A device is said to be time-coherent if it exhibits an essentially linear phase-shift over frequency, i.e., the characteristics of a pure delay. It is not necessary that the frequency response be flat, just that the phase-shift vary linearly with frequency. Apparently, humans can hear time differences in the 5µs-10µs range. A lack of time-coherency is usually problematic in multi-way speaker systems and some microphones. The crossover point in multi-way loudspeakers, without careful control and compensation, causes the speaker to output a multi-lobed pattern, i.e., the main energy output is not in front of the speaker, and the polar pattern is asymmetric. In a microphone, a perfect characteristic impulse response would be where the diaphragm would cease moving as soon as the sound ended. However, bodies in motion tend to remain in motion; in addition, all of the microphone resonates, not only the diaphragm. All of this additional motion contributes to the coloration of the sound by smearing the various arrival times of the various frequencies. By using a very small, lightweight diaphragm housed in a capsule that has a minimum of resonance and reflective surfaces, coloration is reduced.