Improvement using external clock Mutec Ref10 SE120
I read that the word clock input is for using a 75ohm external clock. Is there noticeable difference if connected for discrete and premier DAC?
Reason i am asking is i have a mutec Ref10 clock. Currently connecting to hifirose RS130 and matrix SS-1 Pro network switch. 50ohm clock input.
This word clock is useful when on which audio input?
Hi.
If you already own a precision 10MHz word clock that should be connected to the word clock input on the DAC.
This will provide an external reference that is more precise than DACâs internal clock, on that model.
While some Ethernet switches also have word clock inputs, this makes â forgive my French â no F$&-ing sense because Ethernet is an asynchronous protocol.
Cheers!
A 10MHz clock is not a word clock. A word clock is either 44.1kHz or 48kHz. All of our DACs support a word-clock output to use with other transports. Some transports/servers support a word-sync input.
You cannot connect the 10MHz clock to our DACs for two reasons:
I think the only reason you see 10MHz at all is that itâs the most common clock for test gear. Itâs not useful directly for audio or EthernetâŠitâs just the most affordable way to get a good clock off the shelf.
You can read more about clocks here:
Apologies Jonathan. I meant a 10MHz âreferenceâ clock.
Agreed of course. Again, apologies for the confusion.
Respectfully, I am not sure that this statement is accurate Jonathan. An external 10MHz clock may have higher accuracy than the DACâs internal clock, and therefore can serve as a reference for the word clock. The SRS Perf10 for example, is a rubidium oscillator that quotes an accuracy of ±0.05 ppb. I am guessing that is higher accuracy than the Sentinelâs Femto clock (?) (Iâm sorry, I donât know if MSB publishes that statistic).
Separately, as I am sure you are aware, the dCS Vivaldi stack does support the addition of external reference clock to the separate Vivaldi Clock. That productâs manual states this:
âIf you have a GPS receiver or an atomic clock with a 10MHz output, you can lock the Vivaldi Clock to it. These sources are even more accurate than the Vivaldi Clock, but note that low jitter and good stability are more important than absolute accuracy.â
I greatly appreciate your thoughts on this topic!
(Separately, I stand by my response to @djq that clocking an ethernet switchâat any frequencyâmakes no sense because, again, ethernet is an asynchronous protocol.)
Why Phase noise matters more than accuracy:
When evaluating clocks for digital audio playback, it is critical to understand the difference between long-term accuracy and phase noise (commonly referred to as jitter). For audio, phase noise is vastly more important. The consistency of timing from one digital sample to the next is what dictates sound quality, not how well the clock keeps absolute time over the course of a month.
Accuracy (Long-Term Drift) Letâs look at an extreme example. If a clock is off by a fairly standard 25 ppm (25,000 ppb), it will drift by a maximum of 1 minute over an entire month. While this level of inaccuracy is a major problem for GPS systems or cell towers, it has absolutely no noticeable impact on audio playback. At worst, it creates an inaudible pitch shift of just 0.0025%.
Phase Noise / Jitter (Sample-to-Sample Timing) High long-term accuracy does not mean a clock has low phase noise. Systems like GPS or Rubidium atomic clocks are designed for long-term synchronization. They do not prioritize the microscopic, sample-to-sample consistency that high-fidelity digital audio requires.
A Wall Clock Analogy: Imagine a ticking wall clock in your house. If the second hand ticked unevenlyâsometimes taking 0.9 seconds, sometimes 1.1 secondsâyou probably wouldnât care, as long as it still displayed the exact correct time at the end of the month. That represents high accuracy, but high jitter.
In digital audio, you want the exact opposite. Those uneven âticksâ distort the sound wave. Audio requires every single tick to be exactly the same length, even if the clock loses a few seconds by the end of the month.
The Problem with Syncing to High-Accuracy Clocks Because of this distinction, trying to force an audio system to âsyncâ to a hyper-accurate external clock actually creates problems that degrade audio quality:
Adjustable oscillators are higher noise: One method to sync to an external reference, you must use an adjustable oscillator. These adjustable oscillator designs inherently have higher phase noise than a fixed-frequency low phase noise oscillators.
The Act of Syncing Creates Jitter: To stay perfectly aligned with absolute time, an adjustable clock must constantly speed up and slow down. This continuous micro-adjustment is the definition of jitter. It intentionally changes the lengths of the clock pulses, ruining your sample-to-sample timing.
Frequency synthesizers: External reference clocks rarely operate at native audio frequencies. To use them, the signal must pass through a frequency synthesizer or converter. Synthesizers/converters introduce vastly more jitter than any standard, high-quality audio oscillator.
Finally, just a friendly reminder, in engineering, âphase noiseâ is the term used instead of jitter. Phase noise is typically represented by a graph that shows the full spectrum of these timing errors, while âjitterâ is just a simplified number derived from that graph. Ultimately, keeping phase noise as low as possible is the true key to great audio.
I really enjoy your technical explanations Jonathan. 
Thank you for your responses Jonathan.
Let us assume that you are correct.
Unlike many companies, Esoteric and dCS are not known for âsnake oil.â In fact, IMHO, the best DAC manufacturers in the world are presently (not in any implied order, to avoid controversy): MSB, dCS and Esoteric.
I previously posted that dCS states an 10MHz reference clock could be beneficial to the Vivaldi clock, which is a $20,000 word clock.
Esoteric states this:
âThe G-01XD is a master clock generator equipped with Master Sound Discrete Clock, an OCXO clock module developed in-house to enhance the musical expression of digital players. The G-01XD brings out the full music playback potential of the Esoteric K-01/N-01 series, as well as various digital players that can be connected to a 10 MHz external clock.â
And this:
âThe G-01X can supply a clock signal to a maximum of eight devices. Output can be switched ON and OFF for each terminal. The unit has four 10 MHz sine wave clock output terminals. The other four output terminals (A x 2 and B x 2) enable clock signals to be output at 100 kHz, 10 MHz (square wave), and the basic frequency of 44.1 kHz/48 kHz, up to a maximum frequency of 22.5792 MHz/24.576 MHz, and separate output frequencies can be set for each terminal.
Given the statements aboveâand assuming that both dCS and Esoteric, like MSB, also have very talented engineersâwhy do you think both dCS and Esoteric offer 10MHz reference clock inputs?
(Andâin case anyone is wonderingâI do not work in the audio industry, and consider myself a mediocre engineer (at best), and am asking purely for the pursuit of knowledge.)
10Mhz clocks are mostly useless for audio conversion. If you start with 10Mhz you must translate that to a frequency that the DAC can use (All Chip DACs, all discrete DACs) If you do not want to sample rate convert to a non integer rate (which just moves the jitter to the data where it cannot ever again be removed) then you need a rate related to the original sample rate. These rates must be related to 44.1Khz or 48Khz and be at a high enough frequency that the DACs circuitry can operate. For support of all sample rates up to 768Khz the lowest frequencies that satisfy this requirement are 22.5792Khz and 24.576Khz. If you are interested only in 192Khz maximum then the frequencies of 5.6448Mhz and 6.144Mhz would suffice.
Can I draw your attention to the fact that none of these are not simply related to 10Mhz. The only way to make the proper frequencies from 10Mhz is with a fractional PLL or Frequency Synthesizer. Both of these circuits add significant phase noise just in the frequency conversion. Say you had a perfect zero phase noise clock at 10Mhz, you could never reach the performance of a simple, low cost, commodity, low phase noise crystal oscillator because you would add more phase noise in the conversion than would be created by a simple crystal oscillator at the correct frequency.
The frequency transformations that can be made with very low phase noise addition are integer division (any divisor) and integer multiplication (limited multiples, odd multiples work well). Again 10Mhz cannot be integer multiplied to 22.5792Mhz or 24.576Mhz. Likewise It cannot be integer divided to 5.6448Mhz or 6.144Mhz.
The only possible use of a 10Mhz clock would be to calibrate another oscillator if the 10Mhz clock were more accurate. In audio this is pointless however because even to cheapest of crystal oscillators, that are sub 1 dollar, are plus or minus 50ppm accuracy. This equates to a 60 minute album being played back 0.18 seconds shorter if the frequency is 50ppm too fast or 0.18 seconds longer if the frequency is 50ppm to slow. Any reasonable listener couldnât possibly tell the difference. Most oscillators of reasonable quality (a few dollars) are even better at 20ppm meaning that same hour long album will be 0.072 seconds fast or slow if it is at the ends of its specification.
In conclusion a 10Mhz clock is basically useless for audio conversion.
DCS and Esoteric put their 10 MHz master clocks in a separate chassis. Notwithstanding the fact that both 44.1 kHz and 48 kHz clocks must be synthesised from a 10 MHz clock, the massive electrical distance and junctions between the clock and the analog conversion creates high levels of jitter. Good for selling more expensive boxes, but not optimal for high fidelity digital playback.
By comparison, MSB has separate 44.1 kHz and 48 kHz clocks that are located immediately adjacent to the analog conversion (DAC modules) for the lowest possible jitter.
While this is true in general, Reference Clocks like the SRS PERF10 quoted by @rkaudio demonstrate both, incredible long-term stability and extremely low phase-noise, at better than -155dBc/Hz.
With all due respect, only if itâs not done properly.
Using an ultra-precise external Stratum-1 level Reference Clock, like the PERF10, to improve the performance of a secondary arbitrary Clock is routinely done in Astrophysics and Telecom especially where precision is paramount, using VCOs and PLLs for both Allan Deviation measure and Phase-noise budgets that are magnitudes stricter than anything in the Audio system realm.
As far as audio systems are concerned, thereâs pros and cons for different architectures. Neither is intrinsically better or worse than the other, it just comes down to implementation.
Disclosure: I do not work in the audio industry either, but I do work in astrophysics. I donât currently own an MSB, but hope to one day. Nor dCS or Esoteric.
Correct. No one on this thread is advocating for that. The discussion is about using an external reference 10MHz clock to improve the accuracy of the word clock.
dCS does not manufacture or sell a 10MHz reference clock.
Both dCS and Esoteric do provide inputs for a reference clock into the word clock, whereas MSB does not. (In MSBâs case, if it did, I believe it would be an input to the DAC module.)
That is the reason I was so interested in MSBâs position on this matter, and am thankful for their replies.
I donât own a reference clock for my DAC personally, I am not trying to sell anything, and am purely asking to improve my understanding of this interesting topic.
It seems like @Dustin_Symanski âs position is that the external reference clock could, theoretically improve accuracy, but that the difference is so small that it would not be audible. Is that correct Dustin?
I would have never considered Esoteric as being in the âBest DACâ category. Interesting.
Clock accuracy determines playback speed, even with the absolute worst accuracy modern crystals, clocks, or oscillators (± 50ppm) this difference is absolutely and completely undetectable by humans. 50ppm is 0.05% for playback speed.
If your clock is constantly changing, that condition is called phase noise. Phase noise is absolutely audible because it creates spurious tones that can be extremely unnatural. So clock accuracy is not required for quality audio reproduction ( within reason) but low phase noise absolutely is. So ideally you want a clock that is somewhat accurate, but extremely stable and unchanging. That is what low phase noise means. It doesnât matter at all if the conversion clock is 50ppm fast or slow, you just want it to have low phase noise.
If you have two identical clocks but one is frequency adjustable and the other is not, the adjustable one will always have higher phase noise. It is also really really easy to get some noise on the adjustment signal. If âanyâ noise appears on the clocks adjustment signal that will directly become phase noise.
@Jonathan_Gullman @Dustin_Symanski
You guys explained this so well. 
Your first quote from DCS makes the same assertion that the guys at MSB are making. The quote adds âbut note that low jitter and good stability are more important than absolute accuracy.â Perhaps the question is for DCS to explain why they support this feature while acknowledging there are more important considerations.
Regarding the Esoteric quotes, I agree with @matthewk . I also would not put their products in the top tier of DACs. From what I have read about their design, they are Delta Sigma DACs implemented in an FPGA instead of an ASIC. The quotes donât provide any technical information relevant to the frequency conversion and phase noise issue.
In my opinion, external clock architectures make no sense at all when high quality asynchronous interfaces are so prevalent. Completely eliminating the need for clock synchronization is a simpler solution and it allows close coupling of the clock to the converters where there is a real advantage.
Hi.
I am an MSB fan and presently own a dCS Rossini DAC. This situation is the result of me being introduced to the dCS Puccini product nearly 20 years ago, before I had even heard of MSB. I think MSB is an amazing company and am a supporter.
Re: dCS I have asked dCS and posted what they write in the Vivaldi clock manual on this topic. Their position from their discussion board over the years is that an external 10MHz reference clock could improve performance, but, as you correctly write, lower phase noise is more important.
Re: Esoteric, I wasnât advocating for or against the brand, and I have no affiliation. I was simply noting that like dCS, Esoteric also offers a 10MHz reference clock option for their new reference DAC/Streamer.
Hence, my formulation of the question to MSB, essentially, asking for their position on this interesting topic.
Thanks to @andrewchen, I understand that in precisionâs non-audio applications, like telecom and astrophysics, high accuracy, low jitter external reference clocks are regularly used to improve application accuracy.
Thanks to @Dustin_Symanski âs responses, I understand that MSBâs position is essentially: any accuracy gained from an external reference clock to MSBâs products would be imperceptible to the human ear, and could create other problems if not properly implemented, and therefore is not offered.
If I havenât summarized this correctly, please feel free to correct me.
Thank you!
I went from Rossini Apex with Rossini clock to MSB Reference DAC with Digital Director. Night and day, MSB was significantly more musical and real. I have the Cascade now and itâs in another universe.
Any benefit of greater accuracy from an external clock is totally pointless. An external clock input would decrease performance, regardless of implementation quality, therefore it is not offered. It is not that if it is worse if improperly implemented. Even the very best implementation of an external clock sync input would be flawed.
This is the perfect example of a beautifully adorned and FUD-filled bottle of snake oil offered by dCS and Esoteric. 
Go ahead and add Aurender to the list of manufacturers who doesnât understand the pointlessness of Rubidium clocks, and âreclockingâ asynchronous USB data output.
@ Jonathan or @ Dustin
I wanted to know whether I could inject the Cascade clock signal from any of the BNC clock connectors present on all input cards into a Trinnov processor which accepts a slave clock ? My Trinnov is placed just before the amplifiers, so after the Cascade in the sound flow. My take is that the Cascade clock is by far more precise than the one we can find in the Trinnov processors, be it a ST2 Pro, a MC Pro or a Magnitude.
If yes, then how to know which value is sent to the slave device ( 44.1 or 48kHz or both) by the Cascade ?
And also is there any limitation or influence in the cable length that carries the clock to the slave device ? 1,5m to 2,5m is still okay ? Thx