Now let's see what this implies from the point of view of sound. As we have already seen in its pertaining section, a sinusoidal signal comprises one single frequency [Pure sinusoid ] equal to the number of cycles that the sinusoid itself undergoes per second. However if we consider a signal that has many sudden transitions, it will have additional frequencies. So, a signal that has sudden transitions in time, consists in not just one, but a series of frequencies. The more sudden the transitions are, the higher the frequencies needed to reproduce them. If we were to stretch this last point we would end up with a rectangular wave [Square wave ] which has instantaneous transitions and therefore needs infinite ever-increasing frequencies in order to be reproduced (this is a theoretical abstraction which doesn't exist in real life since amplitude transitions can never occur in a 0 sec time interval).
So we see that by cutting off the sinusoid's peak, the amplifier imposes some transitions onto the signal that are not contained in the original one. This generates new frequencies which also weren't present in the original signal, and this is what causes the distortion. So, unless we wish to use distortion as an effect [Distorter ] , input voltage must always remain within the amplifier's limits stated in the data sheet.