To return to the topic, I show some loopback measurements with a (cheap and rather old, but that is not relevant here) 56 Ω headphone driven from a 100 Ω source. Unfortunately, the drivers are not very reactive, so the difference is not as large as it could be with some other models - HD598 would be more interesting, for example - but it is there. Here is what happens to the frequency response: That is not too much, but it is definitely enough for a positive ABX test. Now the distortion, with the loopback shown in yellow, and the microphone input (at the same loudness) in red. Again, the effect is not major, but at the resonant frequencies (~100 Hz and ~350 Hz) it is within 15 dB of the acoustic distortion, or about 5.6 times less): With 56 Ω resistors instead, it would stay below -96 dB for most of the frequency range.