Noise/interference can come from two sources.

Either wire bourne, i.e. noise from one source travels through common wiring and into another device, or air bourne, i.e. something unsheilded generates electromagnetic interference (EMI) which then creates a voltage/interference in any nearby conductors.

The mains filter should stop any interference from entering via the mains, however your system will still be susceptible to EMI. The point of sheilded wiring is to stop the EMI from reaching the wires within the sheild, and drain any generated voltage to ground. If you connect both ends of the shield, if whatever either connect to is at different voltages, even a very small difference, then current will flow through the shield start producing EMI.
The big issue with plasma, is they're a very good EMI generator due to the high voltages when striking an arc, and any metal nearby will absorb that causing a voltage to be generated in it.

Which reminds me of one thing, have you ensured the plasma table/frame itself is well grounded?
If the table isn't well grounded, then that could be the source of your problems, although I can't decided what would be the best grounding technique. If you use a ground other than via the control box, then you'd have to ensure nothing connected to/within the control box also connected the table to ground, otherwise you'd end up with one very large ground loop. And if you grounded it via the control box, you'd have to ensure all the connections were perfect.

Have you tried over on CNCZone to see what's suggested to combat noise on plasma tables?