Video Satellite LE Sync Problems

[jmattchan]

We're trying to get Video Satellite LE running with HD picture and are having some sync issues and I'm trying to figure out what is wrong with my setup:

Mac Pro Dual Quad Core Nehalem Processors w/8GB RAM
OSX 10.5.5
PT8LE 8cs3
Video Satellite LE cs1
DV Toolkit 2
Decklink HD Extreme III PCI-E (SDI to Projector)
Fed 23.98 Tri-Level Sync, derived from 29.97 master blackburst
Apple Pro Res Quicktimes @ 23.98 fps
Pro Tools LE @ 23.98 with 0.1% Audio Pulldown, no Video Pulldown

For the audio systems:
Pro Tools Settings @ 23.98 with 0.1% Audio Pulldown, no Video Pulldown
1 Admin, 1 Satellite
Both systems are using a Sync IO (not Sync HD) locked to 29.97 blackburst

My understanding is that we could get a framelocked playback with the above configuration. With our current setup, the audio drifts against the picture. Also, it seems to try to "catch up" every once and a while and we get a jump cut.

We also tried to use the syncheck and the results seem to vary from playback to playback by over a frame.

I noticed in the Blackmagic panel that when I connect a 23.98 ref, it sees the signal as valid, and when I just connect 29.97 it does not detect a signal. We've left the derived 23.98 signal connected for now but I'm wonderig if this is incorrect.

Any help would be greatly appreciated.

Thanks,
Matt

[Postman]

Hi, I don't have experience with the Decklink card so cannot help there but maybe I can put us in the right direction.

First question, do you really need audio pulldown? Syncheck audio is meant to play at speed with no pulldown, 48000 not 47520, against 23.976 picture. This might be the cause of your drift?

Second, how are you "deriving" 23.976 tri level from 29.97 bi-level? Does that mean you have genlocked a tri-level generator to 29.97 black that you are using for SyncIOs?

The frame edges of 23.976 will always be changing against 29.97 frame edges, even when the signals are gen-locked. There shouldn't be a drift over a long period, but there WILL be moment-to-moment misalignment of one's frame edges against the other's (due to different frame rates), so I would expect lockups to vary each time you hit play, but staying within a frame's worth of error. Maybe that is the cause of your Syncheck trouble?

[jmattchan]

Thanks Postman!

My Syncheck test were actually done without any pulldown and I am also getting varied results using 29.97 quicktimes. I will take another look though cuz what you are saying about frame-edge alignment makes sense.

The mix itself is for a film and so I am fairly certain audio pulldown is required to sync up with the 23.98 picture (picture is a transfer from the DI, digitized off of 23.98 HDCAM).

For reference we have an NTSC blackburst generator feeding the audio systems' Sync IOs. It is also feeding a tri-level sync box that converts it to 23.98 which in turn feeds the Blackmagic card.

It is very hazy to me as to how I am supposed to genlock all these systems together. I'm not sure where the Ethernet Protocol comes into play with sync-ing all the pro tools timelines and how this relates to genlock.

[Postman]

Okay, I understand your setup now. There should be no drift between the two systems. By "drift" I mean that whatever error is present between them when you hit play should not change until you stop and play again. In your first post you say there is drift. Are we talking about the same thing?

Drift is one issue, frame edge alignment is a separate issue. Your only possibility for perfect lockups is when all three systems (Pro Tools, Satellite LE, Decklink) reference to the same sync signal, which will have to be 23.976 tri-level for this project. There will NEVER be perfect frame edge alignment with your current hardware at 24 or 23.976, only at 25 and 29.97. Unless you can replace both SyncIOs with SyncIO-HDs you're up a creek.

This is complicated further because, last time I looked, Digi does not guarantee perfect AUDIO sync between the main Pro Tools rig and Satellite LE's audio, regardless of the hardware. Maybe Bob Brown can set me straight if I've got this part wrong. So, if you're trying to flange Satellite LE's audio with Pro Tools audio, you may not get perfect results. I haven't tried it, I'm only going by what I've been told.

It is very hazy to me as to how I am supposed to genlock all these systems together.

Just answered, I hope.

I'm not sure where the Ethernet Protocol comes into play with sync-ing all the pro tools timelines and how this relates to genlock.

The Ethernet communications, as I understand it, tells each Pro Tools system exactly when and where to be at the start of playback. Each system is independently expected to actually make that happen. I do not think there is any sort of realtime sample-based or frame-edge sync signal carried via Ethernet.

[jmattchan]

OK I think we've solved it:

Postman (thanks again!) is right in that we are seeing two problems - drift and frame-edge alignment.

Talked to Digi and the frame-edge alignment (that i also assume was causing the stuttering) appears to be mostly solved by actually removing the 23.98 reference from the Blackmagic Card and locking without any reference. Apparently Pro Tools can synchronize over ethernet to something like 20 samples of precision. Syncheck tells us otherwise but the results are now much tighter (within less than 1/2 frame) and way more consistent than before. Also we were told to avoid video sync offsets with multiples of 4 (we had entered 8 and now it is set to 9) due to some odd unexplained bug. If we had two Sync HDs, hopefully things would be even tighter but I haven't had time to test this.

We discovered that the drifting problem was actually solved by releasing the audio pulldown on the Video Satellite LE machine. This doesn't really make sense to me since there was no video pullup or pulldown on that machine. But essentially, the picture was behaving as if it were pulled up and the audio drifted away. Maybe it is some weird thing where the Pro Tools LE system's master clock is the audio's playback rate and any pullup/down is done in relation to that. Couldn't get any verification from Digi on this. But there is no reason to have the audio pulldown engaged on the VidSat machine (they are not *qualified* for simultaneous audio playback). We just put this in to match session settings.

Either way, I managed to play back two reels continuously and the sync was bang on from head pop to tail pop. We were very happy, We are going to reinstall the system in our other theatre today. Hopefully it all works.

[Postman]

Thank you for updating this thread with a resolution! I'm very glad to hear you have a workable solution.

We discovered that the drifting problem was actually solved by releasing the audio pulldown on the Video Satellite LE machine...Maybe it is some weird thing where the Pro Tools LE system's master clock is the audio's playback rate and any pullup/down is done in relation to that.

Thank makes a lot of sense, it wouldn't surprise me if the developers had to make a compromise like that.

Talked to Digi and the frame-edge alignment (that i also assume was causing the stuttering) appears to be mostly solved by actually removing the 23.98 reference from the Blackmagic Card and locking without any reference.

I don't understand how this would have helped but I'd like to. Are you measuring the offset with a 24 frame display or through the Decklink's built-in downconverter?

[jmattchan]

Hey Postman,

OK here's an update - we are measuring the offset with Syncheck, test quicktimes, piped through a Christie CP2000 in a theatre. Our gut feeling now is that there is actually something wrong with our Syncheck. Today, it is showing an offset that is moving significantly but we are not seeing this. The offset is moving at least a few frames over a few minutes and but that isn't what we are seeing and hearing on the screen.

I also do not understand how removing the reference solves this issue. Is it possible there is something in the Ethernet protocol to handle sync beyond what you have described?

Also we have not checked sync off the downconvert although the processing is turned on and feeding a small NTSC monitor here. We will deal with that separately and may have to put in a delay in the audio monitoring chain.

The good news is that we are printing reels now with the mixers and sync is holding perfectly and head and tail pops are lining up perfectly.

Matt

[Postman]

Today, it is showing an offset that is moving significantly but we are not seeing this.

Mine has acted very strangely when the battery had been used for a few hours. A fresh battery fixed it. The thing is not exactly "flea power".

Is it possible there is something in the Ethernet protocol to handle sync beyond what you have described?

Sure, I suppose so, I have no direct knowledge about Satellite protocol. However, Bob Brown mentioned here that the protocol is very lightweight and should not cause significant cpu loading. Also, given the non-realtime nature of ethernet data packets, I cannot see how any realtime clock data could be relied upon. Maybe there are sync milestones or markers of some kind, I don't really know.

[jmattchan]

yeah it would be nice to find out...

i'll try a fresh battery for sure. :)

[Bob Brown]

As part of a the Satellite technology, we developed a "network clock" as a part of our play triggering mechanism. There are several techniques that are being used to provide a high level of accuracy.

If you are synchronizing two Pro Tools HD systems with SYNC's using video reference, the two Pro Tools systems will be "near sample accurate". These systems will be within a few samples of each other and will not drift.

If you are synchronizing a Pro Tools HD system to a Pro Tools LE system, there is a little more inaccuracy between the systems. Currently, there are a few challenges with synchronizing the LE system. The best way to tweak out the LE system is to use the lowest possible DAE buffer setting and put both systems on the same router/switch/hub. This will minimize the error between the systems. When using QuickTime, the video will be synced very tightly. If you were to compare audio between the HD and LE systems (not officially supported of course), you would see some inaccuracies.

I understand the current issues with AV sync on the LE system and we are investigating various things to improve the accuracy of the LE systems in a future release.