There are only 2 external wordclocks that I have used and the cheaper (and probably the most popular) of the 2 is the Aardvark Aardsync II (http://www.aardvark… )

The other one that I used is an nVision unit (was the masterclock for a Neve Capricorn). I can't remember what the price was or what the model number was but they have updated versions of the unit that I had used at http://www.nvision1.com/

Maybe someone has an opinion on some of the Lucent units..........

I'm still learning about this stuff, but I also have a delta 1010, is an external word clock benefical or even possible on these things? I maybe wrongly assumed that this was a Digidesign problem?

so... can the delta 1010 benefit from a word clock?

ive head that it could esepecialy when running with more than one. though i ended up not getting one as there where other things to consider in my studio. also the delta 1010 can use an external word clock the delta 1010lt cannot.

frob wrote: ive head that it could esepecialy when running with more than one. though i ended up not getting one as there where other things to consider in my studio. also the delta 1010 can use an external word clock the delta 1010lt cannot.

I talked to someone at m-audio today on the phone and he seemed a little flaky, but checked with his supervisors and found that much like it shows in all the ads, the 1010 lt DOES have a word clock I/O and can use an external clock. Have you heard otherwise from a more reliable source? I'm thinking about getting one, but could really use the info. Thanks
tony

Very sweeping statement:

If you have only one soundcard or whatever, adding a wordclock will make things worse! Don´t get one unless you have several units that need to be synched. Some sellers want to you to believe different, but hey, it is their way of making money.

The reason is as follows. Assuming that nothing is broken, the sound card runs on an internal crystal oscillator. These are rather easy to design and run very stable. Hence they have very little jitter.

When you connect an external word clock it has to go through other circuits in the sound card. These circuits are very difficult to design and to get to work reliably. That circuit will introduce quite a bit of jitter, more than the internal oscillator had.

Of course your sound card could be broken, then replace it. Or so badly constructed that the internal crystal oscillator creates a lot of jitter, well, don´t expect the much more complicated synch circuit to be any better.

Gunnar

wow, thanks a lot gunnar. Looks like I'll be putting that money into my converter fund.

it has been posted numerus times in this forum that adding say the apogee rosetta and running off the word clock that came with the apogee. i dont like the sound of MOTU's stuff but i wount say that they are built poorly.

ghellquist wrote: If you have only one soundcard or whatever, adding a wordclock will make things worse!

Dear Gunnar,

could you undermine your statement with some actual figures for run-of-the-mill soundcard (short/mid/longtime) clock stability vs. effects of PLL noise?

I'm somewhat confused by the "clock circuit is really easy to design" statement. A clock circuit which has good short (say, a few cycles) low-mid (say, a few seconds), high-mid (say, a few hours) and long (say, a few years) time stability is not the easiest thing to design, especially not in an environment like a computer etc. A PLL on the other hand is normally limited to short-time jitter (which are bigger than those of a XO circuit).

This is not meant to say that your statement is not correct etc., my knowledge comes from entirely different technological areas (and thus frequency ranges), so I might be completely wrong...

Rainer

If a word clock will make a single converter sound worse, will it not do the same thing when you synch multiple converters?

John Stafford

I think you usually don't need an external wordclock on a single unit because it already has an internal wordclock (unless the internal wordclock is crappy). however when you introduce multiple digital units, it becomes neccesary to sync them all up...something most non dedicated wordclocks can't handle.

I use to have a Delta 1010LT (til i got my Presonus Firepod 3 months ago), and the 1010LT can have an external wordclock via SPDIF.

-Blaumph

If you have only one soundcard or whatever, adding a wordclock will make things worse!

You can't make general statements like that and be accurate. It depends on the clock circuit and it's ability to sync and/or resync to external clock sources. Many devices don't sync as well to external WC as they do to their internal sync source. A killer dedicated clock doesn't always mean better performance if the receiving device can't take advantage of the superior clock, As always, it just depends and you need to try for yourself to see if you gain any improvment rather than to just buy into marketing hype. In many cases, you'll gain far more from a better mic and mic pre in the long run.

Don't confuse a dedicated WC with WC distribution. If you have multiple devices that have WC input, they should all have a separate WC connection to the master clock.

IMO- I have a rosetta 800 paired up with my 002R and using the Rosetta as clock makes a very noticable positive improvement in the clarity when compared with the digi clock. This is most noticable with some acoustic guitar recordings. The midrange in the guitar with the 002R as clock sounds "fuzzy" when compared to the Rosetta. This is apparent whether monitoring from the 002 or the Rosetta outputs so I am led to believe that external clock can make an improvement to the stock soundcard.

A few responses to my somewhat too simplified statement above.
But first of all a very general statement again: theory and practice does not alway go together so we still need to use our ears. If it sounds better, it probably is better.

One reference to read who has both the practical experience and also has written about it is Dan Lavry. On his home page and the forum you will find in depth discussions of some of these issues.

http://

I definitely disagree with AudioGaff though:
"If you have multiple devices that have WC input, they should all have a separate WC connection to the master clock"

For most usages a properly terminated word-clock chain, using T-connectors and a terminator on the last slave will be as good as any WC distribution unit.

Gunnar

so it looks like the connections to the word clock are the same as the old tolken ring type network, how similar is the protocal for the word clock versus the tolken ring. i understand that amount of data is quite small, but that data has to besent via some kind of protocal.

Token Ring follows IEEE 802.5 and can use different physical layers, among them a 10Base2 ("thinwire") with a wave impedance of 50Ohms (vs. the word clock's 75Ohm (???)).

The connection topology for 802.5 is different insofar as the ring basically consists of a circular combination of unidirectional point-to-point connections. Thus, the signal is not the same on every part of the network (this is like the token ring works - tokens are either passed on or taken off the network and information is passed on until received and a token is passed forward again).

The word clock functions as a physical bus (i.e. a length of cable with several nodes connected which all see the same signal). Data-content-wise, there is only a clock on the network - nothing above OSI 2 really.

To get really optimum clock distribution, people suggest you use a master clock generator (which does not do "audio things" itself) and distributes the clock to all the audio devices on separate networks.
While this is correct strictly theoretically speaking, I suggest to consider a comparison with other effects. EM waves distribute with the speed of light, which is of the order 10^8 m/s. So if (as an example) your clock bus is 10m long, the delay between the devices at both ends is of the order of 10^-7s. The typical stability of a TCXO is of the order 10^-6, even increased by that of the PLL which divides the XO's frequency (but which I do not take into account as it's most probably smaller), which then reverts to a clock jitter of the order of 10^-4s.

Rainer

Rainer

Sorry, you lost me after the EM Waves part.

-Blaumph

Those with Lynx cards rarely have to upgrade anything! Clocks are fine, converters are fine. Of course if you need more converters, Lynx has you covered there now, too! Maybe a sound card upgrade is what's needed, instead of trying to make what you have work properly?

TG

Whie it is acceptable to chain a few devices with proper T-connectors and terminators, I have learned it still is ideal to not deal with that and have a dedicated connection to a WC distribution box designed for the task just as a wired cable with the correct connectors on both ends is better than dicking around with a bunch of adapters.

frob wrote: so it looks like the connections to the word clock are the same as the old tolken ring type network, how similar is the protocal for the word clock versus the tolken ring. i understand that amount of data is quite small, but that data has to besent via some kind of protocal.

Not quite. As for the protocol there is basically none. Sample is to be triggered on one of the flanks of the (more or less) square wave beeing sent on the cable. I think that the spec is unclear about if you trigger on rising or falling edge, and it does not really matter very much in most cases.

The physical connection is very simple. One master at one end of the cable. One terminator at the other end. And the cable should be a coaxial with 75Ohm as characteristic impedance. And a number of slaves tapping into the signal along the way (none of the slaves should have termination applied). Very basic RF technology, very well known from both practice and theory.

The difficult part is in the slave. Generally the AD converter needs a much higher frequency internally than the sample rate. The sampling is not triggered directly from word clock. Instead the slave has a higher frequency generator internally, and synchronizes this to the word clock. There are probably several available techniques to do this, but I personally only know PLL or phase locked loop. This is also a very well known technology from the RF arena.

The slave, in very simplified description, works by creating an internal word clock. And then comparing it to the external word clock. If the external clock edge comes before the internal clock, this is an impulse to slighly increase the speed of the internal clock. This is a way to sort of run ahead to reach the same point in the future. And in the same way, if the internal clock comes before the external clock, this is an impulse to decrease the clock frequency slightly. These small impulses, sort of nudges the internal clock to be closely in synch with the external clock. Generally, this small nudging will create a certain amount of jitter in the exact timing of when the samples are taken.

There are more and less sophisticated ways of creating a PLL, but it can become sort of difficult if you are on a tight cost budget. Remember, almost all sound cards are designed towards a target budget, and the engineer allocates costs to various parts in order to meet the budget. On a low price sound card, I would not expect the external word clock circuits to receive any larger sum in allocated cost budget. And, just probably, this might lead to less than stellar performance. Of course, without cost limits you could probably create very stable PLL circuits.

On the other hand, it takes very little money to create a short-time stable clock from a fundamental frequency crystal. And even low price crystal comes with a frequency error and long time stability in the order of a few PPM-s. Very generalized, jitter is about short time sample-to-sample stability. A slight difference in base frequency is generally not noticable, until you have two or more sound cards all with slightly different frequencys (as they are bound to have).
Gunnar.

necessary correction to my last post...

blaumph2cool wrote: Sorry, you lost me after the EM Waves part.

-Blaumph

No wonder, I completely fucked up the calculation.The error is indeed much smaller, but the system theory approach was completely wrong, so never mind.

What I basically wanted to say is that syncing the clocks should be always a good thing, because while the PLLs (which we are forced to use wheter we use an external clock or not) do jitter, the XOs have a deviation from their "supposed frequency" with a much larger time constant (some of it static or quasistatic). That means that with time, your non-synced devices will simply drift apart. If converter A's clock is only faster by one ppm (although it is safe to assume that maximum deviations are of a higher order), then it takes about half a million clock cycles to bring the two clocks completely out of sync. If you're recording at 96kHz, it takes about five seconds until this has happened.

If on the other hand your clocks are synced, they might have an error (caused by the jitter of your PLL) which "dyssyncs" your devices, but here the specified p-p jitter is the maximum deviation you will get ever, because the nature of a PLL is to correct for variations. You might have a bias caused by your clock distribution, but this stays constant (in the order of the 10-7s I mentioned earlier). And you will have bias and error caused by the tolerance of your master clock - but at least this affects all synced devices evenly.

Talking about deviations of this order might not even make sense in a lot of studios - does everyone of you use microphone cables of exactly the same lenght all the time?

Rainer

there is nothing wronge with the sound of my converters, i was more or less just interested in the tech behind the wordclock and couldnt find a suitable awneser. but this works thanks for the know.

this is a hypothetical question, as I don't currently have or have ever connected gear through digital connections:

If you take a foucusrite isa428 with the digital card option (which provides 8ch AES/EBU, adat, and s/pdif), and run an s/pdif out to an s/pdif in on the back of a pci interface with the clock on the card set to external, will that work?

What If you take the aes out to an aes in on the back of, say, motu 896hd?

Does there actually need to be a physical connection between the word clock connectors on the isa428 and the 896hd? which would you set to master?

How does this work when using s/pdif (I have a delta omni now and it has the s/pdif but no word clock connectors).

I could use a very simple explaination of connecting 2 pieces of digital gear using the various digital connection types.....

thanks guys

though ive not done this the delta seirees has the ability to run the word clock off the spdif