Calculating Crosstalk: Electrical Currents, dB & Ideas

[patrickewen]

I am currently looking into crosstalk. I have never really heard of this and I wondered if anyone could help me.

If I have two wires wrapped together which have an electrical current running through them, hows to best way to find out the crosstalk? From what i have been reading, the leakage from one wire to another can be capacitive or inductive and am i right in saying its measured in dB?

Any theorectical or practical ideas would be great! :)

 

[berkeman]

There are several kinds of crosstalk, including near-end crosstalk (NEXT), far-end crosstalk (both of those are for adjacent twisted pair cables), and straight capacitive crosstalk.

NEXT: http://www.flukenetworks.com/fnet/en-us/supportAndDownloads/KB/Datacom+Cabling/dsp-4000+series/Near+End+Crosstalk+(NEXT)+-+DSP+4x00+CableAnalyzer.htm

The NEXT and far-end crosstalk come about mainly due to inductive coupling between the twisted pairs. In order to minimize this effect, multi-pair cables use a different "lay length" for each twisted pair -- that is, the twist rate is slightly different for each twisted pair. This keeps the loops from lining up next to each other for the whole cable run, and generally gives pretty good B-field coupling cancellation.

Straight capacitive coupling between PCB traces or adjacent wires in a ribbon cable, etc., is just a straightforward capacitive divider effect, with the source impedance and load impedance taken into account. When you have two PCB traces running together over a ground plane, then the capacitive coupling increases with the capacitance between them (like when they are moved closer to each other in the layout), and decreases with increasing capacitance to the ground plane (like if you use a 4-layer PCB with its internal ground plane layer, versus a 2-layer PCB with the ground plane on the opposite side).