Loop Sync issues

[steins]

Hi,

I have a hard time understaning the Loop Sync feature. I have 3 192 I/O's in my setup. I have an Apogee Big Ben providing sync to my digital outboard (192's, some reverbs and misc. AD/DA's). Am i suppose to use the Loop Sync feature, or it is fine just giving them sync from the Big Ben, clocking them to External Sync? I've read the manuals and searched the net, but there seems to be different opinions on this matter.

Stein Tore

[Top Jimmy]

That is a ticklish subject, and it all depends on your opinion. There are those who opine that loop sync is of the devil and that the only proper method for clocking the interfaces is distributed wordclock. What I can tell you is this. Clocking the interfaces individually is tedious and prone with issues if you fail to jump through all of the hoops properly.

Digi(Avid) does not agree that clocking via distributed word is superior and therefore has not provided a foolproof method to do this. You will have to open the Hardware setup on every session and declare the word input as clock master for each interface you have. Pay close attention to your interfaces. Any time you see a sync light in the wrong position, you'll need to re-declare the word inputs for each of the interfaces. This can sneak up on you unawares and you may not notice that you have a clocking issue until it's too late.

On the other hand, if you use loop sync, you'll just need to set the loop master to word clock and all the other interfaces will fall into line. It's up to you which method you choose.

[mightyduck]

Originally Posted by Top Jimmy

That is a ticklish subject, and it all depends on your opinion. There are those who opine that loop sync is of the devil and that the only proper method for clocking the interfaces is distributed wordclock. What I can tell you is this. Clocking the interfaces individually is tedious and prone with issues if you fail to jump through all of the hoops properly.

Digi(Avid) does not agree that clocking via distributed word is superior and therefore has not provided a foolproof method to do this. You will have to open the Hardware setup on every session and declare the word input as clock master for each interface you have. Pay close attention to your interfaces. Any time you see a sync light in the wrong position, you'll need to re-declare the word inputs for each of the interfaces. This can sneak up on you unawares and you may not notice that you have a clocking issue until it's too late.

On the other hand, if you use loop sync, you'll just need to set the loop master to word clock and all the other interfaces will fall into line. It's up to you which method you choose.

you will also find sync options to set in the session window.

you do not have to use distributed clocking to avoid loop sync. you can simply connect one interface to another using the word clock in and out ports instead of loop sync ports, and terminate the last interface instead of looping it back into the first.

if you have three or more interfaces, you could daisy chain them, or you could use some sort of splitter to sent the clock from the first interface to the others simultaneously.

[James Drake]

loop sync is there because a pro tools rig can be made up of any number of boxes.

if you have loop sync done properly then you can think of your pro tools rig as one big thing, as opposed to a collection of boxes.

then, you can lock your pro tools rig to the rest of the studio in the way you see fit.

so, you could give word clock to one of the 192s.

give video sync to a sync i/o.

lock to one of the incoming aes streams.

whatever.

there is no reason not to run loop sync as designed by digi.

clocking is all about solving problems (like i need to sync to a beta sp deck) and nothing to do with sound quality

[mightyduck]

Originally Posted by James Drake

there is no reason not to run loop sync as designed by digi.

clocking is all about solving problems (like i need to sync to a beta sp deck) and nothing to do with sound quality

i guess you're entitled to your opinion. :)

[steins]

Thanks for replying! I think I understand this matter better now. I will try the different combinations.

Regarding termination; my current setup is distributed wordclock from BigBen to each interface. The BigBen is supposed to give a warning light if there are termination issues. AFAIK, the 192's seems to be terminated internal. The rest of my digital I/O chain are terminated properly, either by software setting or a hardware terminator at the end of the chain.

Thanks.
Stein Tore

[mightyduck]

Originally Posted by steins

Thanks for replying! I think I understand this matter better now. I will try the different combinations.

Regarding termination; my current setup is distributed wordclock from BigBen to each interface. The BigBen is supposed to give a warning light if there are termination issues. AFAIK, the 192's seems to be terminated internal. The rest of my digital I/O chain are terminated properly, either by software setting or a hardware terminator at the end of the chain.

Thanks.
Stein Tore

it wouldn't hurt, [and might help], to terminate the 192s.

but, if i read your post correctly, they are each the only device in their respective "chains". the usual procedure would be to use hardware termination on each of them if they are receiving external clock, but i think i have heard of people getting away with not doing that, especially where the 192 is the only device in the "chain".

with the amount of devices you have to clock, i would not create little sub-chains within the configuration [as in using loop sync for the 192s, or even daisy chaining them with straight word clock].

as long as you have enough outputs on the apogee device to accomodate all of your gear, just send each piece its own synchronous clock [which i think is what you are doing already], using word clock in, and not loop in, on the 192s.

[Top Jimmy]

Originally Posted by mightyduck

it wouldn't hurt, [and might help], to terminate the 192s.

The 192 word in is terminated internally.

the usual procedure would be to use hardware termination on each of them if they are receiving external clock.

Only if the clock input is unterminated internally.

with the amount of devices you have to clock, i would not create little sub-chains within the configuration [as in using loop sync for the 192s, or even daisy chaining them with straight word clock].

Daisy chaining word clock I/O is not a good idea. Each generation would be adding another layer of PLL along with its inherent error. The last device in the chain would most likely be a getting a clock with jitter an order of magnitude or greater than the first device. I once had to run a setup like that where I had to clock 4 devices each to the previous one. The signal from the last device would not pass back to the first one without noise/dropouts likely from serial clock phase error.

At least with loop sync the I/O is paralleled except at the loop master which is providing the clock.

[mightyduck]

Originally Posted by Top Jimmy

The 192 word in is terminated internally.

probably. i remember some discussion where people were annoyed at the way it had been done which stopped them from using "t" connectors.

Originally Posted by topjimmy

Daisy chaining word clock I/O is not a good idea. Each generation would be adding another layer of PLL along with its inherent error. The last device in the chain would most likely be a getting a clock with jitter an order of magnitude or greater than the first device. I once had to run a setup like that where I had to clock 4 devices each to the previous one. The signal from the last device would not pass back to the first one without noise/dropouts likely from serial clock phase error.

i disagree with that. why are you so sure that there is not a simple pass-through from word input to word output in effect when the 192 is receiving external sync? that's the usual plan.

and even if the signal is making it through each interface's pll, that is going to make each interface's output identical because each input will have been cleaned and re-shaped by the exact same pll since all the interfaces are digi 192s.

the usual rule of thumb is that up to 3 devices [maybe 4] in a daisy chain is not probematic.

Originally Posted by topjimmy

At least with loop sync the I/O is paralleled except at the loop master which is providing the clock.

that's not correct. its still a daisy chain, with buffers, and subject to degradation.

and any closed loop system has to have some slop to allow for the downsteam signal to be fed back to the first unit, after first making its way through several devices and buffers.

it seems to me that the digi loop sync plan is primarily in place to encourage the use of digidesign interfaces, especially when there is internal termination which you can't switch off [which makes it impossible to use t connectors].

[Top Jimmy]

Word clock pass-thru? There hasn't been a single manufacturer to do such a thing between the word clock input and output that I am aware of. The word out will be a PLL'd version of the input.

To believe that a PLL "cleans-up" the clock demonstrates a lack of understanding of the inherent issues surrounding the use of one. That's why all clock distribution schemes show a star distribution for terminated inputs, and parallel distribution with termination at the end for unterminated inputs. All devices get a first-generation clock. Digi's loop sync system does not have any PLLs in the loop. Essentially, this is parallel wordclock distribution.

I do not fault the Digi engineers for trying to deliver what they thought was the best "system-wide" clocking scheme for a complete Digidesign system. What I do fault them for is giving short-shrift to anything other than that point-of-view. Digi's(Avid's) downfall has partly been, and will continue to be, about lack of choice. There is a lot of money that can be made just by selling choice. People aren't so stupid as to not be able to tell when they are being sold a closed or restriction-ended system. Trying to clock a full Digi system via distributed wordclock should be a cinch, but as anyone knows who's tried to do it, it is far from it.

Join the vote (for what it's worth). http://protools.ideascale.com/a/dtd/25571-3779