ATEM Television Studio sync drift

We have an issue that's not entirely unexpected, but potentially hard to solve... We have a unit with 5 incoming fibre feeds, into Studio Converters then to individual Hyperdeck Studios for iso recording with the deck outputs heading for the Television Studio for a line cut, recorded to a 6th Hyperdeck Studio (the Pro was delayed and unavailable when we built this). All use floating sync, the 5 iso decks presumably syncing to incoming feeds from the cameras, feeding the ATEM which buffers the inputs to allow switching.

Our issue is that over the course of a 90 minute switched recording, the line cut is dropping a frame about every 30 minutes, and is 2-3 frames out of sync with the iso recordings at the end of the gig, and with its own audio. Video frames are dropped, but audio continues uninterrupted. It's recording a feed of mixer audio (via one camera) which remains in sync with its iso, so it seems to be about the mixer self-clocking and blowing out its frame buffer from time to time.

The cameras themselves (all PDW-700s) remain in sync with each other throughout, so Sony's sync lock seems to be very consistent - but not necessarily the same as other devices in the chain.

An easy suggestion would be to sync all the decks and the mixer with ref, but thinking deeper that may not in fact fix it - if the camera feeds aren't identically timed with the sync generator, frames could be dropped at the recorders rather than at the mixer, and lost forever - at least now, we're recording all the frames somewhere!

Traditionally one would genlock all cameras and ensure that everything danced together - but the Camera Converters don't provide that option (except perhaps by some device able to derive a ref signal from the return video feed, if it's locked to source, which it may not be). Using an HD Studio Pro recording timecode mightn't help, since code would be recorded irrespective of the dropping of frames, and 6 Pros would record identical code with differing pictures.

So I'm thinking this may not be easy to fix, and 2-3 frames an hour is hardly a killer, but annoying in a multicam sync-up. Thoughts?

With the iso recording, the audio sync is OK of each camera?

As you said the obvious possibility are:
If so as you said, maybe consider a sync generator to the hyperdecks and the ATEM.

As you need 7, and BM's can only do 6, look at AJA (sorry BM!) Believe theirs does 7. http://www.aja.com/en/products/mini-con ... /#overview

In terms of audio sync, can look at Behringer UltraCurve Pro http://www.behringer.com/EN/Products/DEQ2496.aspx but that would be more if it is a consistent issue rather than over time.

I'm not 100% sure about what I am about to write here and may do nothing but:
Possibly look at a micro video hub, and have sync connected to it. Connect the cameras as inputs (and the ATEM out as an input as well) and outputs go to your 6 hyperdecks and your 5 ATEM inputs. Also sync to the ATEM. You would have set ins and outs routed but would mean cameras going "straight" to the ATEM rather than looped out of the hyperdecks.

Does not sound like an easy issue to resolve.

Hi Daniel

Thanks for your suggestions - yes, audio sync out of each camera is fine, with each other and with a separately recorded set of mix tracks. The line cut is a frame or two late against audio but that hardly matters if it's consistent. So keeping audio in sync isn't an issue, but avoiding dropped frames as the mixer buffers the camera/deck feeds is. Unless a separate ref feed is run to each camera I'm not sure that adding any amount of ref at the receiving end will solve the issue - in fact, allowing them to free run as we have at least means they'll record everything that arrives - if they're locked to local ref they may not. Yes, AJA's sync generator is a fine little unit - we have one but decided not to use it for this reason. Either way, sending each camera to the Hyperdecks before the mixer should protect the incoming feeds, which as you say, leaves us with this gnarly issue and no obvious answer (other than another cable to the cameras, along with a comms cable since the Studio Converter's minjack comms inputs are pretty nasty - I'm about to throw a Behringer SRC 2496 at that problem for 2-way AES/EBU comms).

No problem! As you said, there is no obvious answer! Assume you have tested with the sync generator at the non camera end with no success.


It does look like running cables may be the only thing that resolves it!

This is only a guess, but I would look at the cables going from the HDS to the Atem it is possible that you have a connection that is causing a problem, or possibly something inside the HDS or near the ATEM is causing a glitch in the signal, since all of the iso's seem ok with no dropped frames it is possible the problem lies post HDS

Sharyn

Have you ever tried to switch one of the Iso recorders with the mix recorder? Just to make sure, that it's not the (6th) recorder who has a problem.

[Gary Adams]

This would appear to be a normal behavior for free running devices. In the analog days, sync had to be accurate +/- 10 Hz at 3.5 MHz, but I would guess devices these days cannot hold to such tolerance. So if the cameras are running slightly (very slightly) fast with respect to the switcher, the frame sync would need to drop a frame as the verticals collide. As stated above the best and highly recommended solution is to genlock everything which implies an additional cable. There is one trick that may be worth trying, although I have not tested this myself. You could take the output of one camera and genlock a local sync generator (requires D/A of the digital signal) and feed the ATEM Reference. Since you say the cameras seem to be similar, this would lock the ATEM to at lest one of them. However, it probably would force a near fixed frame delay for all cameras but it wouldn't drift with respect to one camera, at least. It should be tested first since it has potential of feedback or living near the vertical crossover for long periods of time. Maybe a science experiment but might work or make the dropped frames over 90 minutes a bit less.

Gary

For feeding sync to the cameras I would try to feed a black video over the SDI program feed of the Studio Converter. With a SDI to Analog converter at each camera you should then have a black burst to genlock the cameras, of course you need to genlock the rest of the gear also.