Help me understand external clock specs

[MontyW]

Hi,

 

When looking at the spec sheets of various external clocks; what figures are an indication that 'technically' one will perform better than the other when dealing with jitter?

 

For example one manufacturer states:

 

Jitter: maximum 1 pico second of intrinsic jitter, less than 10 picoseconds of accumulated jitter.

 

Another states:

 

Clock jitter < 6 ps (RMS)

 

So I guess its the lower the PS = the better performer, in real simplistic terms.

 

Monty

 

 

[davidR]

Monty, from what Gordon Rankin, Steve Nugent, and poster JR_Audio have said in the past jitter specs for clocks are very iffy because we really have no standard to follow at this point in time. Tis a complicated matter as we all know from past posts but from what I take from it the less jitter the better generally speaking but that's just one piece of the puzzle. I recognize that first jitter spec you posted from Black Lion Audio's website. I own their Microclock MK2 if you had any questions specifically about it. Matt Newport of Black Lion Audio wrote a short white paper on word clocks. They did have the article on their website but the link is dead now. I have the document saved on my drive so here's the text straight from the PDF document. I take absolutely no credit for this, just passing it along:

 

"Word Clocks: Myth and Reality

When we introduced the Micro Clock, I was rather surprised by the reaction of some

within the pro audio community. Their sentiment is that clocking externally is fraught

with potential drawbacks. Upon reading various popular arguments, I saw a much

misinformation and many half-truths being stated as fact.

 

What I suspect happened was that initially correct limitations to external clocking were

incorrectly expanded upon. It is true that external clocking does not help when applied to

two similar interfaces. But to expand that truth to other areas of pro audio is a mistake

that forecloses many opportunities to improve sound at a low cost.

 

Within a digital audio system, the digital data typically exists in a format known as I2S.

I2S consists of three elements: data, word clock, and bit-rate clock. While it may seem as

though converted analog signals disappear into never-never land where they exist in a

pure, untouchable state, this is not true. Although these three signals contain digital

information, they are actually ANALOG signals, and are subject to much of the same

rules that apply to an audio signal passing through a circuit. Like audio, these signals are

adversely affected by unwanted noise, spectral content, and other factors.

 

Word clock (we’ll call it WCK) and bit clock (let’s call it BCK) are derived from a

master clock, which often (but not always) operates at 256 times or 512 times the

sampling frequency. A master clock operating at 22.579 MHz signal divided 512 times

will yield a WCK signal of 44.1 kHz. Divide that same master clock 8 times for a 16 bit

BCK, and we get 2.8224 MHz for the BCK signal.

 

Imagine then, that you substitute a different WCK into our theoretical I2S component—

one that was generated outside of our theoretical system. Will doing this alter the sound

quality of the converted signal? Absolutely. Whether that end result is better, worse, or

undetectable depends upon a few different things—most notably the spectral content of

the externally generated clock. This is, of course, a bit simplified, but the principle

nonetheless holds true.

 

The reality is that external clocking most often improves sound. If your system has a grainy high

end or has a blurred sound due to the loss of audio image, an external clock most likely will

improve your sound. Interfaces that we know benefit from external clocking range from they

entry-level M Box 2 Pro to the more sophisticated Mark of the Unicorn (MOTU) HD192.

Additional interfaces that are enhanced by an external clock include units from Digidesign and

RME.

 

Let’s take a look as some of the all-too-common, but incorrect beliefs.

 

1) The best way to clock a converter is with its own internal clock, using a

good fundamental frequency crystal as a master clock source.

There are two ideas presented here. The first is, of course, that nothing is better than the

master internal clock when it comes to deriving a word clock signal for conversion. My

opinion is that this is dependent upon several factors:

 

How much jitter exists at the source? If we have a jittery master clock to begin with, we

will end up sending the converter a jittery word clock signal.

 

How is internal division performed? In my experience, clock division is treacherous

ground, and easily one of the largest contributors to jitter. The high speed switching and

large amounts of current movement necessary for signal division has the capability of

generating large amounts of noise and harmonic content which bleed into the clock

signal. A good master clock will be severely limited by poorly implemented division.

 

How much power supply noise exists, and how is the regulation performed? My

experience has shown that power supply noise can increase jitter at the clock source by as

much as 75%.

 

Lastly, the idea that a fundamental frequency crystal is inherently superior to other

crystals is incorrect. In fact, third overtone crystals by nature have less jitter than

fundamental frequency crystals due to a higher “Q.”

 

2) Clocking externally risks adding jitter because of how the word clock is sent

into the system

This is a great big maybe with so many undefined factors that it’s difficult to state it as an

absolute. In an inferior system, clock division and propagation are without a doubt two

huge enemies one has to face. But if one is to inject an externally generated word clock

into the system, we’ve just eliminated half of the problem—division. The word clock

will be divided down by some external and hopefully superior system. That leaves

propagation as the weak link.

 

3) A second-rate internal crystal implementation is going to outdo even a good

external clock implementation.

Like the above arguments, this ignores many crucial factors. As stated earlier, a jittery

source combined with poorly implemented division will result in a jittery word clock

within the I2S component. There’s just no getting around that. A modern multichannel

system will likely use an FPGA to handle clock management, distribution, and

phase/skew issues, and by injecting a low jitter word clock into the I2S component,

conversion will improve, period. Those of us who have heard it ourselves know that this

is an undeniable truth no matter how many misguided engineers claim otherwise.

 

There is a great misunderstanding as to what ‘jitter’ actually is. It is often thought of as

tiny variations within a clock signal’s frequency. This is only a small aspect of the issue.

Overall clock frequency stability is less crucial of an issue than clock spectral content. In

my experiments, I found that minute shifts in clock frequency, although not desirable,

have far LESS of a devastating impact on conversion quality than phase noise or

unwanted harmonics. This is due to the fact that, though the timing of the clock may be

correct, the amount of noise and or unwanted harmonics within the clock can create tiny

timing variations at the converter.

 

Conclusion

 

While addressing every issue involved in the generation, division, and propagation of

clock signals is beyond the scope of this writing, it is my sincere hope that these

unfounded arguments concerning external clocking will be put to rest.

 

This document ©2007 Black Lion Audio. No part of this document may be reproduced without

consent from Black Lion Audio."

 

 

[MontyW]

Hi David,

 

Thanks for posting that article. You are correct in recognizing the Black Lion Audio spec, the other is from a Mutec clock. I am seriously considering buying a 'clock' to deal with my firewire interface and DAC (which is my monitors).

 

Any chance you can post your findings of the BL Micro Clock as this is top of my wish list.

 

Monty

 

[davidR]

As far as the Microclock goes I won't listen to my system without it. Just to give you an idea of my setup I use a Mac Pro 8 core 2.8ghz with 6GB of RAM, Raptor 10,000 rpm hard drive (Not SSD), a Digidesign 003 Rack firewire interface for recording with Pro Tools. Monster pro power conditioning with monitors and computer plugged into it, and my interface plugged straight into the wall as I noticed an improvement in dynamics. My speakers are Event ASP8 studio monitors, the same one's they use for nearfield listening at Fantasty Studios. My room is treated with Auralex bass traps and acoustic foam. We seem to share similar setups. What type of monitors are you using out of curiosity? Originally I wanted to send my unit in to Black Lion because they specialize in heavily modding it with great feedback from trusted folks, but I picked up the Microclock instead for less money just to see if I heard any improvement since my interface isn't an "audiophile grade" DAC to begin with and would return it if need be. When I hooked up the Microclock to my interface I was pleasantly surprised at what I heard. I know describing sound is so subjective but this is what I hear when switching from the internal clock to the Microclock: a wider and deeper sound stage, more defined bass and clarity in the treble and midrange with much less fatigue, it took away "edginess" I was experiencing on brighter/harsher recordings. Instruments are better seperated in that I hear them more individually instead of mushed or jumbled altogether. The sound is "bigger" and less closed in. What I hear overall is more "rich sound" without having to listen harder to hear certain quieter, subtle details. It's just more enjoyable listening to music period. I'm not affiliated with Black Lion in any way, just a very satisfied customer. I still want to send my interface in to be modded, but I've been saving up the pennies slowly as I kind of splurged last year on all my gear at once. Let me know if you have any more questions Monty, I'll be glad to answer.

 

[MontyW]

Hi David,

 

Thanks for the info on your set up. I'm running something very simple:

 

Mac Book Pro 2009 > RME FireFace 400 > Dynaudio Air monitors. I'm soon to add a Dynaudio sub into the mix. I used to go to my monitors using a Benchmark DAC1, but soon realised it was sort of crazy to add another link of conversion, so I now go digital all the way to the monitors and this to me sounds so much better.

 

My room used to be treated with Auralex bass traps and acoustic foam, but I moved away from them and made my own, and with great results, but thats anather story :-)

 

I heard a Big Ben clocking a pair of Air monitors a while ago and I remeber the impact on sound was very much as you described whatr you hear from the Micro Clock, hence my search for a clock and 'trying' to learn a little bit more of the science behind clocking.

 

Thanks,

 

Monty

 

PS: Nice monitors!

 

[Mike in MD]

Could anyone expand on the assertion above that:

 

"Although these three signals contain digital information, they are actually ANALOG signals and are subject to much of the same rules that apply to an audio signal passing through a circuit."

 

The reference was to I2S data:

"I2S consists of three elements: data, word clock, and bit-rate clock."

 

Thanks,

Mike

 

 

[The Computer Audiophile]

That one is out of my league :~)

 

[Happy]

Lots of FUD (fear, uncertainty and doubt on this one).

 

Lots of people mix up their terms here. Confusion between synchornous and asynchronous, mixed with Analog and Digital.

 

Here they are referring to the fact that although sent digitally; some information sent is a strict function of time (clocks mostly). As such any messing with the 'time base' - latency creates problems/opportunities.

 

As one moves to the digital transmission one can move the 'data' without reference to time however when you want to recreate the original analog signal you have to add the time base back. This is why there is so much discussion on clocks, transmission and jitter. If you have an interface e.g. AES that transmits data and clock at the same time (digitally) i.e. mixing 'analog' and 'digital' information one can have problems/opportunities.

 

This is easy to solve in one direction i.e. feeding a DAC - in bidirectional signals e.g. a telephone call it is much... harder - this is why international telephone calls use fibre in 1 direction and satellite in the other - with large echo cancelers - another discussion...

 

/Paul

 

[Mike in MD]

I used to think that digital was all "cut and dried." Take three points, fit a parabola, fill in the spaces, move to the next points.

 

Apparently there is a lot of art here. Good.

 

And now I will try to fire-up my new solid state disk.

 

Mike