I think from your measurements, Planars do not alter the THD+N measurements as dramatically as Dynamic driver headphones.
So its probably a good idea to do such measurements with dynamic driver headphones ?
Just to check though, were the headphone cups covered ? As far as I understand, back-pressure does affect coil load on dynamic drivers, so they should vary with an open ended cup or a covered (like worn on head) test protocol.
This does raise the question of whether having static loads on the amps actually depicts performance that's "real world.". So, though slightly off topic, is still relevant for this thread.
Just a suggestion for
@amirm , perhaps if possible, a THD+N test for headphone amps using dynamic driver headphones that run 32 Ohm, 300Ohm, 600 Ohm, either at 1V (if the headphone is insensitive enough) or at output set to correspond to 90~100 dB SPL, being the upper limits of what's user applicable levels.