The sine cardinal function, abbreviated sinc function, denoted image002z, is defined as

 

image004z.

 

A variation of the sinc function that is usually used in signal processing and used in this work for producing one kind of FIR filters, termed the normalized sinc function, has the form

 

image006z.

 

The design of a Sync FIR filter takes place in the time domain. Three parameters are used for this purpose: (1) the filter type (i.e. low pass, band pass, etc.), (2) the cutoff frequency, or frequencies, and (3) the filters length. The filter's impulse response, image008z, is then given by the following set of equations:

 

(1)Let image010z, where image012z is the cutoff frequency, measured at the ½ amplitude level, and specified in normalized form (a value between 0 and 0.5).

 

(2)Let image014z be the length of the filter.
noteimage014z will affect the transition bandwidth, image016z, according to the approximation image018z, where image016z also depends on whether the filter is windowed or not, and if so, on the window type that was used.

 

(3)Let image020z, then,

 

(4)Low Pass Filter: image022z. If  image014z is odd, we use image024z at image026z

 

(5)High Pass Filter: image028z. If image014z is odd, we use image030z at image026z

 

(6)Band Pass Filter: image033z. If image014z is odd, we use image035z at image026z

 

(7)Band Stop Filter: image038z. If image014z is odd, we use image040z at image026z

 

Where image043z and image045z are the low and high cutoff frequencies, respectively. Finally, a window function may be applied to the resulting filter coefficients in order to improve the filter's impulse response, since we have truncated the normalized sinc function to a finite length, when in reality it extends from image047z to image049z in representing an ideal filter kernel.

 

sinc-function