Let's hope op don't get mad with us taking a longer detour from topic... :)
But as long as P=V^2/R is correct, it doesn't imply the potentiometer is turned down. In fact,
it's usually the opposite in most designs, since the pot is usually placed in the input and,
the greater its value, the bigger is the input (usually dac output) voltage applied to the pre-amp
or 1st stage of the amp. This equation applies to the power available to the phones,
and its equivalent resistance.
About the headphone break event: if the knob was at 12 o'clock, for example. This defines,
regardless the headphone, the maximum power the amp will deliver. However, if the music
is silent at certain point, the power delivered to the phone is nearly zero, at that moment.
Following this tought, in the 1st chunk of the song, the instruments/melody were more
hi-freq content. Due to that music's content, there was less spectral power in this freq,
so less energy was delivered to the phone.
When the bass line entered, a much bigger energy packet was driven to the phone.
Remember that the pot is at the same position. Initially, the power was available, but not in use.
Now, let's extrapolate and think what if a hi-freq synthesized sinus, say 15kHz, was applied to the amp input.
With enough power, it could damage the phone as well, since the amp frequency response includes
this frequency range, them some.
To sum up, if enough energy with right spectral content (within the freq response of
the amp under analyisis) is applied to the amp input,
it will go through the amp, be amplified, and damage the out of spec load,
in this case a headphone.
Edited by xezi - 2/17/13 at 4:55pm