Did anyone official form Chord Electronics wrote something regarding this roll-off on lower volume with low impedance headphones?
The roll-off shown on the graph is fairly over-emphasised, and as Rob's posts indicate, there are some other factors involved in making the Hugo and Mojo sound different.
Only thing that the graph was meant to show there was simply that the Mojo can behave differently to other typical opamp/chipamp based headphone amps.
The most important thing to worry about is that the "impedance" of your earphones are not always constant.
e.g. Almost all BA driver earphones have impedance that increases at high frequency, as you can see from innerfidelity measurements etc.
http://www.innerfidelity.com/content/25-years-making-good-thing-better-etymotic-er4sr-and-er4xr-measurements
So even if the headphones specifications say "16 ohms", that doesn't mean it is constantly 16 ohms across the entire frequency range. (planar magnetic has less of this problem).
In any case, one of the important reasons why professional studio monitors tend to have high impedance (e.g. 600 ohms) is so that it can avoid these amplifier dependent issues.
It's not so much about whether a certain amp sounds well with low impedance or high impedance headphones, but it's simply that low impedance headphones tend to suffer from more sonic coloration from the amplifier performance. In that regard, Mojo is designed exceptionally well to handle both high and low impedance headphones with ease.
e.g. most of my other portable amps either can't drive difficult inefficient loads (e.g. Audeze LCD) or have too much white noise on low impedance high efficiency IEMs (e.g. Andromeda), but Mojo works very well for both cases, particularly for how small and affordable it is.
