Makes sense.Fyi, I did the measurements yesterday and can confirm that you are correct, that the preamp was responsible. While they both (preamp and the E50 dac) output 2X the voltage from their balanced outputs, the final output from the preamp's XLRs remained practical the same whether I selected the unbalanced or balanced inputs to the preamp or not.
There is a very slight (decimal points) difference, measurable but not audible (to me anyway). Cambridge Audio's literature said the preamp was "fully differential" but I think you are right, the internal architecture must have something to do with making sue the XLR output remain at the same level even if the unbalanced inputs (connected to the E50's unbalanced outputs) are selected.
In reading the manual, I know realize the preamp does allow gain trim adjustments for each channel so I could have just use that feature to level match the two amps I am AB'ing without using the dual outputs (RCA and XLR) and variable volume. Not going to waste time now though as I have everything set up now to have the two level matched.
View attachment 219918
It would be interesting is to see if you can perceive a noise-level difference between XLR and RCA. For example, set the preamp at maximum volume level (ensuring to not play anything to preserve your hearing) and try to hear the noise floor. And if it is audible*, if there is an audible difference between XLR and RCA. My understanding is that XLR inherently provides 6dB better signal-to-noise ratio (and hence 6dB lower noise floor at a given matched volume level). That’s an additional advantage with respect to more EMI-robust interconnects…
In my system I can hear some very faint noise if I put my ear right at the twitter. As I use the E50 as (digital) preamp, I can confirm that the noise level is perceptively the same at any volume settings of the DAC.
* some preamp and DAV have an auto-mute function, which invalidate such tests.