2. digital
  3. 134532-106038

How can I tell if I need a better clock for my DAC?

I was interested in the responses to a related post by leemaze this week, saying that a Synchro Mesh was a good way to improve a DAC with subpar jitter.  I have a Cambridge CXU, with an inboard DAC; how could I determine how much jitter it has? 
128x128cheeg
audioengr
3,282 posts
 

There is jitter (lots of low level spurious signal at very specific tones).... probably inaudible but it is there.

This DAC has no reclocker, just an AK4113 receiver. It would benefit greatly from an external reclocker like the Synchro-Mesh if the S/PDIF input is used.  Are you certain that the j-test plot was using the S/PDIF input and not the USB input?

The Benchmark 3 on the other hand does a good job of rejecting jitter compared to prior models. Reclocker not necessary, however you are stuck with the sound of the master clock in the DAC. If you like that, nothing more to do.

Steve N.

Empirical Audio

melm
1,084 posts
 
In response to the OP, jitter never goes away; in the best DACs it is reduced to a minimum. The best way to keep jitter to a minimum is to see that your component has one of the high quality, high specification, clocks. If it doesn’t, sell it and buy one that has a great clock. At the moment the best clocks used in very high quality consumer units are referred to as a femtoclocks with about 80 femtoseconds of jitter.*  At least one manufacturer, Wyred4Sound, makes a femtoclock available as an upgrade to existing DACs. Reports are that the clock change alone makes for better SQ.

That being said, there seems to be little doubt that the quality of the incoming digital signal (jitter, noise) will affect the amount of correction the DAC has to do and so affects SQ. That’s why, for example, some disk players can sound better than others feeding the same DAC, and why some use reclockers in front of their DAC.

As to whether you need a better clock, that's hard to say.  Do you need better SQ?  Will your associated equipment allow a better clock's effect to be heard?  How critically do you listen?  Can you afford?  No one can answer for you.

*There are more accurate clocks but they are used in the space and defense industries and are very, very expensive.
audioengr
3,282 posts
 
The best way to keep jitter to a minimum is to see that your component has one of the high quality, high specification, clocks. If it doesn’t, sell it and buy one that has a great clock.

A good oscillator is a good start, however, there are a lot of other things that take that 80 Fsec and turn it into 200psec. These things include:

1) bad choices for logic family for the associated circuitry

2) poor board design, sliced-up ground-planes and crosstalk

3) Poor power delivery and decoupling caps choices and locations

4) slow reacting power supply and regulators

5) too much sharing of the power between oscillator and other circuits

6) no clever circuit design to minimize jitter

It turns out that these things are actually more important than having an oscillator with 80Fsec of jitter or one with 1psec of jitter.

there seems to be little doubt that the quality of the incoming digital signal (jitter, noise) will affect the amount of correction the DAC has to do and so affects SQ.

Yes, but it’s not correction, it’s simple D/A distortion.

It’s not so much the accuracy of the clock that is important, its the jitter and phase noise specs.

Steve N.

Empirical Audio

gdhal's avatar
gdhal
1,641 posts
 
This Stereophile AES J-test includes a very high level signal mixed with a LSB (smallest signal) and is a great test for interface jitter and any modulation distortion.

Ex 1. Benchmark DAC3 HGC, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 24-bit TosLink data (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

https://www.stereophile.com/images/1117BDAC3fig11.jpg

There is no jitter (spurious signal) visible above -150 dbfs noise floor on the analog output. This means there is excellent interface jitter rejection.

—-&————————

Ex 2. Schiit Yggdrasil, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 24-bit USB data (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

https://www.stereophile.com/images/217Schiitfig12.jpg

There is jitter (lots of low level spurious signal at very specific tones).... probably inaudible but it is there.

But which sounds better? :)
melm
1,084 posts
 
I suppose my response was an "all things being equal" one.  Of course if the rest of the circuit is screwed up, the best jitter clock in the world will be of little use.  So I assumed the circuit was otherwise competently designed and executed.  

A good example of the importance of the clock in an otherwise pretty well designed circuit is the optional upgrade to the femto clock by Wred4Sound.  Apparently the improvement is SQ is undeniable just by going to a better clock.

Yes, "a good oscillator (clock) is a good start".  Without a good start, you have nowhere good to go.  So what is this argument about?

I'm not here to show off what I know about bad circuits.
More to discover