I assume the problem is false stops, and Mach3 trips out when the plasma starts up?

My mk1 router machine had a similar problem (steppers interfered with limits), and I disconnected the limit/home cables one at a time to track it down. The longest cable (z axis home) caused the problem on mine. For mk2 all cables are now shielded, the shields all run to a common earth at one end only (star pattern?), and the cable runs for motor and limits are in seperate energy chains on either LH or RH side of the machine. I've a bit more work to do before firing it all up, but hopefully this will solve my problem.

I'll also experiment with 0.1uF capacitors across the limit terminals on the control board if I still have problems.

Any of this applicable to your machine?