Balanced vs. unbalanced

Eldartford is right on the money.
There is so much confusion about this I used to actually print and hand this pamphlet out to seasoned pros who were still debating the pin 2 pin 3 hot issues.
Handy for the DIY types or for general snake oil detection.

http://www.rane.com/note110.html

I prefer to stay with the design of the system. If the components are not truly balanced to begin with then I try to stick with RCA connectors.

I go with the alternative balanced connections when I am running the signal to a secondary listening area as this is the only alternative since my pre has only one set of RCA and another set of outputs that is balanced.

In reality there are very few sets of components that are truly balanced to begin with.

Bob - you asked "For example, if an op amp has a bandwidth of 10 MHz, what happens when its fed a 100 MHz signal?"

Most likely RF rectification - conversion to DC or low frequency noise if 100MHz (or higher frequency) is modulated. It is caused by different response for positive and negative slope and is present in practically all OP-Amps or Instrumentation Amps.

OP-amp should always have RC filter if front of it to prevent it. Common mode rejection of differential (balanced) connection works at lower frequency range while normal mode rejection (filter) prevents RF rectification.

One more thing - Input does not have to be directly an OP-Amp. It could be, and often is, coupling transformer. Transformer is a very effective method of removing common mode, breaking ground loops, blocking DC and limting rf (filter).

Transformers add a little THD at the lower frequencies. Rowland is using very good Lundahl input transformer in their new Icepower based amps.