This slide shows the effects of aliasing in the frequency domain. The top illustration shows a signal fa being sampled at a rate which is significantly higher than 2fa. The "I" components are the images, or aliases, of fa which occur on either side of each multiple of the sampling frequency. It is convenient to divide the frequency spectrum into "Nyquist Zones", each having a width of fs/2 as shown. The bottom illustration shows the condition in the previous slide, where fa is slightly less than fs. The aliased component falls at a frequency equal to fs – fa, exactly as in the previous time-domain slide. Graph B also illustrates the important fact that in an ideal sampler, with no filtering, any unwanted signal (including noise) outside the 1st Nyquist zone will appear in the 1st Nyquist zone due to aliasing. If aliasing is to be prevented, the ADC input must have an antialiasing filter. This filter is generally a lowpass filter when the signals of interest are located in the first Nyquist zone, but can be a bandpass filter if signals are in other Nyquist zones.