Waveform monitoring: Q I readings

[David Hentz]

So I've run into a brain-scratcher:

My client is having tremendous ghosting/ringing issues in the video feed. After hooking up a waveform monitor to test the feed straight out of the distributor, then testing that same feed after it has gone through an AV Mixer, I have noticed only one real difference. While checking the vectors, the Q and I positions are EXTREMELY different from one another. Coincidentally my color bars look great when coming straight from the distributor, but bloom and bleed over extremely after going through the switcher.

My question:
Does this sound like a proper correlation? And, if so, how can I go about fixing it?


The setup:

The client runs a 5-camera setup. They are all Canon XL-2; four are hard-wired with the 5th being wireless. Each wired camera runs BNC out to a Beamer Analog-to-Optic converter. Optics then carry the signal up to the building. The signal is then run through a Beamer Optic-to-Analog converter to take us back to BNC. The wireless camera is picked up by the receiver and immediately sent over BNC. At this point, all cameras hook into a distributor (the same distributor.) From there, video feed is sent into a Grass Valley Indigo AV Mixer (with updated firmware) (Video is being fed in via composite in). Video feed seems to be fine at this point.
From here...it starts to get dicey. The client sends multiple versions of the feed out. An SDI output is sent to a ClickEffects Crossfire; picture is extremely crisp. Among other outputs, a program feed is also delivered, via composite output, to an Edirol 8v switcher (analog input only.) This is where we see video issues.
Into this Edirol not only is the Grass Valley feed being used as an input, but each of the original 5 cameras. When the Edirol is pulling a feed from the distributor directly (the same distributor the Grass Valley pulls camera feeds from), the picture quality is great; at least, as good as a Canon XL-2 using analog output can get. However, when the switcher pulls a camera via the Grass Valley feed, ghosting and ringing occur. It is through this feed we see our Q and I dots move to opposite the blackburst, instead of in between our other colors.

Edirol 8v switcher is analog; there is, regrettably, no Ref-In. Thus, timing cannot be achieved via this switcher.
Client is using a Blackburst generator to sync signals across production (minus the Edirol 8v switcher). Waveform monitor verifies sync of Grass Valley feed before it enters the Edirol. Camera feeds coming directly off of the distributor are NOT synced.**** Individual cam feeds look great, though. It is only the feed that comes from the Grass Valley which looks bad.

****We ran camera feeds into/through an Ocean Matrix 8x8 Router (which has a Ref-In). Even after referencing cam feeds before sending to our distributors, we saw no change.

We use a DPS-165A Frame synchronizer (which is properly referenced by the BlackBurst) to sync our Grass Valley feed into the Edirol. (We also use this device to make other adjustments. i.e. recover Video IRE, etc.)

again:
Grass Valley Indigo AV Mixer
Edirol 8v switcher (analog. no ref-in)
BlackBurst
Ocean Matrix 8x8 Router (OMX-SW8x8NT)
Ocean Matrix 5x1 Distributor (OMX-7015)
Canon XL-2
Beamer Analog-to-Optic and Optic-to-Analog converters.



Thoughts? I would especially love to hear/learn about Q and I measurements/corrections.
Thanks guys! (and Dolls)