audio aero capitol reference - XLR or RCA ?

Hi Masilu
The factory rep information is correct. The Capitole's single-ended outputs are converted to a balanced signal by going through a Burr Brown special line driver that splits the signal into positive phase and negative phase (pins 2 an 3 of the XLR connector).

I suggest that you use the RCA outputs of the Capitole, unless your cables are very long.
In your system, because your amp has RCA inputs, the RCA to RCA connection is the shorter signal path.
A XLR to RCA adapter will certainly degrade the sound quality.

Those with amps that have a direct XLR input will get better sound from the Capitole's XLR output, again because it is a shorter signal path.

Bob Reynolds, you are correct.
Common mode noise reduction (CMRR) will only happen if the source AND the receiving end are balanced. Adapters at any end of the cable do not bring any of the balanced IC benefits and degrade sound quality by means of an extra connection.
Balanced input transformers at the amp or preamp do provide CMRR, if the player's output is balanced.

Since the Capitole uses a custom-designed balanced line driver, its balanced output will drive longer ICs and get the benefit of common noise rejection if connected to a REAL balanced input. By real I mean two symmetrical, identical, opposing polarity input circuits, not a XLR socket with connections only to pins 1 and 2 or with an adapter.

In my previous post, I tried to simplify the issue by writing shorter signal path. I meant apples should connect to apples and oranges should be connected to oranges. Your post covers more ground and is technically more correct.

Zenieth, for the purposes of interconnection CMRR, truly balanced means two symmetrical, identical, opposing polarity OUTPUT AND INPUT circuits, not balanced DACs, which may or may not provide (depending on board layout & power supplies) an improvement in DAC internal S/N ratio.

There is too much confusion and WAY too much adoration of balanced connections in the high-end community.
In my humble opinion, end-to-end balanced circuits cost double and provide about 10% better bass and much less natural midrange timbre/soundstaging. The reason is quite clear once one realizes that transistors can be matched for gain (Beta) at ONE specific volume level, not at CD's 64,000 different discrete levels. Resistors can be matched at 0.5% and so on.
Best case, a well-implemented balanced circuit will have an average dynamic (not bench-measured, static) imbalance of 1% or more between the opposing phases.
AFAIK, 1% imbalance in a push-pull circuit translates to 1% harmonic distortion at the output.
Please correct me if I am wrong.