A balanced view of impedance
Why do we want balanced interconnects? Basically there are two reasons:
- If we run two wires that are closely twisted together, AND that have identical impedances to ground, any induced noise or hum voltage will be equal in both wires. Then if we use at the receiver end a circuit that only amplifies the difference between the wires and not the common (noise) signal, we have cancelled the noise and hum;
- If we, in addition to the two signal wires, run a separate shield as a ground between two pieces of equipment, any ground currents between the equipment generates the same induced junk on both wires, and then the above argument becomes valid. Also, since that screen or shield is not used as reference on either side, the actual noise voltage that exists across the shields' impedance as a result of ground current is immaterial.
Two things are immediately clear: You MUST have equal source impedances for the two signal wires, otherwise the noise becomes unbalanced, and appears at the receiver as a difference signal and will become part of the signal. Also, there is NO requirement for the signal at the two wires to be balanced. In fact, one of the wires may even have no signal at all, without any negative effect to the working of the balanced interconnect.
This is important. It is true that in balanced output stages you often see the two signal wires driven in anti-phase, with equal amplitude signals. The reason for that is that it gives you the double signal level with a given supply voltage. With a standard +/-5V supply you can get say 4V peak undistorted, single ended, which is about 2.85VRMS. But if you drive a balanced cable with +/-4V peak, the receiver sees 8V peak which is of course about 5.7VRMS. That extra 6dB helps again to increase signal-to-noise ratio over the line. But it is important to keep the facts straight: differential, balanced signal drive is not required to obtain the advantages of balanced interconnects.
For those who want more detail, there's an excellent write-up on balanced issues, much better than I can do, by Graham Blyth.
Back to balanced-to-single-ended cable page