Complimentary ColdFire® Digital Signal Processing Library


Roll over image to zoom in


  • Software Architecture
    • The core component of the ColdFire® DSP Library is a group of DSP algorithms implemented in assembly for optimal computational performance. In order to make these assembly functions more user-friendly, custom data structures and initialization functions are included. As a result, the assembly functions are C-callable with a minimum number of arguments. The user must only initialize a data structure through the use of its associated initialization routine, requiring no intimate knowledge of the data structure implementation or the assembly code. With the assembly algorithms implemented as functions rather than macros, the instruction code is not replicated in memory even if called multiple times.
  • Supported Platforms
    • ColdFire ISA_A platforms with an on-board MAC are supported. The library was developed and tested using M52221DEMO hardware evaluation board and CodeWarrior® 6.4 Integrated Development Environment (IDE). EMAC platforms are supported as long as the assembler consistently uses the same accumulator (typically ACC0). This was tested with CodeWarrior 6.4 on MCF5227x.
  • IIR Filter Configurations
    • The ColdFire DSP Library includes a large set of IIR filter configurations that span a wide range of applications. These predefined configurations allow a user to quickly select a specific frequency response by making three simple decisions.
  • Filter Configurations
    • The following table identifies all filter configurations included in the library. Each combination of filter shape, order and cutoff constitutes a single filter configuration. Each configuration comes with four parameters – filter coefficients array, numerator scale factor, denominator scale factor, and filter order. The configurations below are all characterized as Butterworth IIR filters.
  • Hardware Validation
    • All filter definitions have been tested in hardware for the recommended input ranges. They have been validated against a floating-point model to have minimal fixed-point errors, in both the RMS and absolute maximum measures. Filter definitions that are highly susceptible to fixed-point errors are intentionally excluded from the library. Hence, very low or very high digital frequency cutoffs, especially for higher order filters, are excluded..

Supported Devices


Quick reference to our documentation types.

4 documents