jeffx said:
Any idea how to fix the ground issue if it is at the capture card?
Finding a ground loop isn't very difficult if you have an ohm meter or continuity checker, but it can be time-consuming. I would try it with everything powered off. Based on the symptoms, I would also concentrate on the shield of the coax.
Check the continuity between the coax shield at the PC and the PC chassis. They will likely be connected. If not, a ground loop looks less likely.
If there is a connection, then that's pretty normal. But now we need to make sure that it is the ONLY ground connection for the shield. At one of the cameras, check for continuity between the shield and each power pin. There is a strong likelihood that the shield will be connected to the negative terminal. If not, you don't have a ground loop, unless the shield connects to ground somewhere else along it's run, like at a lightning arrester.
If one of the power pins is connected to the camera's shield, then you are tying the two shields together by using a single power supply. That may or may not cause a problem, depending on what happens at the other end of the cables.
But now you have to check the cameras' power supply. Depending on it's design, the power-supply's power cord may have two or three prongs. If it is designed to have two prongs, then it is isolated, and you should not have a ground loop at that point. If it has three prongs, you need to check whether the ground prong is connected to either power lead. If it is, you have a ground-loop between your computer and the cameras' power-supply.
The best solution to a ground loop is to get the grounding correct. If for some reason you can't, as a last resort a video-isolator (often called a "humbucker", after a company that makes guitar pickups) can be put in-line to 'break' the ground loop. Video professionals always have an isolator with them for ground-loop 'emergencies' and for testing, so it may be worth getting one to try. It can often come in handy.