Most balanced transmissions used in consumer audio equipment are pseudo-balanced transmissions using op amps. The circuit can be easily realized at low cost, but in truly harsh environments, pseudo-balanced transmission easily loses its function.
The author of the article is simply stating the legendary story of balanced transmission that is widespread among audiophiles, and does not appear to have much understanding of the true purpose of balanced transmission.
As allude by
@Toku, a key thing most people misunderstand (including me until very recently) about balanced audio is that it is simply two wires of equal impedance, so that the receiving end can reject any introduced common mode noise by subtracting the signal on one wire from the other wire. As such, it would still be a balanced output even if the device sent 2 Vrms on the positive wire and nothing (0 Vrms) on the negative wire — for a total of 2 Vrms signal from the balanced output, just like the 2 Vrms on the unbalanced output. And in such a case, the measured SINAD of the balanced and unbalanced output would be virtually identical.
By contrast, if my understanding is correct, professional studio equipment might have a much higher SINAD delta between the balanced output and the unbalanced output. Not because the output is balanced, but because the 15V power supplier of the professional device allows for output at +24 dBu (professional audio) nominal level on the balanced output, while limiting the total output to nominal line level appropriate for consumer audio gear on the unbalanced output. In which case, an output of 12.3 Vrms vs. 2 Vrms can allow for higher SINAD primarily
because it is a high signal level amplification against the fixed noise floor of the device, not because it is using balanced outputs. Of course, nearly all professional audio devices will will be splitting that voltage across the 2 impedance-matched wires using a differential amplifier... but NOT ALWAYS! A simplified pseudo-balanced output (i.e., designed to plug into either a balanced or unbalanced audio device on the receiving side) could just send the entire 12.3 Vrms down the positive wire and 0 Vrms on the impedance matched negative wire.
In the case of the E50, the SINAD of the two outputs are comparing 2 Vrms vs. 4 Vrms at 0 dBFS. As such, the balanced output is not that much higher of a signal level at full scale volume. Of note: the maximum line level on the balanced output from the E50 is limited by the 5V power supply. On the other hand, the cost of an E50 is substantially lower than professional audio gear that operates at +24 dBu with a 15V power supply.