Synthesis of anti-aliasing filters for IF receivers

Digital IF receivers face a trade-off between the required selectivity of the frequency response and the value of the IF frequency: higher IF frequencies make it easier to reject the image frequency of the mixer, but increase the required Q values in the anti-aliasing filter. However, high-Q filters are hard to implement in IC technologies. The conventional approach to filter design is to convert the band-pass filter synthesis problem into a low-pass filter synthesis problem, and apply transformations to obtain the frequency response of the desired band-pass filter, which will have as many zeros in the origin as complex poles. However, this process may result in very high values of Q: we thus designed the filter directly in band-pass form. We show that alternating low-pass and band-pass filters, thus reducing the number of zeros in the origin, enables better aliasing rejection, achieving 4-5dB additional attenuation. This implies that the conventional approach of cascading low-order band-pass filters is not optimal, as it results in lower attenuation of the aliasing bands, given constraints on the number of stages and maximum Q values.