10-03-10. Phd: The Audio Research amp that I purchased is a true balanced design. But more importantly I like the way it sounds. I used to own it two years ago and never took advantage of its balanced technology but was thinking that this time I should try it in its fully balanced mode because maybe I missed something but now I'm not sure.I could be wrong, not being familiar with the specific design, but I suspect that you were getting much of the benefit of the amp's fully balanced internal design without realizing it (contrary to some of what has been said above).
Since the amp is fully balanced internally, and provides rca as well as xlr inputs, I would expect that when the rca inputs are used there is circuitry at the front end of the amp that performs a single-ended to balanced conversion. That would allow the benefits of fully balanced INTERNAL design to be retained, at the expense of the possible sonic side-effects of the extra circuitry (which I would expect to be minimal or negligible in an ARC or other good design).
Some of those benefits, which have not yet been mentioned in this thread, are cancellation of some forms of internally generated distortion, and loading of the power supply that is more constant than with most unbalanced internal topologies (other than Class A). That in turn reduces internally generated noise, and possibly also inter-channel crosstalk in the case of a stereo amp.
Those benefits would NOT be realized, btw, in the case of a fully balanced amp that does not provide rca inputs, if a simple rca-to-xlr adapter were used at the inputs, which is an approach that would make no sense with a fully balanced amp. They WOULD be realized if something like a Jensen transformer were used to convert single-ended to balanced.
As far as the benefits of balanced EXTERNAL connection are concerned, I would add greatly reduced susceptibility to ground loop issues to what has been mentioned. That will sometimes be significant even with short cable lengths. That benefit, and also the common mode noise rejection that has been mentioned, will be realized regardless of whether the components are balanced internally or not. Although if they are not balanced internally, the possible sonic effects of the extra stage required to perform single-ended to balanced conversion may present a tradeoff.
All of those factors, of course, can be outweighed by many other variables in any given system, so I agree with the several people who have suggested experimentation and going with what sounds best. For initial experimental purposes, at least, you might want to consider purchasing a pair of inexpensive pro-oriented Mogami xlr cables.
Best regards,
-- Al