When is a word clock necessary?

[Noize919]

I have a 96io (adat ins and out) hooked up optical to a Presonus DIGIMAX. I'm also running a Control 24 with its 8 pres going into the analog inputs on the 96.

how should i clock my gear?...and i may add some more digital equipment down the line.

[malditoyanki]

Any time you have two or more digital devices "talking" to each other with digital signals you need a clock source (usually the sender).

Easy answer: get a Big Ben, slave all digital devices to it and you'll probably never need anything else for clocking.

[theblue1]

That said, using the 'lead' converter's internal clock will likely provide the best quality for that converter.

Why?

Some folks seem to think that supplying a supposedly superior external clock will help a converter that has a supposedly inferior internal clocking implementation.

Unfortunately, this thinking seems often based on the false notion that the external clock replaces the internal clock's operations. But the internal clock in a conventional converter is always the direct soruce of timing for internal operations. If external clock is supplied, it actually makes the internal clock circuitry work harder at a more difficult job -- synchronizing its own internal timing to the incoming clock signal.

Because of that, the slaved clock circuitry is always trying to correct itself to the incoming clock -- and that almost invariably tends to produce more jitter.

What about those who claim that simply hooking a Big Ben or similar external clock to a single stand alone converter seems to improve the apparent quality?

People as divergent as converter design legend Dan Lavry and engineers at Digidesign have suggested -- perhaps somewhat sardonically -- that some people would appear to simply prefer the sound of extra jitter.


At any rate, malditoyanki is correct -- any time you have two converters yoked together via direct digital audio connection (say, via ADAT-lightpipe, AES-EBU or S/PDIF), the converters must be synchronized or they will drift apart due to tiny differences between the crystals that lie at the heart of all conventional converter clock circuits.

You can desginate one converter (typically the 'best') as the master or you can rely on an external standalone clock source like the Big Ben, Timepiece, etc.

And, for complex rigs, a standalone clock with multiple connection options may well be more convenient and easier to work with, even if it increases jitter in the converter you might otherwise have designated master.

[themikesonata]

Originally Posted by malditoyanki

Any time you have two or more digital devices "talking" to each other with digital signals you need a clock source (usually the sender).

Easy answer: get a Big Ben, slave all digital devices to it and you'll probably never need anything else for clocking.

Big Ben Saves Lives... it's that good.

[psycho_monkey]

Originally Posted by themikesonata

Big Ben Saves Lives... it's that good.

I'm sure it does, and it is...but it's very expensive, and the OP doesn't actually need one.

If he had the money to spend, he'd be better off replacing the 96i/o first (or getting some better pres!).

to the OP - word clock isn't needed in your setup at the moment. It's more usual when you've got multiple sources that need syncing but can't be daisychained eg multiple AD converters feeding one digital source - there's no "straight through" chain so generally either the receiving device or one of the AD converters would be the clock source, and supply word clock to the other devices, as well as all the devices supplying digital audio to the receiving device.

If that makes sense.

[Lagerfeldt]

Originally Posted by theblue1

Some folks seem to think that supplying a supposedly superior external clock will help a converter that has a supposedly inferior internal clocking implementation.

I see I'm not the only one fighting this myth

[synthoid]

Here's my take on this question. (I have a Big Ben, but I wouldn't recommend getting one unless you have a really compelling need for it.)

One situation where you need a word clock generator is that you have so many devices slaved to word clock that a daisy chain would be too long or impractical to cable together.

Another situation is where you need format conversion in the word clock itself. For example, you need to derive a word clock from an AES/EBU signal, but to send it to a device that has a BNC word clock input.

Another situation is that you need to multiply or divide the word clock frequency -- to clock converters at 96KHz, but synched with a 48KHz word clock.

Another situation is that your word clock source is unsteady, and the "unlocking" of slaved devices creates noise or some other kind of nuisance. There are some pieces of gear that have word clock outputs, but the word clock outputs either stop altogether or "glitch" when the gear changes state. I've used timecode synchronizers that have this problem, and SP/DIF connections have this problem too. This can create annoying pops and clicks at the slave device. A word clock generator solves this by always producing a steady word clock output, so that slave devices seldom or never need to lose their lock to it.

Simply repeating a word clock signal to a few slave devices is seldom a good reason to get a word clock distributor. If you have a really corrupted word clock source, you probably shouldn't be using it as a master word clock in the first place.

-synthoid

[theblue1]

Originally Posted by Lagerfeldt

I see I'm not the only one fighting this myth

Yes... I've gone many rounds on this issue, all righty. Me and Max Gutnik (marketing guy from Apogee) have hit the mats more than a couple times and I finally managed to get him to state things in a way that I could (sort of) live with:

For the record, we do not recommend buying Big Ben for the sole purpose of improving the percieved accuracy of your converter. Most often, the folks looking to do that, would benefit much more by just spending a little extra money and buying an Ensemble, or Rosetta 800. Clocking is just one aspect of what makes a converter sound good, so if you can improve the analog section, filtering and PSU as well as the clock, you are better off, even if the clock is not quite as good as the one in Big Ben (on the other hand, if you upgrade to the AD/DA-16X you get it all). But that does not mean the thousands of people who have heard the difference Big Ben makes are crazy. As we have attested to, they are not.

Max Gutnik (Apogee) Good Clock - Bad Clock?

And he is on record as conceding that slaving a given AD to an external clock will tend to increase jitter over the same unit using its internal clock as master, as well. (Although I believe he suggests that Apogee's proprietary noise-shaping technology somehow makes the increased jitter sound more accurate than the same unit in a more jitter-free standalone clocking mode; this is an aspect of his argument that I utterly fail to comprehend.)

But, as Max indicated, he still maintains that many people perceive the sound as more accurate and, he feels, that's most important.

And, of course, none of my comments should be perceived as suggesting that the Big Ben or other units are not necessarily excellent at doing the job they were designed to do: provide a master clock in complex rigs where daisy-chaining from the 'lead' converter is inconvenient or impractical.

And, as synthoid indicates, a tool like the Big Ben can be a heavensent in a complex rig where there are the kinds of issues he describes.

[elambo]

Max's quote in no way supports your theory. Nor does Max. Or the vast majority of the industry.

I'm having trouble making any sense of it myself. The clock working harder? So what - what else does it have to do? If the end result is an improvement, let the clock sweat a little. There are fans for that.

[theblue1]

Briefly -- as has been explained at length elsewhere -- the key to the problem, elambo, is that the sample timing of most conventional AD runs off an internal crystal clock. Crystals have a fixed oscillation rate and this varies in very small -- but still significant -- amounts from one crystal to another and that is why we normally must use a phase locked loop in order to try to synchronize more than one AD unit when they will be run in tandem. In a simplified sense, the PLL is continually resetting the clock circuitry of the slaved unit in order to sync to the incoming clock.

And that tends to produce increased jitter -- timing inaccuracy in the sample taking -- relative to the unit's own internal clock, which, as you will no doubt see, means decreased sample accuracy. This is a fact of multiple AD life and must be lived with when using multiple AD units yoked together.

EDIT/appended: It might help to think of clock accuracy as having two, separate, distinct, aspects: 1) regularity of spacing from one sample to another and 2) long scale accuracy, ie, accuracy against some external standard (for instance, International Atomic Time) The problem we have is that trying to synchronize a crystal based clock to an external source in order to gain greater objective accuracy (#2 above) we end up sacrificing the regularity of sample spacing (#1 above) and that manifests as dynamic inaccuracy.

Max Gutnik, as you'll see if you follow the threads below does not dispute these basic facts, although he does suggest that Apogee's noise shaping technology makes the increased jitter sound better than the actual, accurate sampling to test subjects in his company's testing -- unfortunately, those tests have apparently not been published or peer reviewed, so we'll have to take their word. (Updated info on that is welcome!)

At any rate, subjective preference is just that -- and I would be the last to try to tell someone what they prefer.

But jitter is something that can be easily measured and such measurement tells us that slaving-related jitter is a fact of life.


Below are links to more threads on the topic, elambo.

You'll be able to see everyone's thinking.

You're absolutely free to your own opinion and interpretation -- but I hope you will not promote unsubstantiated opinion as fact. I suggest that you find some credible, authoritative sources to back you up.

[BTW, I think you will see an evolution in how Max phrases things over time, incorporating more careful qualification to his statements as the arguments become more refined.]


Good Clock - Bad Clock?

Low End Gear High End Clock...

But here is the mother of all threads on the supposed benefits of external clocking as well as some very pointed and science based refutations by converter design legend Dan Lavry and others of some of the claims:

PSW Recording Forums: Dan Lavry => Proper word clock implementation

(Max and others at Apogee appear in that thread; make sure you follow their dialog with Lavry, which is pretty interesting if you're interested in this stuff.)

[Jason West]

Originally Posted by elambo

Max's quote in no way supports your theory. Nor does Max. Or the vast majority of the industry.

Is it true that it's the "vast majority"? I would've thought it's the reverse: the vast majority thinks external clocks do not improve the sound.

Not arguing with you. If it's true that most (90% ?) engineers think that BigBen/Antelope will improve their sound, then it is what it is.... but I don't remember a poll about this.

[elambo]

I'm entirely familiar with jitter and have spent a mint taming it in the studio and even my home theater. I've put in a lot of reading and listening time with different types of clocks with different points of origin (internal, external, synced from one source, daisy chained, etc.) and my opinion is that one source dictating to others is best.

You are right - it's one man's opinion - but unless things have changed dramatically since the time that I dug in deeply a couple years ago, then it's also a common opinion that a simplified clock structure received from one master is best. But if this topic is more about the clock design itself, not the syncing of multiple clocks, then I'll quietly duck out and be on my way.

For the record, in each of my studios, there is one master to which each subsequent digital device slaves and this has been the method which sounds bests. But I'll gladly reconfigure if those who I consider authorities can show that there's a better method, enough so that it would be worth the time and effort to experiment.

[elambo]

Got it. All this talk of ADA.. it's my ADD that's getting the best of me. I should read more carefully.

The question is the benefit of one device's clock from an external source, NOT the syncing of multiple devices.

I'm out

(although tomorrow I WILL test this theory)

[Geert van den Berg]

Originally Posted by Noize919

I have a 96io (adat ins and out) hooked up optical to a Presonus DIGIMAX. I'm also running a Control 24 with its 8 pres going into the analog inputs on the 96.

how should i clock my gear?...and i may add some more digital equipment down the line.

You have the ins and outs connected? Which DigiMax version do you have?

I thought it's just a preamp with digital output (on ADAT connector). So that would mean you only have one ADAT cable connected. In this case you could do without connecting the wordclock cable, but then you'd need to tell Pro Tools to sync from the ADAT input in the session setup window (command-2).

BUT, that isn't what I would do. It's more practical to let Pro Tools be the master, so I would, connect the wordclock from the 96io to the preamp and select external clock (always make sure your preamp is set to the same sample rate as your Pro Tools session!).

If you have more digital devices you'd need an external clock, if the devices in question cannot sync from their digital input.

[theblue1]

Originally Posted by elambo

Got it. All this talk of ADA.. it's my ADD that's getting the best of me. I should read more carefully.

The question is the benefit of one device's clock from an external source, NOT the syncing of multiple devices.

I'm out

(although tomorrow I WILL test this theory)

No prob!

This topic makes my brain itch, anyhow.

Have a good evening!


PS... check out that thread in Dan Lavry's old PSW forum I linked to above. And, really, if you've got something that works for you, I wouldn't be busting my ass changing it, that's for sure.