A way to minimize the effect of aliasing is filtering the analog signal by means of a low-pass filter (see PAK Academy: Anti-Aliasing Filters). Such an analog filter is built out of electrical components, such as resistors and capacitors. In order to attenuate the frequency range above the Nyquist frequency sufficient enough, the slope after the cut-off frequency should be steep which means high order filters are required. This can only be realized with electrical components having very tight specifications, and are therefore very expensive.
Moreover, as we are talking about electrical components with very tight specifications, the properties of the filter become temperature dependent due to slight variations in, for example, the resistance of the resistor. So in the past, users had to wait before their data acquisition systems had come to operating temperature to be able to use them.
Figure 1. shows the filter characteristics as used in the old days. The filter kicks in at about half of the sampling frequency. Aliasing effects caused by frequency content close to the sampling rate are minimized, but only to a certain extent.
Nowadays, these problems are circumvented by using a so called ‘Multi-Sampling Rate’ technique. Instead of sampling at slightly more than double the frequency range of interest (order of magnitude kHz), the analog signal is sampled much higher (order of magnitude MHz or GHz).
As a result, only frequency content close to the high sampling rate will mirror back into frequency content of interest. This allows the slope of the low-pass filter to be less steep (lower order filter) and therefore less electrical components with lower specifications are required.