Dave, sorry, I'm morbidly curious as to your choice of wiring the CAT-x UTP cable with each pair connected together. There's nothing much wrong with this and it probably presents a clean and cheap 4-core cable and bi-wiring each pair like this will reduce the resistance of the wire. But you used the term twisted-pair, and that has a particular connotation with respect to EMI - I guess you're aware that the way that you've wired that is not exploiting the twisted-pair?

Anyway, that's casual banter. If I'm going to reply I'll try to add some value. Jazz is correct - you're having to short the analogue and digital ground together for the VFD. I did do this on mine, also a NVEM, and it worked, though I did get some fluctuation in final spindle speed that may or may not have been associated with this and a ground loop. I never did find out what because I replaced the analogue speed control with a £2 RS485 serial link to the VFD.