For the NVEM driver (I have one sat on a shelf somewhere) - that's a single-ended output drive heading to the TB6600 which is a simple single-ended opto-isolated input device - a twisted pair designed to reduce signal noise by the merit of common-mode noise reduction doesn't really work particularly effectively. That's not to say that it won't work, but just that it won't add significant noise immunity. To be brutally honest, the drive requirements of the TB6600 (forward current draw) means that it's not particularly susceptible to noise in any case. Good wiring strategies (avoiding parallel runs with high frequency/high current signalling, and/or shielding) would be, broadly, as effective.

My one concern with the NVEM device driving the entry-level stepper drivers is borne from practical experience - the step-outputs are only 2.5us wide and can have a poor/unreliable performance with the stepper drivers. You should be able to find a post by me on this subject elsewhere on the forums.

Jazz: cba googling (need to get motivated to go to work), I think the major discriminator between cat 5/6 is the number of turns /inch in the twisted pair, and possibly also the screen coverage - but as you say, don't quote me on it. I've certainly got a reel of cat-5 in the shed that is solid core.