Interaction of SMPTE & SampleRate during active audio transmission

[LukeFromBerlin]

Here's one for the specialists...

I have 2 Macs synced for ...

(1) for song position: via LTC -> Mac 1 plays the sequencers (logic) and gives out SMPTE as LTC. Mac 2 is a protools HD 2 system and the active session receives the LTC via a Digidesign Sync i/o for song position and tempo info.

(2) for audio: The audio is transferred digitally via a couple of AES/EBU multicores and ADAT Lightpipes, whereas the ProTools HD system works like a digital mixing console with all channels on active input (later for recording the tracks into the session for the final mixing)

(3) for sample rate: the Sync i/o is the general clock and gives out wordclock to Mac Nr 1 as well as the HD hardware (96 i/o and 192 i/o) on Mac Nr. 2 via Avid Loopsync.

Now the question (drum roll ):

I know how crucial a good clocking is for "good" digital signals with little jitter, no clicks and pops etc.

This works fine when the Protools system is just on active input and "resting" (not in play or record mode). But what I am unsure about is what happens technically when both systems are running, which is the case when ProTools is actively recording the tracks (for final mixing etc.). In that case the protools session has to keep the sync reg. song position and speed (via SMPTE / LTC) that it receives from the logic mac playing the music , while at the same time its the master reg. sample rate (to sync the digital audio stream)...so: does or doesn't the syncing, meaning the minimal changes in speed of the to systems (protools trying to follow the logic song on mac nr. 1 as precisely as possible) during recording, possibly interact or mess up the audio "live" stream (jitter or bad quality due to little digital errors evolving from trying to keep both "speeds" in sync) ? .

In other words, when there are e.g. minimal timing issues on Mac Nr 1 (sending LTC) who is controlling the speed of the logic song (which is also full of audio files besides the virtual instruments) or little timing issues on Mac nr. 2 (receiving LTC and having to keep in sync with the incoming SMPTE data) - could this interact with the sample rate and cause losses in quality of the digital audio "livestreaming" ?

Or is SMPTE so much less precise (frames) than wordclock (samples) that you would never hear or have jitter in the audio... because the "jumps" (in the nanoworld) SMPTE is making all the time because of it's much poorer resolution... (?)

It's hard to hear / judge for me, I don't notice obvious jitter (at least I think I don't ) but sometimes I have the feeling the groove and timing is not in 100% sync... and I am insecure whether the audio could sound much better if the logic song and the protools session wouldn't have to try to stay in sync via SMPTE during the streaming (?)...

Just making ProTools the LTC master as well (not only for wordclock) is not really possible because I would have to start/stop the song on the protools mac everytime then... and wait for around 5 second for logic to sync up every single time then ...

So: any expertise on the interaction of sample rate and timecode during active digital audio transmission is greatly appreciated !!

Thanx !!!

[James Drake]

Hmm,

You seem a little confused.

One thing to note is that it is not possible to get tight sync with logic.

Originally Posted by LukeFromBerlin

Here's one for the specialists...

I have 2 Macs synced for ...

(1) for song position: via LTC -> Mac 1 plays the sequencers (logic) and gives out SMPTE as LTC. Mac 2 is a protools HD 2 system and the active session receives the LTC via a Digidesign Sync i/o for song position and tempo info.

What hardware do you have on the logic rig?

Remember that LTC can only give positional info (in hours/mins/secs/frames), no tempo or other musical info.

(2) for audio: The audio is transferred digitally via a couple of AES/EBU multicores and ADAT Lightpipes, whereas the ProTools HD system works like a digital mixing console with all channels on active input (later for recording the tracks into the session for the final mixing)

I would avoid ADAT if possible, I have never been able to use it without having digital 'ticks' every so often. Use AES or some other format whenever possible.

(3) for sample rate: the Sync i/o is the general clock and gives out wordclock to Mac Nr 1 as well as the HD hardware (96 i/o and 192 i/o) on Mac Nr. 2 via Avid Loopsync.

This is fine. What hardware do you have on the logic rig?

Now the question (drum roll ):

I know how crucial a good clocking is for "good" digital signals with little jitter, no clicks and pops etc.

There should be one clock master and everyone else should lock to this. But even still, with a correct clock setup and no flashing lights I have never been able to use ADAT and fully trust it.

This works fine when the Protools system is just on active input and "resting" (not in play or record mode).

There is no difference between 'resting' or playing. The dsp is dsp is a mixer. It doesn't have different 'modes' of working if the transport is running or not!

If there is audio coming to/from HDD then this is just more audio coming to/from the mixer.

But what I am unsure about is what happens technically when both systems are running, which is the case when ProTools is actively recording the tracks (for final mixing etc.). In that case the protools session has to keep the sync reg. song position and speed (via SMPTE / LTC) that it receives from the logic mac playing the music , while at the same time its the master reg. sample rate (to sync the digital audio stream)...so:

The speed of the system is set by proper clocking, so in this case your sync I/O being clock master.

Speed is not set by the LTC. Timecode just tells you what current place you are NOW, and it just happens that it is constantly updating.

In your case the LTC should be coming from logic which is locked to word clock, so the LTC should be in time with the word clock. But you can't trust logic sync so...

does or doesn't the syncing, meaning the minimal changes in speed of the to systems (protools trying to follow the logic song on mac nr. 1 as precisely as possible) during recording, possibly interact or mess up the audio "live" stream (jitter or bad quality due to little digital errors evolving from trying to keep both "speeds" in sync) ? .

In a properly clocked system, there is no 'minimal changes in speed'. Everyone should be running together at the same rate.

In other words, when there are e.g. minimal timing issues on Mac Nr 1 (sending LTC) who is controlling the speed of the logic song (which is also full of audio files besides the virtual instruments) or little timing issues on Mac nr. 2 (receiving LTC and having to keep in sync with the incoming SMPTE data) - could this interact with the sample rate and cause losses in quality of the digital audio "livestreaming" ?

Or is SMPTE so much less precise (frames) than wordclock (samples) that you would never hear or have jitter in the audio... because the "jumps" (in the nanoworld) SMPTE is making all the time because of it's much poorer resolution... (?)

It's hard to hear / judge for me, I don't notice obvious jitter (at least I think I don't ) but sometimes I have the feeling the groove and timing is not in 100% sync... and I am insecure whether the audio could sound much better if the logic song and the protools session wouldn't have to try to stay in sync via SMPTE during the streaming (?)...

Just making ProTools the LTC master as well (not only for wordclock) is not really possible because I would have to start/stop the song on the protools mac everytime then... and wait for around 5 second for logic to sync up every single time then ...

So: any expertise on the interaction of sample rate and timecode during active digital audio transmission is greatly appreciated !!

Thanx !!!

I'm not sure what the problem is.

If you hear no problem there is no problem.

Bear in mind there is no way to get proper sync with logic either as master or slave so....

[LukeFromBerlin]

Dear James

thanks a lot for trying to follow my descriptions & the comments !!!

Originally Posted by James Drake

What hardware do you have on the logic rig?

MacPro 8-core with 2 PCIe cards, one RME for AES/EBU and one for ADAT Lightpipe

Originally Posted by James Drake

Remember that LTC can only give positional info (in hours/mins/secs/frames), no tempo or other musical info.

But due to the connection of time and positional info it -indirectly- is also the info reg. the tempo of the song I guess... as protools doesn't "know" about tempo, it's just a DSP recording audio. But -and that's the crucial point- doesn't ProTools need to know where to "place" the recorded Audio while recording (the PT session "moves ahead" when in play or record mode)... or does it just record an audio file, regardless of the LTC and "places" it into it's timeline afterwards ? That would mean PT shows kind of a "visual demo" of where theregion would end up afterwards but does not really "place" it while still recording it ?
Thats' what I am trying to figure out. When PT records not only onto HD but also "into the song/session" it would HAVE to react to even slightly out of sync LTC while recording the audio file(s) in that moment... because it is constantly updating it's own song position where it is recording to...

Originally Posted by James Drake

I would avoid ADAT if possible, I have never been able to use it without having digital 'ticks' every so often. Use AES or some other format whenever possible.

A agress, especially because of the sample rate limitations of ADAT (48 KHZ max) because ProTools HD STILL does not support SMUX (any changes there ? I would open a bottle and party immediately !!!)
That I use ADAT when I have more than 16 Ch of Audio has to do with the much higher costs for AES/EBU. I use 2 ADAT bridges I got from ebay which is a very good bang for the buck if you want to stay HD and not switch to native... and get 32 more digital channels...
And so far it seems to work fine, I trust more in AES/EBU but so far I have not hear a single tick, click or pop using ADAT lightpipes (knock on wood)

Originally Posted by James Drake

What hardware do you have on the logic rig?

I think you mean the ProTools ?
96 i/o with an ADAT bridge on the legacy port
192 digital i/o with a second ADAT bridge on the legacy port

192 for AES/EBU only
96 i/o for analog

Sync i/o as a masterclock, no daisy chaining, all T-adapters. All Avid hardware synced with loopsync, the adat bridges using the old digi-clock format

Originally Posted by James Drake

There is no difference between 'resting' or playing. The dsp is dsp is a mixer. It doesn't have different 'modes' of working if the transport is running or not!

That's once of the crucial questions I guess (see above). When it's recording and the rcording should sound excatly the same like when resting it would have to record the audio file / region directly into the timeline... which would lead to a direct connection of sync and audio. Otherwise it would have to "place it afterwards"... which could sound different in case there were very small (on sample level) inaccuracies in the sync during the recording process... and what I heard while recording.

Originally Posted by James Drake

The speed of the system is set by proper clocking, so in this case your sync I/O being clock master.

Speed is not set by the LTC. Timecode just tells you what current place you are NOW, and it just happens that it is constantly updating.

(see above question reg. how PT handles external syncing.. which is indeed just constantly updatet position info... but what if the position "steps" are not always 100,0000 % exact ... which is the case when there are small timing inaccuracies...

Originally Posted by James Drake

In your case the LTC should be coming from logic which is locked to word clock, so the LTC should be in time with the word clock. But you can't trust logic sync so...

How do you mean that ?

And doesn't logic sync itself reg. timing/tempo/song position even if the sample rate is being clocked externally (to make sure the audio arrives phase-correct etc.) ? Here I am really kinda

Originally Posted by James Drake

In a properly clocked system, there is no 'minimal changes in speed'. Everyone should be running together at the same rate.

(see above. Can I or do I have to tell logic to sync to Wordclock ? The audio cards are of course clocked (by the sync i/o), so the audio they send into the lightpipes and AES-cables is clocked. But doesn't logic "time itself" while playing a song (regarding internal Midi and WHEN it releases the audio samples (which themselves are clocked by wordclock) ?

Originally Posted by James Drake

Bear in mind there is no way to get proper sync with logic either as master or slave so....

You mean compared to protools as a recording sequencer ?

I produce mainly electronic music with heaps of plug ins that are not availabe for PT, so Logic is my "instrument" and ProTools my "mixer" due to the FAR FAR better mixer (when summing up) and the great TDM Plugins... but it's not a matter of recording acoustic audio with logic (I use PT of course for adding vocals and natural instruments...)

And since Logic is not "following" SMPTE any more (which was a disaster) but is the master reg. playing the song (not regarding wordclock) it seems to work so far - timingwise. Although there ARE timing issues sometimes which add to my insecureness reg. the LTC/Clock subject.... and I began to think about where these inaccuracies could come from and sometimes wonder if it could sound different or better reg. the transmission of the audio. So that's why I ask all these questions... I think I should really try to know and understand what is happening inside my setup reg. clock AND sync... and how they interact (or perhaps not ) ...

Hope I sound a LITTLE less confused or confusing, and appreciate your help and advice !!

cheers

[James Drake]

Originally Posted by LukeFromBerlin

Dear James

thanks a lot for trying to follow my descriptions & the comments !!!

no problem, but i think you would do well to do some proper research on how sync systems work and how they should be used.

and THEN understand that logic will not offer tight sync even if you do everything by the book.

MacPro 8-core with 2 PCIe cards, one RME for AES/EBU and one for ADAT Lightpipe

so, you have two interfaces on the logic rig? you need to make sure that these are locked properly to the sync i/o without any mistakes with word clock termination.

also, are you using the RME box to output LTC? in which case logic will be generating MTC in software and sending it to the RME box, which will convert to LTC. the RME box SHOULD endeavour lock the LTC output to the wordclock (don't know if it does), but i don't know how it would deal with the MTC from logic which is probably all over the place.

But due to the connection of time and positional info it -indirectly- is also the info reg. the tempo of the song I guess... as protools doesn't "know" about tempo, it's just a DSP recording audio.

Yes, PT DOES know about the tempo.

You have to have the same tempo map on the logic session and the PT session.

They both now have the same relationship between bars/beats and min/secs and so if the timecode says 'now at 03:22', both computers end up at the same bar/beat.

But -and that's the crucial point- doesn't ProTools need to know where to "place" the recorded Audio while recording (the PT session "moves ahead" when in play or record mode)... or does it just record an audio file, regardless of the LTC and "places" it into it's timeline afterwards ? That would mean PT shows kind of a "visual demo" of where theregion would end up afterwards but does not really "place" it while still recording it ?
Thats' what I am trying to figure out. When PT records not only onto HD but also "into the song/session" it would HAVE to react to even slightly out of sync LTC while recording the audio file(s) in that moment... because it is constantly updating it's own song position where it is recording to...

pro tools will place recorded audio on the timeline correctly in relation to incoming LTC when recording in sync.

when generating timecode, pro tools will output LTC and audio at the same time with the correct relationship as shown on the timeline.

A agress, especially because of the sample rate limitations of ADAT (48 KHZ max) because ProTools HD STILL does not support SMUX (any changes there ? I would open a bottle and party immediately !!!)
That I use ADAT when I have more than 16 Ch of Audio has to do with the much higher costs for AES/EBU. I use 2 ADAT bridges I got from ebay which is a very good bang for the buck if you want to stay HD and not switch to native... and get 32 more digital channels...
And so far it seems to work fine, I trust more in AES/EBU but so far I have not hear a single tick, click or pop using ADAT lightpipes (knock on wood)


I think you mean the ProTools ?
96 i/o with an ADAT bridge on the legacy port
192 digital i/o with a second ADAT bridge on the legacy port

192 for AES/EBU only
96 i/o for analog

Sync i/o as a masterclock, no daisy chaining, all T-adapters. All Avid hardware synced with loopsync, the adat bridges using the old digi-clock format

be careful with t-adaptors and wordclock and terminating correctly.

That's once of the crucial questions I guess (see above). When it's recording and the rcording should sound excatly the same like when resting it would have to record the audio file / region directly into the timeline... which would lead to a direct connection of sync and audio. Otherwise it would have to "place it afterwards"... which could sound different in case there were very small (on sample level) inaccuracies in the sync during the recording process... and what I heard while recording.




(see above question reg. how PT handles external syncing.. which is indeed just constantly updatet position info... but what if the position "steps" are not always 100,0000 % exact ... which is the case when there are small timing inaccuracies...



How do you mean that ?

I don't know what you mean about these timing inaccuracies. If you are generating timecode correctly then the frame edges will be locked to the frame edges of your system.

And doesn't logic sync itself reg. timing/tempo/song position even if the sample rate is being clocked externally (to make sure the audio arrives phase-correct etc.) ? Here I am really kinda





(see above. Can I or do I have to tell logic to sync to Wordclock ? The audio cards are of course clocked (by the sync i/o), so the audio they send into the lightpipes and AES-cables is clocked. But doesn't logic "time itself" while playing a song (regarding internal Midi and WHEN it releases the audio samples (which themselves are clocked by wordclock) ?

logic has no sync.

when generating timecode, it will output timecode which is sort of related to the point on the timeline, but not exactly.

it will be different every time you hit play. and be different depending on how many plugins and what hardware you use etc.

there are even cases where logic can 'lock' to MTC but seem to varispeed the audio in and out of sync. in which case it just must ignore the relationship between the sample grid on the timeline and the sample edges of the audio interface and just be sample rate converting everything all the time or something.

You mean compared to protools as a recording sequencer ?

I produce mainly electronic music with heaps of plug ins that are not availabe for PT, so Logic is my "instrument" and ProTools my "mixer" due to the FAR FAR better mixer (when summing up) and the great TDM Plugins... but it's not a matter of recording acoustic audio with logic (I use PT of course for adding vocals and natural instruments...)

And since Logic is not "following" SMPTE any more (which was a disaster) but is the master reg. playing the song (not regarding wordclock) it seems to work so far - timingwise. Although there ARE timing issues sometimes which add to my insecureness reg. the LTC/Clock subject.... and I began to think about where these inaccuracies could come from and sometimes wonder if it could sound different or better reg. the transmission of the audio. So that's why I ask all these questions... I think I should really try to know and understand what is happening inside my setup reg. clock AND sync... and how they interact (or perhaps not ) ...

Hope I sound a LITTLE less confused or confusing, and appreciate your help and advice !!

cheers

in your case, if you are worried about tight sync, i would:

use logic as a soft-synth host (or try mainstage).

it just sits there doing nothing, just being a synth.

do all your audio record and midi programming in pro tools.

in this way logic will just play noises when you send it midi notes, and your whole project will be in sync, because you are not trying to sync pro tools and logic.

[LukeFromBerlin]

Originally Posted by James Drake

so, you have two interfaces on the logic rig? you need to make sure that these are locked properly to the sync i/o without any mistakes with word clock termination.

RME wordclocks it's cards internally (with a mini-cable inside) and they seem to be very good at it. I had a RME-guy on the phone once who explained to me how it works. Pci-e Card #1 is connected to sync i/o via a -proper- cable with the right terminators. Pcie-e Card #2 is connected internally (wouldn't have jacks for this anyways).

All this works perfectly well, rme software always shows (the right) samples rates and sync etc.

Originally Posted by James Drake

also, are you using the RME box to output LTC? in which case logic will be generating MTC in software and sending it to the RME box, which will convert to LTC.

Actually logic is not capebable of giving out SMPTE other than via MTC (miditimecode). As Protools (at least the sync i/o) is not capebable of receiving MTC I had to find a workaround... which works fine so far and which I can recommend to everyone with the same or similar problems: I recorded TC in various sample rates and place the right one on a logic audio track within the song. The analog out goes into the LTCV in of the sync i/o. Bingo.
But I hate the hassle of setting the track up etc. so I want to get a Rosendahl MTC/LTC converter soon. I've only heard good things about it so far.

Originally Posted by James Drake

You have to have the same tempo map on the logic session and the PT session.

Sure thing.

Originally Posted by James Drake

pro tools will place recorded audio on the timeline correctly in relation to incoming LTC when recording in sync.

EXACTLY. That's what I am wondering about. The only way ProTools can do this while recording a long piece of audio (e.g. a 5 minute stem-track) while syncing to an external LTC would be to change it's sample rate OR the song tempo according to possible inaccuracies (frames / subframes ?) in the LTC it receives. Like when the logic songs speeds down just a tick when multiple audiotracks jump on , each with 3-8 plugins on them etc..
My bigger concern is that this could lead to bad digital audio (jitter, quality loss, phase problems) rather than timing issues, because I guess ProTools will always try to keep it's session's timing as accurate as possible by any means... but the logic song might fade in and out of it's 100% sync at the same time...

And -as far as I know- logic will always be it's own sync-master, regardless of the audio wordclock connections. Unless it acts as a slave and that is indeed a nightmare...

Originally Posted by James Drake

I don't know what you mean about these timing inaccuracies. If you are generating timecode correctly then the frame edges will be locked to the frame edges of your system.

If Protools wants to keep its timing accurately in sync while a whole 5-minute song is running (and ProTools is recording) it will have to do SOMETHING do compensate for a possible discrepance between LTC and Wordclock timing it encounters on the way. That's what I also wonder about. Either it will have to vary the sample rate (like logic when acting as a slave) or it will have to place the samples differently on the timeline (BUT they are all part of ONE audiofile...)

Originally Posted by James Drake

logic has no sync.
when generating timecode, it will output timecode which is sort of related to the point on the timeline, but not exactly.

Well, even after I bought a Unitor-8 interface to achieve this I had to find out that the "SMPTE OUT" jack there is for STRIPING only. So far it seems like logic can ONLY generate MTC. Which is OK, but "hidden" in all the other midi info and hard to "extract" or change into LTC. That's why I use the workaround with recorded TC (from the unitor ;-) and am saving money for the Rosendahl

Originally Posted by James Drake

there are even cases where logic can 'lock' to MTC but seem to varispeed the audio in and out of sync. in which case it just must ignore the relationship between the sample grid on the timeline and the sample edges of the audio interface and just be sample rate converting everything all the time or something.

...or something, yes. I dont know what the hell its doing but same experience here... the Samplerate would vary like by 10-20% or so
That's the "nightmare" situation I described earlier, no option.
And for my workflow logic needs to be master (reg. sync, not Wordclock) anyways.

Originally Posted by James Drake

in your case, if you are worried about tight sync, i would:

use logic as a soft-synth host (or try mainstage).

it just sits there doing nothing, just being a synth.

do all your audio record and midi programming in pro tools.

in this way logic will just play noises when you send it midi notes, and your whole project will be in sync, because you are not trying to sync pro tools and logic.

Thats a very clever solution James (seriously). I will consider this, only so far I am much happier with doing my very complex midi work in logic... one of the few things where decades of experience make logic superior to new-to-midi-protools (which I prefer for almost everything else, last but not least audio editing etc.). Also it can get a hell of complex to categorize all this midi data as I do a LOT of automating of parameters of Instruments, Plug in's etc. That would quickly be 50-90 midi channels that need to be set up... and can cause timing problems too (like in my good old analog days with more midi-than anything else cables lying around all over the place... an cascading midi-interfaces etc... also a risky business...

thanks again for your expertise

[FajitaTone]

Are you SURE the Sync can't accept MTC?

[LukeFromBerlin]

Yes... although it would make me VERY happy if someone could proove me to be wrong

There is only a MTC OUT jack / connection on the back ...and that is the only Midi connection on the whole interface too.

Seems like this has also not changed when Avid brought out the "Sync HD I/O" (I am using and referring to the old "Sync i/o", which seems to be more or less the same or just as fine)

[FajitaTone]

maybe put a MIDI interface on the PT rig to accept MTC and not have the MTC/LTC conversion

[SBP]

you can do midi over ethernet for positional reference
-Check out Audio Midi set up utility in your Mac.

Locking the 2 machines audio speed clock etc
Sync HD will not generate word clock only distribute.
So
Either get a work clock generator -and clock both to that
Or clock one of the machines (the slave) to the digital out of the other (The master)

Hope this helps

Bruno

[LukeFromBerlin]

Originally Posted by FajitaTone

maybe put a MIDI interface on the PT rig to accept MTC and not have the MTC/LTC conversion

Hi FajitaTone

I've tried that in the past too... strange thing is:
I am using the avid sync i/o as a master clock distributor, so I HAVE to activate it in the "setup -> peripherals" window by checking "enable sync peripheral"

But as soon as this box is checked, the the "MTC Read Port" option - where I could tell PT from which alternative interface to receive MTC from... grays out !

When I switch the "enable sync peripheral" box OFF .... my midi-interfaces (like e.g. a small Midisport) immediately appear as an option - not only for generating MTC (that's always possible) but now as a "MTC reader port" too.

But as the Sync i/o does not have a MTC input … it seems like it's an either/ or thing.... either MTC available through other interfaces > but no sync i/o (for clocking etc.)
or
Sync i/o available > but no other Midi interface (that can receive MTC) accepted anymore (grayed out)


Same seems to happen when trying to do it via ethernet (which I will try again soon more thouroughly, but I dought that the "MTC reader port" option suddenly becomes available as soon as a ethernet midi connection through the apple "audio-midi-setup" is up and running... ?)

Thanks everybody for your tips. Seems weird that the sync i/o doesnt accept MTC but doesn't allow other midi peripherals for MTC in to run at the same time ... any idea where that comes from ?