Ideally a headphone should quickly and smoothly dissipate energy following an input transient. Little insight into this can be gained by looking at its impulse response (IR) on a linear amplitude scale but with a logarithmic (decibel) amplitude scale we can see its behaviour more clearly. Because the external ear has resonances associated with it, which are duplicated in the artificial ear used for the measurement, the decay cannot be instantaneous. But in the best headphones – usually open-back designs – the impulse response decays to -40dB in less than 2 milliseconds (ms) and to -50dB in less than 3ms.
The log IR is not derived from a single measurement but from all 10 frequency response measurements performed on each capsule (with the headphone being removed from and reseated on the artificial ear between each measurement). It is created by converting the 10 IRs to a decibel amplitude scale, normalising the peak amplitude to 0dB, and for each time step choosing the highest amplitude of the 10 to create a worst-case composite. This is overlaid on the calculated impulse response (green trace) of a perfect headphone drive unit having a maximally flat frequency response between second-order high-pass and low-pass roll-offs at 20Hz and 20kHz, which is shown as a benchmark. Further insight into the time domain behaviour of the headphone is provided by the CSD waterfall.
Test signal: pink-spectrum periodic noise
Sampling frequency: 96kHz
Bit-depth: 16 bits
Time resolution (measurement): 10.42us
Time resolution (graph): 10.42us
Reseats on the artificial ear: 10 per capsule
Measurement averages: 10 per re-seat
Trace smoothing: none