View Single Post
Old 22nd October 2017
  #6
Here for the gear
 

Quote:
Originally Posted by JohnRoberts View Post
For audio interfaces all (most?) well designed gear will have a finite source impedance (build out resistors), so the dominant cable frequency response effect will be a simple LPF formed by the build out R and the cable C.
Yes. Standard source impedance for audio line level is in the 100 to 600 ohm range, and load impedance is 10K ohms or higher (some older gear as much at 100K).
Quote:
Characteristic cable impedance for modest length audio cables is mostly insignificant at audio frequencies. Even DCR is not very significant except for speaker cables where the current can cause IR voltage losses in long speaker runs. For passing RF and Digital signals the wavelengths are short enough wrt cable length, that characteristic cable impedance and proper terminations do matter (reflections inside the cable interfere with signals).
Yep. This is why we don't "match impedance" for line-level audio. So cable impedance is a non-issue. For more, see http://www.rane.com/note126.html . I think they are a little cavalier on this point; what I learned was that you should start thinking about transmission lines when your cable length was only about 1/10 of the wavelength. (i.e. if a 10th of a cycle or more will fit in your cable.) But that still means you're not worrying about it for audio unless you're working on analog phone lines.

Regarding HF rolloff, here's a real-world example: I'm running 50 feet of Belden 8761 STP between balanced output (ok, it's only impedance-balanced) and balanced input here. Belden shows this stuff to be 47 pf/ft in that configuration. The output impedance of the mixer that's driving it is 120 ohms. If you work the formula from the Rane article, you'll find that the 3dB down point is up around 500 kHz! I'm not losing any sleep over that.

One other point that no one has mentioned: Using twisted pair has a real benefit even for unbalanced connections: The twisting reduces the magnetic loop area of the cable. This reduces inductively-coupled noise pickup. It doesn't do a thing for capacitively-coupled noise (ie electrostatic, ie most RFI); that's what the shield is for. And inductively-coupled noise is usually the lesser problem, unless you have a LOT of current flowing through AC power wiring near your cables. But STP is cheap and readily available. Certainly, if you are having an inductive-noise problem, it's an easy thing to try.

n.b.: The folks over at Blue Jeans Cable say otherwise. See http://www.bluejeanscable.com/articles/balanced.htm . They're correct that twisted-pair gives no common-mode noise rejection when used with an unbalanced input, but then they conclude that it has no benefit at all with an unbalanced input, that coax is superior. They're mistaken. They are also correct that such cable, used that way, will have a higher capacitance than simple coax, but you could add a lot of C to the cable before HF rolloff becomes a problem.

Last edited by RickBrant; 22nd October 2017 at 04:27 AM.. Reason: added detail