[old_tv_nut]

Don't worry about 1% tolerances - this is a much greater effect.

As dtvmcdonald says, this could be a problem in your source.

But to make sure, I would check that the complete path through R64 and L13 is Ok (2200 ohms), and check C1C by temporarily bridging an approximately 40 microfarad capacitor from line 64 to ground. Then, if the problem is still there, investigate your source.

As dtvmcdonald says, this could be a mistermination of a long cable somewhere, but I think it would have to be a really long cable, like 200meters (656 feet). The luma delay line is roughly one microsecond delay. A back and forth again echo time would be twice that. If you assume the visible width of your picture is about 50 microseconds, the ghost is indeed about two microseconds.