Speaking to the issue of dry-pairs - the IT folks at the college I work for have all but abandoned a very clean copper plant. The only apps remaining on it are MAC machines and some security phones and related apps. So, I have probably 45 Jensen JT-112 transformers around the joint to haul audio from almost every building on campus. Just for grins, I ran SMAART (live sound, FFT-based, audio analysis program) on my longest run, which is approx 2800' and crosses (5) punchblocks: the audio was essentially flat from 50hz past 20khz and overall level loss was about 4dBv. I then looped the circuit back (5600') and measured again: the results were a roll-off starting around 16khz and down 5dB at 20khz with an overall attenuation of 9dBv. That line was driven by an ATI DA out. Needless to say, I am a big fan of dry-pairs when dealing with a distributed facility. The xfmrs are about $45/each - I build the terminations on either a rack panel or in a project box.
-D