| MadSci Network: Computer Science |
Hello Lol,
Twisting the two wires of a twisted-pair cable is a way of reducing interference. There are lots of ways in which the outside world can disturb electrical signals on a simple wire. This is very a big problem if you want to send a lot of data down a cable very fast, and be able to read it twenty kilometers down the road!
The simplest way of sending data signals would be to have the signaler at one end of a single wire applying voltages, and the listener stand at the other end with a (grounded) voltmeter:
wire
apply voltages --> --------------------------------------- -> read
voltages
However, this wire will behave like an electrical antenna: the listener
will see the signal voltage, plus the local radio station, plus electricity
from the subway, plus 60 Hz from the power lines overhead ... it's a mess.
Here's solution number 1: use two wires, one carrying the signal and one
carrying a constant voltage (which may as well be ground = 0 volts)
signal wire
apply voltages --> ------------------------------ -\ read difference
apply ground ----> ------------------------------ -/
ground wire
Now, all of the noise sources that change the voltage on the
signal wire will also change the voltage on the ground
wire. When you subtract the two voltages, the difference will be the
original signal. However, now you've got an enourmous loop of wire - a
long signal wire, and a long ground wire, and a bit of space in between
them. If you know Maxwell's Equations, you'll remember that a changing
magnetic field going through a loop will make a voltage go around
it. So if a magnetic field (from radio stations, power lines, trolley
lines, whatever) goes between the signal and ground wires, it can raise the
signal voltage and lower the ground voltage (or vice versa). There's a way
to minimize that, though: twist the wires!
__ __ __ __ __ __
apply voltage --> --\ / \ / \ / \ / \ / \ / \__
\ \ \ \ \ \ __ read difference
apply ground --> --/ \__/ \__/ \__/ \__/ \__/ \__/
Applying a changing magnetic field to this picture will try to pump
voltage around any loop, say, clockwise. Every one of the twists behaves
like a little loop of its own. But a clockwise current in one loop will
exactly cancel with the clockwise current in the next loop, and in the
average over many twists, the effect of changing magnetic fields also
cancels out!
This is why twisted pairs have very low noise. The things to think about, from an electromagnetism perspective, are capacitance and inductance. A single wire is a capacitor, and can turn ambient electrical noise into voltages: adding a ground wire cancels out the capacitance. But a loop of wire is an inductor, and can turn ambient magnetic noise into voltages: twisting the wire cancels out the inductance.
Good question! I can't think of a reference for twisted pairs in particular, but you can read about inductors and capacitors in any physics textbook (like Halliday, Resnick & Walker), or electromagnetism text (like Griffiths, "Introduction to Electrodynamics").
Sincerely,
-Ben Monreal
Try the links in the MadSci Library for more information on Computer Science.