon on a Vespa 125 with a flat tyre on bad luck Monday, in bad luck land,on a very bad luck day

[Clockmeister]

One couldn't possibly comment 😁