Internal or External Clock

I know you’re not asking me, but I agree that this seems logical — we probably wouldn’t see Vivaldi Ones paired with Vivaldi Clocks so often otherwise.


Definitely value your input Ben!

You already have your answer :grin:;


Apologies for being late to this thread, we’ve been at CanJam London this weekend so it has been a busy one!

Yes, that is correct. We use high quality crystals in all our products, and the crystals in the Vivaldi One (or any Vivaldi product) are not lesser than those in the Vivaldi Clock. However, as has been written elsewhere, the absence of any DAC, CD mech, networking etc. means that those crystals inside the Clock have less impacting them and can naturally run more accurately.

It is correct to say that for optimum operation, the DAC needs to have a high quality clock circuit inside the same unit, optimised so that clock signal reaches the DAC circuit cleanly. This is still the case with a dCS DAC even when it locks to an external Master Clock - the DAC clock then slaves to the external clock as opposed to synchronising to the timing information in the audio signal (which has its own problems).

We do this synchronisation by way of a very capable Phase Locked Loop. It is a bespoke dCS design, using our own hardware and software, that lets us get the best performance out of any source, or clock input. Every step in the clocking chain is important for audio playback, and this extends beyond just internal clock performance. The PLL inside each product in the chain needs to be of top quality, and each needs to be kept synchronised. Using a Master Clock to do this avoids effects like intersymbol interference which can induce jitter into the signal.

We’ve been using Master Clocks in audio systems, both consumer and professional, for decades. I think it is fair to say that if we had found a better way of clocking audio in that time, we definitely would have taken it! :slight_smile:


Thank you for the thorough response @James.

@James The questions remains why Vivaldi clock will be very different (or better performing) than Rossini Clock. Is it because of lesser quality crystal ?


@sourav no, we don’t compromise on crystals. The Vivaldi Clock has a more advanced chassis meaning it provides better mechanical isolation from physical vibrations which would otherwise impact the crystals.

The control board which runs the unit is more advanced as well, which (among other things) allows us to perform gentler DSP on the clock signal itself. More delicate handling there provides better performance.


First of all, many thanks for all those complete and very clear answer. It shows how Dcs experts are.

Finally, best way to know is to listen and compare.It’s what I did when I added a U Clock , than moved to Scarlatti and now with Vivaldi Clock. And I added a 10mhz clock , and this one sublimed the dcs clock.

Still want to dive deeper on this because… Well, that’s what these boards are for and if someone wants to start a 30 page thread on Ethernet cables, who am I to judge ; )

These two statements remain in conflict in my mind and I’m just trying to reconcile them technically:

James/dCS: “However, as has been written elsewhere, the absence of any DAC, CD mech, networking etc. means that those crystals inside the [Vivaldi external] Clock have less impacting them and can naturally run more accurately.”


MSB: “A clock sent over an ultra-high quality cable will still increase its jitter considerably. The best solution is to create the lowest jitter clock as close to the DAC as possible.”

It seems logical that if one has multiple devices that need to be synched, a high quality external clock would be the best way to do this (i.e., upsampler, DAC, transport), as long as the increase in jitter from the cables did not offset the accuracy gain from the isolated clock.

However, if you just have one box with no transport, e.g. MSB Premier, or MSB Select (in fact there are multiple boxes but the other box is simply the power supply, not needing synchronization, and omitting the "digital director which is a new product), it would seem to that putting a high accuracy clock right next to the DAC inside the unit would be best.

Any further thoughts on that @James (in general, not with respect to MSB in particular)?

Thank you,

(Not to pre-empt James or anything, but…) I’m curious what exactly has led you to believe dCS doesn’t have high accuracy clocks built-into each DAC? :thinking:


James if fact says there is

1 Like

Hi @barryr1 and @Anupc , I think it’s clear that dCS’s strategy is to put one (high) quality clock in the unit and one (higher) quality clock in the separate clock, per @James prior explanation.

Many participants on this forum extol the benefits of the external clock and and also
the importance of HQ clock cables. My default position is that they are correct. This was my experience with my prior system.

I’m asking technical questions about this approach.

This is an image of my former Puccini u-clock (attached below). I understand that since my former Puccini had a noisy, moving transport, a better clock could be located in an external unit. That seems obvious and logical.

If that unit didn’t have a transport, I don’t understand why this mechanism–or better, a Vivaldi quality external clock, for example–couldn’t be, or shouldn’t be, located directly inside the unit attached to the DAC, without the need for a separate chassis, addt’l milling, display, additional labor, etc.

Asked differently, if one just had an upsampling DAC–imagine a Vivaldi One with no transport–just put a great clock next to the DAC and be done. Couldn’t one avoid the additional jitter created through the cable and achieve a superior result?

I’m sorry if I’m diving too deep.

I don’t think you’re diving too deep (if anything, it’s not nearly deep enough with actual specifications etc. :wink:)

But I’m still curious what’s your basis for assuming the Vivaldi One (or DAC) doesn’t already have a great Clock built-in? What are you benchmarking it against exactly?

I think James is in essence saying here that putting the identical clock into this alternate functional environment with another functional device interferes with the purity of the clocks isolated environment and the transport of the signal from its separate isolated environment via a decent cable still beats the non isolated clock. Putting the better vivaldi clock in with the vivaldi dac still results in a clock that can be improved on by a vivaldi clock … I think :thinking:

Yes, this is correct. And that is what we do.

This would only be true if the clock signal being sent from the Master Clock was being used to actually clock the DAC. It isn’t - it is being used as a reference to synchronise the DAC’s own internal clock to. Whether any jitter on this interface is passed onto the DAC comes down to how the PLL inside the DAC is configured.

If a DAC is locking to an external Clock and has a fixed PLL filter bandwidth, and that filter bandwidth is wide, high frequency components in the clock signal (jitter) will be passed on to the DAC’s clocks and cause problems in the audio. If the filter bandwidth is narrow, the changes come and go before the PLL reacts to them, so the jitter is not passed on, but there are other problem which can arise. It is all a trade-off.

If you opt not to use an external clock whatsoever, and to simply lock your DAC to the incoming audio signal (ignoring for a moment USB and network data), you need to look at the timing information embedded in the signal. This is a difficult thing to do well, and may not even be something the DAC manufacturer has control over (if they are using off the shelf SPDIF or AES decoder chips in their product).

Furthermore, an effect called Intersymbol Interference will cause jitter in the clock purely from interactions within the data itself. This effect is not present when using a Master Clock sending a Word Clock signal, as the signal is regular.

If you did want to take a deeper dive, take a look at this Audio Precision paper:

The story is pretty much the same, just that it applies to everything inside the DAC, not just the CD mech. If you were to put that board in the same chassis as the DAC, there is no physical isolation of that clock circuit from all the circuitry inside the DAC which will still be emitting EM energy inside the product. A separate chassis completely isolates this clock circuit from anything else within the DAC.

It will also be running off of the same power supply, that is also running the DAC, the FPGA, the networking, the output stage. This will cause the voltages being sent to the clock crystals to vary, which diminishes performance.

I should note that this is essentially what we do do if you set your Bartók for example to Master mode and connect the Word Clock Out to another piece of equipment - you are then using the Bartók’s internal clock as the master for the system. It does work, and provides an improvement over using no clock connection whatsoever (due to intersymbol interference) but it still won’t be as good as a full dedicated Master Clock.

I hope that makes sense - there’s a lot of pieces in an audio system’s clock path, and they all work together in different ways. Can make it tricky to explain :slight_smile:


Synchronisation of clocks/oscillators is more audible than the absolute quality of a clock. It’s only when units (or the multiple local oscillators within a unit) are synchronized that the quality of the clocks themselves is worth addressing. Clocks are like flavoring to a dish - they all have their sound, and yes, that may be thought of as fascinating. But flavoring is what comes last creating the perfect dish - same principle.

Greetings from Switzerland, David.

Thank you for the thorough explanation @James. The above were exactly some of the details I was trying to understand and now do. Greatly appreciated!


Very late and tangential into this fascinating debate (not all of which I completely grasp with my limited digital tech awareness), I acquired an external Vivaldi Clock even though it cannot link with my built-in Meridian active speaker DACs. However I do input music via CD from my Oppo 205 and hi-res albums from an Innuos server into the Vivaldi Upsampler (replacing a previous Network Bridge) and I have noticed sharper transients and depth to the soundstage (in a good recording) although it is incremental. I am assuming my findings link with James’ explanation. I would be interested to know if this is the case and not an example of the ‘emperor’s new clothes’!


Note that a clock is not a clock, if you will.

dCS’ clocks output 44.1 KHz and 48 KHz clock signals as TTL square wave outputs.

So if you are going to sync your external device with a dCS clock or a dCS device with a third party clock it’s important to find out which clock frequencies and modes the master clock can generate and which the recipients expect.

For example, the Jay’s Audio CDT3-MK3 Transport only accepts a 10 MHz clock in, so it could not be clocked by any dCS clock.

However the VIvaldi Clock can accept the signal of a 10 MHz clock as a reference clock to which its generated clocks will be locked.

It sounds even more insane than most things in high-end audio, but I know people who have their Vivaldi or Esoteric master clocks locked to a 10 MHz clock generated by a GPS receiver.

I’m also surprised no one in the audio community is yet offering a device with a chip scale atomic clock, unless they’re less accurate than the current offerings:

CSAC SA65 Chip-Scale Atomic Clock

They’re not as expensive as you might think.

Thanks for the further guidance which is sadly way beyond my understanding. I am not at the cutting edge of (or even close to!) anything remotely to do with digital technology. Taking advice from my dealer who further referred to dCS, I am working on the basis that the dCS Clock will link and is compatible with their Vivaldi Upsampler. I am hoping there will be additional benefits incorporating this particular Clock with inputs from a disc player and music NAS outputting from the Upsampler to a stereo pair of active speakers (where individual DACs are obviously found). Those cannot be externally clocked as I understand it.

Apologies if this all sounds very simplistic but I was always more comfortable with analogue technology back in my youth! Graham