All very true - in fact I wrote Sound Devices tech support and asked if the audio sample rate clocked itself off of incoming timecode when the recorders were supplied with external code and their response was they do not. But its internal TC clock is derived from its audio clock and is highly accurate. That's one reason they suggest using the recorder as the timecode master.

The only time TC will help with a camera that doesn't have timecode I/O and genlock is if you use the recorder as master and send LTC to a camera audio track, then when you import the video clip you instruct the NLE to align the clip into the timeline using the recorded LTC and ignore the TC recorded with picture. Then when you drop the BWF from the recorder into the timeline it will align itself to the LTC. Unfortunately the only editors I've heard of that can read LTC and align to the timeline based on it are Final Cut Pro and Avid and if we're not using either of them recording the LTC is pointless.

You can also jam a smart slate and when importing the cip into the timeline, align the timeline with the numbers shown on the slate, then when you drop the BWF onto the timeline it will autoalign. But unless you have a h*ll of a lot of clips to sync, that seems like an awfully expensive alternative to just visually finding the frame where the sticks of an old fashioned slate bang together and lining it up to the spike on the audio waveform.

You are also absolutely right that unless you can slave the video sync on the camera to the wordclock on the recorder or vice versa nothing you do with the timecode does anything to preserve sync over long shots. At best all TC can do is help with the initial alignment but when you move away from that alignment point, sound and picture will eventually drift out of sync unless the audio and video clocks are slaved to each other. How fast they'll drift out depends on the quality of the clocks - with some cameras you might get away with an hour before you're a frame off but with others it might only be 5 minutes.

Absolutely, but not many, in fact none that I'm aware of, prosumer cameras allow TC input. That seems to be reserved for hi-end studio cams.

Canon XL-H1 and XH-G1 have TC I/O and genlock but they're the only cameras I know of under 10 grand that do.

Its well worth the extra money to buy the 702T in case one purchases a TC capable camera, hopefully "scarlet". I did two short films using the good old line up the audio/slate and probably will continue to do so. In the future, even when using the smart slate, the clapper of it would still help in case the Timecode does drift. Im looking for new ways to save time, hopefully timecode will help with this, but i still wouldnt fully trust it cause machines aint perfect either :)

BTW, SD says their timecode recorders use the frame rate to calculate subsequent timecode values (pg 30 of the 744T manual). So you're right with trying to get it right.

... And then there is "F Samping Rate Modes" which applies when recording for material that will have a 3:2 pulldown removed. I will have to worry about and understand this setting, but my brain hurts eough as it is.

The F timecode rate modes were included on the Deva to help when telecine had older Fostex DV40 machines. There was an issue in them that prevented them from reading and placing the proper timecode from the audio, so Zaxcom after getting complaints from mixers about it, introduced the F version of timecode specifically for the Fostex machines. SD when they introduced the 744T, included the F format since many post houses tend to be SO slow in updating their equipment and it's still an issue. Unless the people doing telecine for you tell you they are having issue with your audio and timecode, then you shouldn't use the 'F' flag on your timecode.

Wayne

My head hurts now but moving on, if im shooting HD at 23.976 and using this setting on the 702T as the master and sending out 23.976 timecode out to a smart slate and to the TC in on the camera,
where is the bottleneck here, other than maybe code drift after a certain time?

The "F" mode on SD recorders helps facilitate direct importation to an NLE timeline in order to maintain sync with film that has been telecined MOS. The recording is made at 48.048Khz, 30ND frame rate, while at the same time being stamped as 48Khz, 29.97ND. This automatically introduces the 0.1% "pulldown" that will keep sound and picture in sync.

Question for Wayne (or whoever): Do you know of a workaround for the Edirol R4 Pro, which does not have an "F" mode, that would accomplish the same thing? (I mention the R4 Pro, since you recommended it as a viable alternative to the 744T. How does it compare with SD on the basis of sound quality? Thanks.)

See the following Tech Note on the Sound Devices web site ...

http://www.sounddevices.com/notes/re...sr-frame-rate/

In short, you should be recording at 48kHz with 29.97 NDF timecode for HD picture shot at 23.976 and edited in an NTSC environment. I thought you said your camera didn't have a TC input?

no it doesnt steve, but when i'm making my soon camera purpose its gonna be a must, thats why iam getting the 702t. so when editing in a 23.976 timeline, doesnt it make sense to record the audio at that rate? unless of course we wont be removing pulldown upon import

I'll confess that there are moments when I think I understand all this with crystal clarity and then the very next moment it's as clear as a pea-soup fog! But if you're going to edit in an NTSC environment the video will be converted to 30 FPS so for your audio frame count to match the eventual video frame count, it needs to be at the NTSC rate.

alright thanks. When are we talking about the video being converted, when rendering perhaps? when importing into the NLE and removing pulldown from the 29.97, it'll be 23.976. so, if one edits at this rate, how does the audio match?
its only gonna match when reinserting 2:3 pulldown, aint it?

i think my question is good, just maybe you misunderstanding me

No, it's an excellent question. I recall seeing somewhere - I think it was Wolf Seeberg's site - just within the last couple of days that with most, if not all, HD cameras generate timecode at 29.97 non-drop even though they're shooting at a 23.976 frame rate. So while 1 second of video has actually recorded 24 frames, the timecode recorded for that same second counts it as 30 frames. In other words, a frame of video is not equal to a frame of timecode - could it get any weirder!

So, my gathering is, its best to record the audio at the 29.97 non drop on the 702t, to edit on the 23.976 timeline which will later match when rendering as pulldown will be reinserted to 29.97 NTSC. yes?

That's what Sound Devices recommends. The 23.976 timeline is used where the video is slated to ultimately be converted to film for release. If your release format will be video you'll be working on a 29.97 FPS timeline.