Sounds to me like you may be bottoming-out the voice coil. A loud transient can push the speaker out pretty far and the recoil can cause the piston to hit the motor assembly.
It would be worthwhile to verify that the speaker polarity is in phase with the amp. In other words that the positive output from the amp is connected to the positive lead on the speaker. You can watch the cone to make sure that it is pushing air on the initial attack of a note. If it seems to be moving inward on the initial attack you are more likely to bottom out on transient peaks and damage the voice coil.
It would be worthwhile to verify that the speaker polarity is in phase with the amp. In other words that the positive output from the amp is connected to the positive lead on the speaker. You can watch the cone to make sure that it is pushing air on the initial attack of a note. If it seems to be moving inward on the initial attack you are more likely to bottom out on transient peaks and damage the voice coil.