FIR滤波器的实现方法(转)

源：http://blog.sina.com.cn/s/blog_493520900102uy26.html

内容来自于上篇博文，第七章，FIR滤波器

http://blog.sina.com.cn/s/blog_493520900102uxzu.html

 

使用 FDATool工具，构建一个Kaiser窗低通滤波器，采样频率8kHz，生成滤波器系数：

 

1）例程1

 

文件 fir_fltcoeff.h，以浮点形式表示的量化后的滤波器系数

//define the   FIR  filter  length

#define   N_FIR  40

 

float   h[N_FIR]  =  { ‐ 0.004314029307,‐ 0.013091321622,‐0.016515087727,

                  ‐0.006430584433,  0.009817876267, 0.010801880238,

                  ‐0.006567413713,‐ 0.016804829623, 0.000653253913,

                   0.022471280087,  0.010147131468,‐0.025657740989,

                  ‐0.026558960619,  0.023048392854, 0.050385290390,

                  ‐0.009291203588,‐ 0.087918503442,‐0.033770330014,

                   0.187334796517,  0.401505729848, 0.401505729848,

                   0.187334796517,‐ 0.033770330014,‐0.087918503442,

                  ‐0.009291203588,  0.050385290390, 0.023048392854,

                  ‐0.026558960619,‐ 0.025657740989, 0.010147131468,

                   0.022471280087,  0.000653253913,‐0.016804829623,

                  ‐0.006567413713,  0.010801880238, 0.009817876267,

                  ‐0.006430584433,‐ 0.016515087727,‐0.013091321622,

                  ‐0.004314029307   };

 

实际电路用AIC3106，每次采样中断来临后读入一组数据，分别为左右声道的数据，各16位。中断服务程序将左声道的每个数据送入滤波器，然后从DAC电路输出滤波后的结果。右声道数据不变。

 

文件 ISRs_Plus.c ，中断服务程序：

// Welch, Wright, & Morrow, 

// Real-time Digital Signal Processing, 2011

// Modified by Mark Wickert February 2012 to include GPIO ISR start/stop postings

 

///////////////////////////////////////////////////////////////////////

// Filename: ISRs_fir_float.c

//

// Synopsis: Interrupt service routine for codec data transmit/receive

//           floating point FIR filtering with coefficients in *.h file

//

///////////////////////////////////////////////////////////////////////

 

#include "DSP_Config.h"

#include "fir_fltcoeff.h"   //coefficients in float format

 

// Function Prototypes

long int rand_int(void);

  

// Data is received as 2 16-bit words (left/right) packed into one

// 32-bit word.  The union allows the data to be accessed as a single 

// entity when transferring to and from the serial port, but still be 

// able to manipulate the left and right channels independently.

 

#define LEFT  0

#define RIGHT 1

 

volatile union {

Uint32 UINT;

Int16 Channel[2];

} CodecDataIn, CodecDataOut;

 

 

float x_buffer[N_FIR];       //buffer for delay samples

 

 

interrupt void Codec_ISR()

///////////////////////////////////////////////////////////////////////

// Purpose:   Codec interface interrupt service routine  

//

// Input:     None

//

// Returns:   Nothing

//

// Calls:     CheckForOverrun, ReadCodecData, WriteCodecData

//

// Notes:     None

///////////////////////////////////////////////////////////////////////

{                    

WriteDigitalOutputs(1); // Write to GPIO J15, pin 6; begin ISR timing pulse

int i;

float result = 0; //initialize the accumulator

 

 

  if(CheckForOverrun()) // overrun error occurred (i.e. halted DSP)

return; // so serial port is reset to recover

 

  CodecDataIn.UINT = ReadCodecData(); // get input data samples

 

//Work with Left ADC sample

//x_buffer[0] = 0.25 * CodecDataIn.Channel[ LEFT];

//Use the next line to noise test the filter

x_buffer[0] = 0.125*((short) rand_int());//scale input by 1/8

 

//Filtering using a 32-bit accumulator

for(i=0; i< N_FIR; i++)

{

result += x_buffer[i] * h[i];

}

//Update filter history

for(i=N_FIR-1; i>0; i--)

{

x_buffer[i] = x_buffer[i-1];

}

 

//Return 16-bit sample to DAC

CodecDataOut.Channel[ LEFT] = (short) result;

// Copy Right input directly to Right output with no filtering

CodecDataOut.Channel[RIGHT] = CodecDataIn.Channel[ RIGHT];

WriteCodecData(CodecDataOut.UINT); // send output data to  port

WriteDigitalOutputs(0); // Write to GPIO J15, pin 6; end ISR timing pulse

}

 

//White noise generator for filter noise testing

long int rand_int(void)

{

static long int a = 100001;

 

a = (a*125) % 2796203;

return a;

}

 

可以看出，中断服务程序是逐点响应的，每次来一个数据，都要做一次滤波器运算，并将原有的数据前移一位，将新数据追加在缓冲区末尾。

float x_buffer[N_FIR]; // 用于保存现在的新值和过去的数值，长度与滤波器系数相同

x_buffer[0] // 用于保存当前值

x_buffer[1] // 保存过去前一点的值，以此类推

h[0...39] // 滤波器系数

result += x_buffer[i] * h[i]; // 当前点滤波器输出结果，循环从0到39