Originally Posted by PhilNYC
So if you use a Neutrik digital XLR, it should be fine, right?
And aren't there additional benefits to running a balanced signal, particularly for long runs? BNCs are just for single-ended SPDIF, right?
Maybe. The connector's still not 110 Ohms, and the cables aren't particularly great at maintaining a constant impedance especially when subject to mechanical stresses. As for distance, due to the impedances, and signalling conventions, shielded twisted pair AES/EBU can maybe go 300 feet, but AES3id, which uses coax, can go over 1000 feet.
Balanced is useful for rejecting noise in so far as the sending and receiving circuits maintain their impedance balance over the frequencies of interest. For high frequencies like digital audio signals, that's at least 10 MHz if not more. I don't think there are any balanced circuits that have matched impedances to that high a frequency, so balanced doesn't buy you much for RF. That's why high frequency transmission uses coax instead of twisted pair.
For coax, there are certain construction techniques well-known in industries where noise rejection matters that are used to deal with noise impinged on the cable's shield. The important thing is to never let common-mode noise enter the chassis. One example is a chassis-mounted BNC jack whose ground is capacitatively coupled to the chassis to prevent low-frequency ground loops, but still allow a low-inductance path for RF to the chassis, which in combination with a ferrite bead on the signal conductor and the capacitance of the shield of the input pulse transformer which is tied to ground forms a pi-filter for preventing common-mode RF noise from entering the chassis.
Shielding is important because jitter is just one effect of noise injection into the system. There are lots of things out there (eg. cell and cordless phones) that can put RF noise into the system and not only introduce jitter in many parts of the design, but can also mess up other parts of the system. Audiophile designers seem to like using high-speed opamps that often don't have well-designed power supply bypass structures (eg. through-hole parts, bypass caps far away from the the device, etc.), and that's an invitation for RF to cause trouble.