I have some problems with several aspects of Tyll's THD+N graphs, the vertical axis is in % of THD+N, it seems correct, depending on the headphones involved, high levels of Harmonic distortion is expected for the bass, however how come we have peaks at 20/200/2000 Hz for every headphone. Other transducers for which THD has been measured with a frequency sweep also don't show such peaks. Also if we look at the peaks just after 200 Hz, they appear at the same frequencies regardless of the headphone, I don't explain this.
They could be measuring artifacts, induced by the boxed enclosure that Tyll uses, or by the measuring gear, but then it would make sense to eliminate them from the final graph to increase readability.
Alternatively, it looks like FFT of the headphone's response to a tree tone signal (20/200/2000 Hz), the problem is the other peaks don't correspond to harmonics, or even IMD. Also, it wouldn't make sense to have a different floor noise depending on the headphones. And finally, I trust that Tyll wouldn't have mislabel led the axis legends.
Could I have some precision?
Also, from the LCD2-3 graphs.
Both show that the THD+N figure drop from to 80 to 90 dB SPL, there's no reason for harmonic distortion to drop when increasing volume. It means two things: that THD didn't increase because of higher level (good power handling) and that the THD+N figure is dominated by its noise component which explains the drop as background noise is constant. The second means that the harmonic distortion component of the LCD2-3 is buried in the noise and cannot be picked up by the measuring gear. And as I said earlier, there's no reason for transducer to show little peaks in %THD+N at specific frequencies shared by all transducers, those peaks should be disregarded.
