If you examine an XLR connector closely it has three pins. One carries the original signal, another the exact inverse (or opposite) of the original, the third is the ground. The theoretical advantage is that at the end of the reproduction chain the signal that is on the two non ground pins is examined and anything common to those pins must be noise (something that is not part of the original signal.) This

You got it!

On a purely practical level, my listening experience over the years is that with unbalanced interconnects, certain equipment connections (unit A to unit B) will sound good with cable A, better with cable B, and godawful with cables C and D. This would not seem to make any sense, but it is clearly audilble in many cases.

The problem is magnified with cable lengths over 1 meter in length. Some equipment is less sensitive to this; some very touchy.

With balanced interconnects, in my experience, you get absolutely no loss of quality from the interconnection every time (assuming the equipment is truly using balanced circuitry). The cables can be 10 meters long, and there is never any difference in sound between cable A and cable B that is audible to me.

As stated in the post above, in a balanced interconnection the two signal conductors are shielded by a shield that is truly grounded at both ends, and the two signal conductors are driven by a differential amplifier, and the receiving circuit will normally have a very high CMRR, or ability to reject all noise from any source and not amplify it.

With unbalanced interconnects, the fact that the shield carries signal current means that it cannot be truly "grounded" at both ends or there would, de facto, be no way for signal current to exist. Both ends of the shield are a "floating" ground in most cases, and this creates a complex circuit network consisting of the complex impedance to ground of the output circuit at one end, the complex input circuit impedance at the other end, and the resistance, capacitance, and inductance of the cable in between. This is an extremely complex network that is difficult to model with a computer. Trying to measure what is happening with a scope or other instument is also problematical, because the instrument cables and input characteristics modify the network characteristics by their very presence when connected.

This complex network changes due to small variations in cable geometry (construction) from cable to cable which affects capacitance and inductance somewhat.

This complex network is highly subject to producing ultrasonic oscillations which can overload the input circuitry of the device at the receiving end (preamp, amp, etc.). This ultrasonic garbage is obviously not audible, but it can swamp the audble signal and cause the input circuitry to go into non-linear modes of operation, causing the audible signal to be somewhat distorted. In severe cases the sound will be seriously affected, in other cases the problem will be more subtle; a very slight deterioration in sound quality.

These ultrasonic oscillations may be continuous, but in most cases they are short bursts of ultrasonic energy, stimulated by various amplitude peaks of the audible signal, and dying out quickly, but recurring frequently.

All unbalanced connections are a complete mess from an engineering standpoint; highly subject to high noise levels and oscillations.

There are two ways to minimize this problem.

One is for the equipment manufacturer to design a "hard" low-pass filter at the input of the circuit to supress anything above 20KHZ; very few do this. Some even advertise that their frequency response goes up to 50 or 100 KHZ, which is a very stupid design decision.

Another is to design the unbalanced interconnect with one internal conductor to carry the "high" side of the signal, and a second one to carry the "low", or "floating ground" side of the signal, plus a shield that surrounds them both and is tied to "ground" at one end of the cable ONLY. This changes the network in ways that make oscillation less likely and makes the shielding more effective. This is not nearly as good as a balanced interconnection, but it minimizes some aspects of the problem and makes severe sonic degradation less likely.

There are a number of commercially available cables that use this 2-conductor plus shield geometry, such as Audioquest Viper or Diamondback, to name two examples. They generally have less problems than cables with only one internal conductor and a shield. It is CRITICAL that the shield only be connected at one end and NOT connected to anything at the other end. Merely having 3 conductors does very little to help the problem.

Incidentally, with power amplifiers, this ultrasonic energy can cause huge amounts of ultrasonic power to be generated, causing the amplifier to overheat for no apparent reason, sometimes damaging output circuitry or blowing fuses. If your heat sinks or case are too hot to touch comfortably when volume levels are moderate...be suspicious of a problem of this nature.

Also; claims that such materials as ultra-pure copper or silver in cables create improvements is absolute nonsense, in my opinion; these do not affect network characteristics significantly. What MAY be helpful is certain dielectric or shielding techniques in the cable construction, such as teflon insulation or better separation of conductors; these CAN and DO change the network capacitances significantly.