UM0585

16-bit FIR filter

Doc ID 14988 Rev 2

13/25

4

16-bit FIR filter

4.1

Description

The finite impulse filter (FIR) is a digital filter that is linearly dependent on a fixed finite

number of input samples.

The FIR filter can be defined by the number of coefficients to be processed (also called

taps), which gives the number of MAC (multiply-accumulate) operations to be done.

Figure 2.

Block diagram of an FIR filter of length N

The FIR filter of the DSP library is a direct-form real FIR filter that uses an array of M 16-bit

coefficients to filter N 16-bit samples.

Let us put:

●

a, the output vector of length N

●

c, the coefficients vector of length M

●

x, the input vector

So, x must have a length of M + N – 1.

4.2

DSP library function

The FIR function of the DSP library is an optimized assembly function that takes into

account the load-store architecture of the Cortex™-M3. Therefore, and as an optimization

constraint, the number of taps and the number of output samples must be a multiple of 4.

c

0

c

N-1

c

2

c

1

+

+

+

+

ai15448