Digital filters are for the most part identical to their analogue counterparts on which they are based. However one of the way in which they can differ is when they are carried over to the digital domain by means of the ‘bilinear transform“‘. As long as the sampling rate exceeds the frequencies of interest by a comfortable margin, the bilinear transform provides a very accurate way to design a digital filter.
The problem with the bilinear transform only presents itself with certain filter types and only when the filter is applied at the top end where some of the cut or boost would pass beyond half the sampling frequency – the Nyquist frequency which represents the highest frequency which can be reproduced by the system at that sampling rate.
A consequence of using the bilinear transform is that the amplitude must be at unity at the Nyquist frequency. Because of this the symmetrical shape of bell filters applied at the high end will be distorted, squashing the slope of the bell above the centre frequency more and more as the centre frequency is raised.