Cybershaft rubidium clock...low priced option?

[Elberoth]

My friend just bought the $1000 CyberShaft Premium OCXO:

 

CyberShaft OCXO Premium

 

It already has much lower jitter than the $23.000 Esoteric G-01 Rubidium Clock, not to mention the $6000 Antelope crap:

 

 

He will test it with his dCS Vivaldi stack (directly connected to the 10M clock input on the Vivaldi clock).

 

If it performes as expected, we will try the $2800 Cybershaft OCXO Limited IP17 clock:

 

[Elberoth]

Thank you. I have a couple of questions:

 

1. Is the jitter performance claimed, or measured?

2. The clock output is 10MHz. Some DAC's which allow external clock output need to be fed clocks at a base sample rate - e.g. 44.1, 48, 192kHz, etc. How do you convert the sample rate to match these DAC's?

 

1. I did jitter calculations myself. Jitter is calculated from the phase noise. It is a simple math.

 

You have to enter the Carrier Frequency in MHz, Integration Bandwidth (1 to 100Hz) and Offset & Amplitude data (Freq and phase noise figures).

 

I have used this jitter calculator:

 

Abracon Phase Noise/Jitter Calculator

 

2. The dCS Vivaldi clock has a suitable 'master clock' 10MHz input.

[Elberoth]

BTW, instead of the OCXO Limited, would the "double clock mounted" be a good step up from the Premium?

 

ƒ_ƒuƒ‹ƒNƒƒbƒN“‹Ú

 

 

Looking at the numbers, which I have seen somewhere, it would ba a major downgrade. The Cybershaft Limited uses Rakon HSO-14 clock (which is basicly Oscilloquartz BVA-8607 XO) which I believe is the lowest plase noise clock available on the market - far exeeding any Rubidium clock I've seen.

[Elberoth]

Keith_W - there is a nice diagram that shows what JohnSwenson have explained (borrowed from pinknoisemag.com):

 

 

So as you can see, there is a regular XO inside the Rubidium Oscillators plus a Rubidium cell, photo detector and lots of electronics - all of which is useless for audio.

 

If you take that XO out of the Rubidium Oscillator, it would perform much better with regard to phase noise figures. Which is why regular XO have better phase noise figures.

 

So called 'Atomic clocks' for audio is nothing more than marketing. They use the Rubidium oscillators for bragging rights only. It doesn't mean some of them aren't good and can't make your system sound better. Clocks like Esoteric G-01 ore very good and many Esoteric and dCS users reported excellent results with using those. The problem is, that for far less money, you can get even better performing OCXO !

 

The rubidium oscillator used by Esoteric - Stanford Research Systems PRS-10 - costs $1300 in OEM. That is the cost of the Rubidium Oscillator itself. The very same company - Stanford Research Systems - makes OCXO that far exceeds their PRS-10 rubidium module on phase noise performance, for 1/3 of the price.

 

If Esoteric decided to switch from the SRS Rubidium oscillator to SRS OCXO, they would instantly make their clock sound better. But they won't, as all the bragging rights would be gone, too !

 

For those interested to read more why atomic clocks are not suitable for audio, there are two very informative articles (written for non techies) that I highly recommend:

 

Grimm Audio on Atomic Clocks

The Future of Clocks: Clarifications in the Audio Clocking Paradigm

[Elberoth]

First off, many frequency standards are sine wave output, a lot of DACs that have external inputs want a square wave not a sine wave. Make SURE the reference and the DAC will work together before spending any money.

 

That is a very good remark.

 

All Esoteric DACs and CDPs accept 50 Ohm, 10MHz sine wave:

 

ESOTERIC/Dual Mono D/A Converter D-02 | ESOTERIC COMPANY

 

On the other hand, the affordable TEAC UD-503 requires 50 Ohm, 10MHz rectangle wave:

 

http://www.teac.com/content/downloads/products/935/ud-503_om_efs_va.pdf

[Elberoth]

The best engineering practice is to use the very best local XO, as close as possible to the DAC chip itself. If you have to use a separate transport, you can always send the clock signal back from the DAC to the transport.

 

The dCS and Esoteric's (who followed dCS steps) soultion is flawed. Sending clock from an extrernal box, via extra cables and PLLs, always degrades the clock signal quality. I belive dCS had chosen this solution since they started as a proaudio company, and that is how things are done in the recording studios. In a studio reality, you have to synchronise multiple digital devices (not just the CD transport), hence the need for an extrenal 'house clock'.

 

So even though the concept of an external clock is flawed, but it doesn't mean, you can not improve it in the systems that rely on it.

 

To give you a car example: I think we can all agree, that an independent suspension is a better solution for a sports car, than a live axle. But it doesn't mean, that you cannot improve the handling of the older Mustangs that have live axles.

 

Same here. The solution with an external clock is not ideal, but under certain conditions (very high quality clock), you can still improve systems that use one. And this is what dCS and Esoteric owners report hearing, after adding the external clock.

[Elberoth]

One thing that puzzles me a little about the Clock in the Cybershaft. Clocks are not cheap, and unless you buy them in the thousands, the minimum asking price by Rakon is several thousand EUR. I don't understand how the Cybershaft can sell their device at less than 2k, when the clock alone is over their MSRP.

Are they seconds or something? Would it be possible to see a photo inside showing the Rakon clock?

 

It seems that I made a mistake, stating that their OCXO Limited clock uses Rakon HSO-14 oscillator. Their site is in Japanese and I must have made a mistake in translation. The original Oscilloquartz BVA-8607 OCXO (Rakon HSO-14 predecessor) module retailed for over $10k if I'm not mistaken, so I would be surprised if HSO-14 was any cheaper. I have discovered that myself yesterday, but was waiting for a quote from Rakon on the HSO-14 OCXO module, to update my initial post (still waiting for the quote).

 

Cybershaft declined to disclose the information on what OCXO they use when I asked them yesterday, stating only that they come from Rakon.

 

The OCXO module that perfectly fits the external dimensions (which I measured on my example) and freq criteria from Rakon is their ROX5252T1 model, from their High End Telecom Discrete OCXO series:

 

Rakon - Synchronising connectivity everywhere

 

I believe they both use the same modules for both the Premium and the Limited clocks The difference beeing a better PSU in the Limited model and hand selected, best performing clocks from the lot (hence the -117dBc/Hz @ 1Hz specs for the Limited IP17 model, where Premiums only do -110dBc/Hz @ 1Hz).

[Elberoth]

Here is the phase noise plot for the Rakon OCXO that is most likely used in the CyberShaft clocks:

 

 

Those are close to state of the are numbers and we can assume some samples measure even better than this.

[Elberoth]

Here is the revised clock graph from the first page. I have also added the Sforzato PMC-01 BVA clock, which is the only clock in the world based on the Oscilloquartz BVA-8607 OCXO module:

 

 

The numbers used in this table, are based on minimum manufacturer guaranteed values. The actual samples may exceed those numbers.

 

The perofrmance of the Sforzato PMC-01 BVA clock is so good - it should be considered the curent king of the hill in clock world. No wonder there were reports people in Asia dumping the Esoteric clocks for those.

 

Antelope on the other hand, looks like a complete garbage.

[Elberoth]

Sure looks like it. But unless I am mistaken, Antelope is the only company that synthesizes 44.1, 48, etc from the 10MHz base clock? Some of us (like me!) can't use 10MHz base clocks.

 

Both the new Antelope XM and Esoteric G-01 can do that. G-01 can output a 10 MHz clock (sine wave or square wave) as well as 22.5792 and 24.576 MHz master clocks.

 

The above data is for their oscillators only.

[Elberoth]

I just got first feedback from a friend who tried the CyberShaft OCXO Premium clock on his Vivaldi stack. He has been very impressed - adding the master clock brings similar improvement to adding the Vivaldi clock itself. More resolution and articulation, with even greater smoothness.

 

So it seems this thing definately works - at least with dCS gear.

[Elberoth]

The CyberShaft OCXO Premium clock is $1000 shipped, not $3000.

 

The CyberShaft OCXO Premium clock outputs 10MHz signal and cannot be used directly on dCS components as dCS clock replacement, since all dCS grear requires 44.1/448kHz Word Clock signal.

 

The CyberShaft OCXO Premium clock is used ON TOP of the exisiting dCS clock - be it Vivaldi or Scarlatti - as their Master clock. dCS clocks have a suitable 10MHz Master Clock input at their back.

 

Hope this helps.

[Elberoth]

Here is the image of the dCS Vivaldi clock. On the left you can see the Master Clock input ('Reference IN'). This is where you connect the CyberShaft master clock.

 

 

I agree this is amazing $1000 clock can improve $20k dCS clock. On the other hand, if it was housed in a dCS like box via a regular dealer network, it could easily cost $15-20k, not $1k.

[Elberoth]

True. I've heard from respected digital audio designers that the most important is phase noise in the <1Hz range. Very difficult to get the XO specs down to 0.1Hz though.

[Elberoth]

I plan shortly to install onboard the following OCXO:

 

http://www.pulsarclock.com/ds/Pulsar_Clock.pdf

 

The RMS jitter has been calculated at 20.174fs.

 

The part is about $400.

 

I'm actually very happy with the Crystek CCHD-950 which I recently installed, and is much cheaper.

 

I'm not sure how you have calculated those jitter values.

 

Using the same calculator and range I used to calculate the jitter of the above clocks, I calculated rms phase jitter of this Pulsar XO at 446fs, not 20.174fs. 20x difference.

[Elberoth]

See post #253 here:

 

LKS Audio MH-DA003 - Page 17

 

The tool focusses on jitter in the 10Hz-10Khz range.

 

Some calculated values appear in post #274.

LKS Audio MH-DA003 - Page 19

 

As stated, I currently use the Crystek CCHD-950, replacing the stock CCHD-575 in my machine. The 950 has a worse RMS jitter spec, but sounds better in my DAC. However, I have the 575 in another of my DACs and it sounds just fine. This suggests that experimentation is the best way to find the best XO for a particular DAC, rather than raw specs.

 

Installing the following DIP adapter on the board allows for fairly easy XO rolling:

1107741 Aries Electronics | Crystals, Oscillators, Resonators | DigiKey

 

This makes comparisons of sound a lot easier; easier, certainly than having to re-solder an XO every time.

 

I can't comment on external clocks as my DAC's don't support them (and I don't have the $$$ to play that ball game anyway).

 

But those measurements are for diffrent ranges ! One was in the 1Hz-100Hz range and the other one - 10Hz-10Khz range.

 

You cannot compare XO jitter figures in different ranges.

 

Jitter is not unlike car acceleration. The measured car acceleration (measured in seconds) only gets any meaning after you apply a speed delta to it.

 

What you just did here, was comparing car A acceleration in 0-60 range, with car B acceleration in 60-120 range. Makes absolutely no sense at all !