I'm using an XL H1 (classic; not the S model) and an Edirol R4 Pro to record dual-system sound. I've done various tests to find the best way of synchronizing them — camera set to rec-run, roll camera first then roll sound; R4 set to internal, camera set to external TC; etc. — but I always get the same result: The sound is always somewhere between three and ten frames ahead of the picture. This is true regardless of whether I'm shooting 60i or 24F.

Now, let's be clear: Being able to get sound within half a second of sync automatically before slipping it into place by hand is a great improvement over not having any automated synchronization at all. But it's not perfect, since I still need a reference of some kind to sync my sound. For drama, that's no big shakes, since every take gets a slate anyway. But I shoot a lot of documentary material, where slating is awkward or downright impossible.

My workaround is to record on-board sound as well as outboard sound and use the on-tape sound as a sync reference. Which is fine, I'd be doing that anyway just in case, but it'd still be great to figure out how to sync these two devices in such a way that they're frame-accurate.

Does anybody out there have any pointers for me?

(I omitted almost all the details of the various tests I've done, just to keep this short.)

Jeff,
This may be inherent to the way Canon is processing the picture. You don't say you are using HDV or DV in your tests but there is a noticeable amount of picture delay from reality when you look at the output of the camera on a monitor. Maybe this is what you are dealing with as the Edirol doesn't delay the sound for processing as much as the camera does. There also may be a difference in the delay between DV and HDV. I am not sure why Canon doesn't take this into account in the output timecode of the camera but we know the XLH1 cameras are from the consumer camera part of the company so maybe they didn't design this function properly. Sending the audio to the camera is probably the best record of what should be the sync. Sometimes you can set up an offset so the sync matches up but I don't know what your software hardware combos are capable of. Different frame rates and types of recording could end up with different offsets. Maybe you should try and do a test to figure out the offset for each rate and whether it is stable. When you are doing double system you should try Free Run and if you can send the deck sync as well as timecode you should get the most reliable results.

Thanks very much for the reply, Daniel. You're right, I did neglect to mention that I'm shooting HDV.

The XL H1 handles timecode very strangely in 24F mode — or maybe it's the combination of XL H1 and Final Cut Pro. Rather than giving you 23.976 media files with 24 fps timecode, you end up with 23.976 media files with 30 fps timecode in a particular drop-frame fashion. I forget exactly which frame numbers get dropped, but it's one frame out of every five in a consistent cadence.

I thought this might be the problem, so I did some tests at 60i. No joy. I got the same result.

I think you're on to something about the inherent delay in the camera; I wouldn't expect the camera to do all the necessary processing in less than one frame. But I have to wonder … exactly what's the point of a timecode-out spigot if what's coming out doesn't match what's being laid down to tape? What else would you use the TC OUT for, if not synchronization?

I spent some time googling around to see if anybody had written about timecode-out latency on the XL H1, and I came up empty. But that doesn't mean it's not real; it just means I didn't find anything.

Thanks again for the reply.