|On this page…|
Typically used with multirate filters, multistage filters can bring efficiency to a particular filter implementation. Multistage filters are composed of several filters. These different parts of the mulitstage filter, called stages, are connected in a cascade or in parallel. However such a design can conserve resources in many cases. There are many different uses for a multistage filter. One of these is a filter requirement that includes a very narrow transition width. For example, you need to design a lowpass filter where the difference between the pass frequency and the stop frequency is .01 (normalized). For such a requirement it is possible to design a single filter, but it will be very long (containing many coefficients) and very costly (having many multiplications and additions per input sample). Thus, this single filter may be so costly and require so much memory, that it may be impractical to implement in certain applications where there are strict hardware requirements. In such cases, a multistage filter is a great solution. Another application of a multistage filter is for a mulitrate system, where there is a decimator or an interpolator with a large factor. In these cases, it is usually wise to break up the filter into several multirate stages, each comprising a multiple of the total decimation/interpolation factor.
As described in the previous section, within a multirate filter each interconnected filter is called a stage. While it is possible to design a multistage filter manually, it is also possible to perform automatic optimization of a multistage filter automatically. When designing a filter manually it can be difficult to guess how many stages would provide an optimal design, optimize each stage, and then optimize all the stages together. DSP System Toolbox™ software enables you to create a Specifications Object, and then design a filter using multistage as an option. The rest of the work is done automatically. Not only does DSP System Toolbox software determine the optimal number of stages, but it also optimizes the total filter solution.