##
Melitta - Designer

The filter designer allows to compute scaled filter parameters for a range for digital filter families and types given the appropriate input parameters. The types are as follows:

- Low pass filter,
- High pass filter,
- Band pass filter,
- Band stop filter.

The filter designer offers means to design the filters in an interactive manner (a shell), as well as in a GUI-supported manner. The filter design itself can be carried out using several methods as follows.

###
Design by analogue reference filter

This design method uses analogue reference filters in order to generate digital filter parameter sets for the following filter families:

- Butterworth filter,
- Chebychev Type I filter,
- Chebychev Type II filter (inverse Chebychev Type I),
- Elliptic Cauer filter,
- Bessel filter.

The designed analogue filter transfer function are transformed into the z-transfer function by using:

- Impulse invariant transformation,
- Step invariant transformation,
- Bilinear transformation (with frequency prewarp option),
- Matched s/z method,
- FIR approximation.

###
Design in the discrete time domain

FIR filters only are considered in this design, either with design in the frequency domain or through windowing.

####
Approximation in the frequency domain

An ideal digital filter transfer function is approximated in the frequency domain by variation of the FIR filter parameters according to certain algorithms:

- Frequency sampling,
- Least Square Minimisation,
- Chebychev (RezMez),
- Maximally Flat.

####
Window method

A finite sample of the impulse response of the ideal filter (infinte Sinc(n) function) is used to compose a FIR filter. The finite sample is shaped using the following windowing functions:

- Rectangular,
- Triangular,
- Hamming,
- Hanning,
- Blackmann,
- Kaiser,
- Chebychev.

###
Filter decomposition

The transfer function of the digital filter can be decomposed into a product, or sum of first and/or second order stages. An option will define the ordering of the stages by pole location and/or gain.

###
Filter parameter implementation

The last step in the design is the computation of the scaled filter parameters of the respective IIR or FIR filter for a concrete digital filter with a given finite word length through scaling. The output can be written into C source or header files.

###
Filter presentation

It is possible to generate Pole/Zero plot, a Nyquist plot, a Bode plot as well as a step and impulse response for all filters. It is also possible to compare all plots mentioned before.

Back to Melitta

Back to homepage

Copyleft
2005,
Peter Wurmsdobler.