Mutec REF 10 Masterclock

[barrows]

On 5/18/2017 at 2:29 AM, Octagon said:

1 Hz:               ≤  -116 dBc/Hz
10 Hz:             ≤  -142 dBc/Hz

@julian.david, These are some impressive numbers.  As we know, all Xtal based oscillators vary some.  Do you do provide individual measurements of each unit produced for phase noise?  Or is there an actual guarantee that these numbers will be achieved?  And is that in a Lab setting at the ouptut of your box, or with X length of coax cable, or?

[barrows]

On 6/20/2017 at 7:06 PM, seeteeyou said:

According to the discussions from that 100-page (and counting) megathread, the best practice would be picking the clock that's closet to the DAC itself if we're interested in replacing that with sCLK-EX.

 

However, we've gotta ask the $64,000 question in order to find out if the clocks inside DirectStream DAC were holding back the ones inside a modified Singxer SU-1 with sCLK-EX.

 

In other words, is DirectStream DAC operating in master mode or slave mode for I2S? My understanding could be incorrect so hopefully someone could chime in to correct me. If DirectStream DAC were operating in slave mode for I2S, it's taking the clock(s) from the source (i.e. Singxer SU-1 with sCLK-EX) and that's the only way to justify the investment IMHO.

 

Otherwise, DirectStream DAC could be accepting the inferior clock(s) inside the DAC itself if that were operating in master mode for I2S. Consequently the performance of sCLK-EX might be held back by the inferior clock(s) behind that and maybe we aren't exactly getting what we've paid for?

 

FYI - the differences between master mode and slave mode for I2S could be found here as well

 

https://www.allo.com/sparky/kali-reclocker.html

https://www.allo.com/sparky/boss-dac.html

The DS DAC from PS Audio always operates in master mode for its converter section, as it is an asynchronous DAC by design, that is it resamples all incoming data asynchronously to its single internal masterclock.  This does not matter what input you use.

[barrows]

4 hours ago, BigAlMc said:

Does this mean that the directstream DAC is unlikely to benefit from the signal being reclocked upstream, or less likely to benefit as much as other DACs? 

I would roughly estimate about half of the DACs on the market operate asynchronously, re-clocking all inputs to their own internal clock master.  Theoretically, these DACs are "jitter immune", but in practice they still seem to perform better with a low jitter source.  I would suggest that the PS Audio DS, and other async re-clocking DACs will benefit less from the application of extreme re-clocking methods in the source, but they are still likely to benefit some.

DACs which derive their internal master clock from the source will benefit more from a low jitter source.

 

The PS Audio DS has a very good internal masterclock (Crystek VCXO).  Rather than reclocking the source, the best improvement could come from replacing the internal masterclock with something better (but internally, not via a cable etc)  This would have to be a very good clock source though, and one would need an experienced tech to make such a mod.  I have worked on some DS DACs, and would suggest money would be better spent by making internal mods, rather than adding an external clock to the source.  For the price of the Mutec stuff, you could probably find someone to add a better clock internally.

[barrows]

34 minutes ago, SwissBear said:

From a philosophical standpoint, there is a time where buying more gears must cease

Each additional component must also be powered, and each power supply adds to the overall noise profile.  Noise is cumulative, so there will come a point where additional components are doing more harm than good.

[barrows]

2 hours ago, romaz said:

 

Here is John Swenson's response to asynch USB DACs:

 

"But what about asynch USB, isn't the DAC in control? Overall yes, the DAC has its OWN FIFO and also checks it, but instead of changing a clock frequency it sends a command back to the computer which tells it to speed up or slow down the average sample rate. So even though the local DAC clock is in ultimate control of the sample rate, as far as the MC3+/USB is concerned the USB data stream is in control, it just passes it on down to the DAC."

 

 

 

RE the PS Audio DS, I was not referring to its USB input in this case, which is of course async as are all commonly used USB inputs.  I was actually referring to operation of the rest of the DAC after the USB input .  The I2S feed comes from wherever (any of the inputs) and is then oversampled into an entirely new clock domain controlled by a separate masterclock oscillator.  The masterclock in the Directstream is the only clock which has any involvement in what happens after that. 

[barrows]

2 hours ago, austinpop said:

Asynchronous USB only means the receiver controls the average sample rate. The data is still clocked by the source.

This is factually inaccurate.  Asynchronous USB does the following:

 

Data stream coming in is buffered and the buffer is controlled by software, the software sends commands upstream to the serving device only to manage the buffer such that it does not get too full or too empty.  This is done so that there are always samples available to the output.

The output of the buffer is directly clocked by a free running oscillator, and the only thing which determines the timing of the samples is the accuracy of this oscillator: output jitter has no relation to anything going on further upstream (except perhaps due to low level noise effects affecting the local clocks jitter). 

[barrows]

5 minutes ago, simonklp said:

Thank you @austinpop for your valuable information.

 

This is the answer that I want for the question that I had asked about "why music playback by asynchronous USB process is still affected by accuracy of the clocks in the chain at upstream of the DAC" in several weeks ago.

 

At that time, instead of not getting this answer, I got a feeling in some of the subsequent discussion that people was always challenging the observations in this thread.

 

After all, I would like to point out that progress in many aspects (e.g. science & etc.) is made by questioning, making hypothesis, conducting experiments, observation, drawing conclusions and etc. By reiterating this process, human beings have achieved the advancement in these many aspects nowadays.

 

Thanks again to @austinpop for your valuable information.

 

Wish you a nice weekend. Cheers.

That information is inaccurate.  This is not how async USB works.  Just because something is the answer you "wanted" does not actually change how async USB receivers work.

[barrows]

15 minutes ago, simonklp said:

@barrows Ok. Fair enough. I welcome open minded discussion of counter theory.

How async USB audio actually works is not a "theory".

[barrows]

38 minutes ago, simonklp said:

Ok, noted with thanks. Cheers.

The reason i am so adamant in my reponses is because it does a disservice to the community to suggest "theories" which are factually incorrect.

 

As I mentioned in my first response, the influence of noise coming over USB may be able to reduce clock performance in the USB receiver/DAC influencing jitter caused artifacts in the DACs analog output, hence the great lengths we go to with Sonore Rendu products to reduce that noise.  I have no doubt that USB source qualities can influence the overall outcome of system performance, but the important point here is that the clocking of Async USB is not reliant on the upstream clocks, the only clocks which matter (for this) are those which clock out the I2S feed from the USB interface buffer.

The upstream clock(s) appear to exert what I would term a second order effect which may effect DAC output quality, but the important distinction is that this effect is not directly related to how the I2S data is clocked out of the USB interface.  This is why Sonore uses a "femto" clock in the ultra and Signature Rendu products.  The only plausible explanation for this second order effect is noise coming over the USB input, causing some disruption in the DAC (ground bounce, etc).  

[barrows]

9 minutes ago, simonklp said:

So, do you mean that the plausible explanation to the effect of upstream clocks is due to their lower noise instead of their lower timing accuracy? Am I understand it correctly?

The only plausible explanation is that the result of a better clock in the upstream digital USB source (say an ultra Rendu) is that the improved timing reduces noise on the USB feed to the DAC.  I will not rule out the possibility that there might be some other advantage to a better clock in this position, but at this time it appears to be extremely unlikely, to the point of being virtually absurd.  Note the same does not apply to digital sources where one is using a SPDIF output, that is an entirely different thing, I am referring to USB output devices here.

As this is the mutec clock thread, i would also suggest that it is crazy to use such a high quality clock on (like an SMS-200 ultra) a USB output device unless first the actual DAC clock has been brought up to at least the same low level of phase noise.  Upgrading the DAC clock is going to offer an exponentially better improvement than upgrading the clock in a USB source device ( or router, switch etc).  For the price of the Mutec, one could get a competent tech/modder to put a really awesome clock(s) upgrade in their DAC for a much larger performance increase.

[barrows]

6 hours ago, austinpop said:

The reason for my statement was that there is a pervasive, yet incorrect, impression out there that in "asynchronous USB," the target supplies the clock for the data flowing from source to target. In fact, as John mentioned, the target only uses an explicit feedback mechanism, effectively a flow control mechanism. Both source and target have free running clocks.

Of course the above is how it works: BUT there is nothing wrong with that!  The clocking of the data stream only matters where the data is clocked into the DAC itself, this is the only place where any timing variances can produce an audible problem: no where else!

As long as the USB interfaces' buffer is managed such that is does not overflow, or run out of samples, it is doing its job perfectly, upstream timing is not a factor.  You guys are misinterpreting John Swenson.  

Again, I am not saying that upstream clocks will not have an influence on sound quality, but the mechanism by which they can does not have anything to do with the timing of the audio data, the only place the timing of the audio data matters is at the input to the DAC (that is the DAC chip itself, or discreet converter section, not the USB input).  I am also not suggesting that async USB is perfect (we make the Rendu products specifically to enhance the performance of even well implemented USB interfaces, and the ultraRendu and Signature Rendu SE both are governed internally by "femto" clocks).

The only plausible explanation for better clocks upstream improving sound quality is that they reduce noise getting into the USB interface/DAC, where it could effect the DAC clock, or perhaps even the analog output stage.  The mechanism of how this noise is produced is not understood (by me, anyone?), but I suspect it is related to imprecise timing resulting in more processor power being used, perhaps, perhaps because more error correction is necessary.

I would also stress that unless you have a system such as @romaz, it is also a bit daft to suggest that "huge gains" will be made with adding a $3K clock to a USB output device.  Someone with a more moderate system will be much better served by improving speakers, or DAC by that price delta.  Indeed, for $3K, one could have a couple of really good OCXO installed in their DAC, along with a good dedicated power supply for them, by a competent tech for a much bigger gain in sound quality.  Remember there are a lot of members here who do not have DACs in the $6K and up range (where DACs really start getting very good) and a thread like this is going to scare them away from computer audio.

[barrows]

40 minutes ago, ElviaCaprice said:

Sonore, you should have just kept out of this thread.  Your comments here come across as making you look unprepared/desperate to match the challenge of SOtM and Mutec. 

Our ears will guide us over any technological reasoning.  Romaz, thanks for the great thread, lot's of food for thought.  Austinpop, I look forward to your findings with the Cybershaft. 

 

You are quite comical, last time I looked this thread was started by octagon, and is about the Mutec clock generator, not Romaz...  I posted here because I was interested in the Mutec clock, I asked of Julian if Mutec individually tests their clock generators and provides a phase noise plot for each one, as there will be some variation (every crystal is a little different) and at the very low levels of phase noise specified it would be nice to now what one is actually getting..  I can buy a very good OCXO from a supplier in Europe, for about $500 USD, and they will provide a phase noise plot for each individual unit.  I also offered the suggestion that one would be better off providing a clock of this accuracy for the actual DAC (internally) where it could offer a very much larger benefit than applying it to a USB output device.  This is good technical information, which audiophiles would benefit from.  None of it has anything to do with Sonore products.

Yes, you certainly seem to not want to be bothered by any "technological reasoning", LOL.  

[barrows]

@romaz  Exactly, but i think you are missing the point I was trying to make:

 

For someone like you with a Digital front end >$20K (blu+DAVE) improvements in your D/A converter are going to be hard to come by, perhaps impossible, so going all out with experiments on source feeding that combo may be justified as your last area to achieve gains in sound quality.  But what has happened over in the other thread (and some here, as they seem to have been linked) is that readers are getting the idea they are going to get "big gains" by doing this.  I am only trying to bring back some sense of scale and relative importance to the discussion.  For example: spending thousands of dollars on re-clocking a router or switch, or a Regen type device is going to be wasted money when much bigger gains could be had to the audiophile who is using a lower level DAC.  People with more moderate DACslike an Ayre Codex or Mytek Brooklyn would eb better served by first putting that money into a better DAC first (even more so better speakers if they do not already have Vivid Audio Giyas or something similar).

Maybe my perceptions of the value of "upstream tweaks" is somewhat ameliorated by the USB source I use, but I am not going there fully as this is not a Sonore thread.  I do value some upstream tweaks in my own system, like the custom built ultra low noise power supply I built for my router, which we will be using at RMAF, but if you read this thread there are some folks here who fully do not understand even what kind of influence upstream clocks can have (noise as you and I agree, and not better data timing in the direct sense). 

 

In addition my posts are not meant to be any criticism of the Mutec components, for their intended purpose I expect they are excellent, and the clock itself, if it actually meets the claimed specs is excellent for its intended purposes (clock distribution necessary in Pro audio environments), and I agree the price appears to be quite fair as well for a non asian origin product.

[barrows]

45 minutes ago, austinpop said:

I wish more DACs enabled a reference clock input

I strongly disagree with this.  Asking customers to shell out more money to add an external clock when a really good clock should be installed internally is a money grab in my book.  Especially when one considers that an external clock will never be as good as the same clock applied internally without cable/connection losses (which are significant).

A better way for a DAC manufacturer if they would like to allow for an upgrade clock would be to implement the internal clock(s) as a module, on a small daughter board which plugs into the main DAC board.  Then they could offer an upgrade path of better clocking at much more reasonable prices and with higher performance.  (Considering that chassis and cables are a large portion of any components expense).

[barrows]

11 minutes ago, romaz said:

While it appears Sonore has not yet substantiated the impact of clocking in these "spaghetti" devices, I would encourage you guys to give it a go

This is a exactly what John S. is doing.  Hopefully his research will be able to determine the mechanism for any improvements, and then audiophiles can move forward in an informed way.  Of course we are well aware of the value of a better clock in our Ethernet Renderers, we first started experimenting with that long ago, which led to the development (among other things) of the ultraRendu and Signature Rendu SE.

But, if I am going to use clocks as accurate as the Mutec claims to be, they are going in my DAC first as a replacement for the audio clock(s) where they can do the most good.  Of course audio clock frequencies will not have as low phase at low frequencies as a 10 MHz clock can (phase noise scales with clock frequency all other things remaining equal) but using a fixed frequency clock at audio frequency will perform better than a 10 MHz reference clock used to generate (digitally) audio frequency clocks.  There are now some pretty nice audio frequency clocks available at semi-reasonable prices.

[barrows]

5 hours ago, romaz said:

Also the clock must be physically close to the active elements,with dedicated stripline PCB routing with proper termination. Running the clock externally is a crazy thing to do, as you are simply adding more jitter and noise and an extra PLL in the system

Exactly that.  Actually it appears that Mr. Watts is extolling the value of very low phase noise at very low frequencies, and internal clocking as close to the active elements as possible, and I could not agree more.

I only use the term "femto" myself because it has come to mean something to some audiophiles (hence the quotes), eventhough it does not really describe anything: is a clock with 400000 femto seconds of jitter low jitter?  Of course not.  When I talk about premium clocks for audio, I am referring to  clocks with very low phase noise at 10 Hz and below.  Although I disagree that OCXO cannot be advantageous for audio, they do address long term stability, which does not matter for audio, they are also better in some cases at achieving low phase noise at low frequencies.  I believe the Mutec is based on an ovenized oscillator, and it appears to achieve very low phase noise at low frequencies. 

[barrows]

4 hours ago, lmitche said:

The best response, and the one we all want for you, is for you to come back to us with a bigger, better more beautiful solution.  If you can measure and prove scientifically why it is better, that would be icing on the cake. We all have respect for all your team have done for us in this hobby,  Sonore, Barrows, John S and yourself.

I would suggest that, perhaps, this already exists.

 

Anyway, i have never bashed a component made by anyone.  I have suggested that some approaches, which may be used by some manufacturers are of questionable value: like using an external clock instead of using a good one internally.  Hey, look, Rob Watts, of Chord (a designer I respect very much) said the same thing...

My posting here has been towards two purposes, try to learn something myself about the Mutec products (some of which has been achieved) and try and educate some posters when they are making claims that are just plain technically wrong (like how async USB audio works).

[barrows]

3 hours ago, SwissBear said:

some of these satisfied customers very lucky enough to be able to test the Ref-10 as an improvement to the performances of the MC-3+ USB and I am one of those. Having no financial interest in Mutec, I sincerely advise MC-3+ USB users to test for themselves and appreciate the improvements

And as Jesus has pointed out, this is a sensible use for the Mutec Ref-10, as the MC-3+ is an SPDIF output device and can certainly achieve better performance with a better clock, no one questions this approach here that I am aware of.

[barrows]

5 hours ago, romaz said:

his pulse array DACs are inherently immune to jitter, he believes all sources sound the same with his DAVE and that the microRendu is no better than a Windows laptop.  Here is what Rob shared with me last year shortly after I bought my microRendu and I told him I was hearing an improvement:

So who is "right", Rob Watts or you?  It appears that you and I both know something which Rob Watts does not, that the source still matters despite using a well isolated async USB interface?  We agree, so I am really unclear as to what your point here is.

As far as Sonore goes we make products to both achieve the convenience of computer audio, and to improve the performance of the associated DAC by providing as good a USB feed as possible at the associated price point.  I doubt that anyone in this discussion thinks the USB source does not matter.

[barrows]

@julian.david Hi Julian,  for sure I in no means meant to criticize your product in any way, and a number of times in this thread I have praised it.  Sorry if my posts came of as critical of Mutec, what I was concerned about is how and what some people are using the Mutec for.  As we both know, crystal based oscillators can vary a bit in performance and I was just asking two questions: if you test each individual unit to make sure it meets spec, and whether you provide that test information to the customer.  I think doing so would add peace of mind that the customer knows for sure what they are getting, and would make sure you know what they are getting as well.  I say this specifically because the performance in the specs is very, very, good.  I am not trying to infer that I doubt it though, just a little reassurance all around would be nice.

Thanks for the kind offer of a demo as well, but at this time there is nothing in my system which could benefit from a 10 MHz reference, maybe sometime.  I tend to use audio frequency clocks directly in DACs and interfaces without the addition of a DDS.  Although it is interesting that there are relatively new developments in DDS tech which makes this approach more viable for attaining really low phase noise at low frequencies where it matters most.

 

Just saw the second post that you test each unit to be within specs, that is great news, and I am sure of value to those considering purchasing the product.  Thanks for providing confirmation of that.

 

BTW, barrows is actually my real first name (it is given from my mother's maiden name)  I use precisely for transparency on forums.  the entire name is : Barrows Worm de Geldern, and I usually go by Barrows Worm.

[barrows]

4 hours ago, zoltan said:

Rob Watts: "You can't make assumptions as to something sounds different or not. You have to do the hard work and do the listening tests."

Exactly what I did just now (again). Switched off the REF10 feeding an MC3+USB and a SOtM SM-200 ultra. Both back to their own clocks. Audibly less intimate details, 'liveness', dynamic sound, etc. The REF10 makes a difference for the better whether it should or shouldn't. 

Are you using SPDIF/AES output to the DAC?  If so I am not surprised that generating the SPDIF feed via a better clock makes a difference, as this is the audio clock.  Even with a DAC which re-clocks onboard, there will be less jitter created artifacts the lower the jitter of then coming SPDIF signal.  This is a quite different situation than if one was just using USB input to the DAC. 

[barrows]

8 hours ago, One and a half said:

I would have thought a clock needs to go through a PLL and compare this to the incoming data, whether the clock is external or internal, the function is the same.

An external clock has the plus of its PSU removed from the DAC, so this bodes well for the REF10  

A PLL is not necessary in every case.  For example if one has a DAC with local masterclock and an isolated asynchronous USB interface the clocking would look like this:

 

USB receiver-isolation-masterclock(s), re-clocking (flip-flop)-DAC.  With a synchronous clocking arrangement for the DAC (bit clock and masterclock are both produced from the same masterclcock reference) the bit clock and the masterclock are already synchronous, so no PLL is needed to line them up.  The same masterclock is used to create the data timing (it is sent back through the isolation to clock the data at the USB receiver) as is used to re-clock it before conversion.  This is the "best" way to get the lowest possible jitter at the input of the DAC, where it matters.

[barrows]

17 hours ago, afrancois said:

https://www.audiophonics.fr/en/cables-numeriques-cables-numeriques-coaxial/oyaide-db-510-coaxial-bnc-cable-75-ohm-silver-07m-unit-p-9117.html

Why would you use a 75 ohm cable for a 50 ohm impedance clock line?  Or do I have the Mutec's specs wrong?  I am pretty sure it is a 50 ohm impedance device, as are most clock inputs...  If so, you would want a 50 ohm cable.

[barrows]

Thanks for additional clarity Julian.

[barrows]

7 minutes ago, ElviaCaprice said:

Which is a negative in my opinion.  Such as the Chord DAVE.  But I live with it.

An internal power supply is better than an external supply, given that the supplies in question are of the same quality.  DC does not like to travel distances, and the added L, C, & R of sending DC power over a cable between boxes reduces performance.

This is especially true for digital or high speed circuits of any kind, a clock source running at 10 MHz is certainly a high speed circuit.  The problem is that power supply impedance increases over distance, and digital circuits perform best with the lowest possible power supply impedance.  

Less boxes is often better.