DSP_学习报告_定点FIR滤波器的仿真与实现.doc

《DSP_学习报告_定点FIR滤波器的仿真与实现.doc》由会员分享，可在线阅读，更多相关《DSP_学习报告_定点FIR滤波器的仿真与实现.doc（9页珍藏版）》请在淘文阁 - 分享文档赚钱的网站上搜索。

1、DSP学习总结陈松 4在这门课当中，管老师不仅系统的简述了DSP的整体框架，也深入浅出的介绍了其他微处理器例如单片机和RAM的框架，以及与DSP的联系与各自的应用前景，当然不止这些。总之，学习了这门课让我受益匪浅。而DSP在编程中有很多关键的知识点，如下：1.变量的作用域以及生存期.2.CMD文件编写，要求熟悉DSP的存储器结构。3.用定点型的DSP将浮点数转化成定点数计算，即Q格式等的应用。4.matlab的运用。为了串联着知识点，我在这里做一个从MATLAB仿真到CCS验证的定点FIR滤波器设计来完成，并将这分报告作为我学习本门课程的总结。做得不好的地方还希望老师指出。一、FIR数字滤波器

2、的基础理论滤波器就是在时间域或频域内，对已知激励产生规定响应的网络，使其能够从信号中提取有用的信号，抑制并衰减不需要的信号。滤波器的设计实质上就是对提出的要求给出相应的性能指标，再通过计算，使物理可实现的实际滤波器频率响应特性逼近给出的频率响应特性。FIR 数字滤波器系统的传递函数为： （1）由此得到系统的差分方程： （2）若FIR 数字滤波器的单位冲激响应序列为h(n)，它就是滤波器系数向量b(n)。传统的滤波器分析与设计均使用繁琐的公式计算，改变参数后需要重新运算，从而在分析与设计滤波器尤其是高阶滤波器时工作量特别大。这里应用MATLAB 设计FIR滤波器，根据给定的性能指标设计一个H(z

3、)，使其逼近这一指标，进而计算并确定滤波器的系数b(n)，再将所设计滤波器的幅频响应、相频响应曲线作为输出，与设计要求进行比较，对设计的滤波器进行优化。设计完成之后将得到FIR滤波器的单位冲激响应序列h(n)的各个参数值。二、MATLAB软件仿真1. 利用matlab生成FIR低通滤波器的系数，并绘制器滤波器性能图。FIR滤波器性能：截止频率：50HZ通带：50100HZFIR滤波器代码：fs = 1000; %采样频率f = 50 100; %通带a = 1 0;dev = 0.05 0.01;N,Wn,beta,ftype = kaiserord(f,a,dev,fs);w = kaise

4、r(N+1,beta);%wvtool(w)h = fir1(N,Wn,w); %滤波器系数wvtool(h);ha,w = freqz(h,1024);plot(w/pi*400,abs(ha);title(幅度谱); %添加标题figure;plot(w,unwrap(angle(ha);title(相位谱); %添加标题参数的FIR滤波器系数：h=-0.0019,-0.0018,-0.0008,0.0010,0.0036,0.0062,0.0081,0.0082,0.0060,0.0011,-0.0061,-0.0142,-0.0212,-0.0249,-0.0230,-0.0137,0

5、.0034,0.0276,0.0566,0.0871,0.1149,0.1362,0.1477,0.1477,0.1362,0.1149,0.0871,0.0566,0.0276,0.0034,-0.0137,-0.0230,-0.0249,-0.0212,-0.0142,-0.0061,0.0011,0.0060,0.0082,0.0081,0.0062,0.0036,0.0010,-0.0008,-0.0018,-0.0019;FIR滤波器性能图：2. 在matlab上产生带有噪声的信号，经FIR滤波器滤波，对比滤波前后效果。验证信号频率：20HZ验证信号噪声：200HZ验证FIR滤波器效

6、果代码：clc;clear;% SIN信号产生fa=20; %信号频率fz=200; %杂波频率dt=0.001; t=0:dt:1; %采样频率fs=1/dtn1=length(t);x1=zeros(1,n1);x1=10000*(sin(2*pi*fa*t)+sin(2*pi*fz*t);plot(t,x1);title(滤波前信号); % FIR滤波%h滤波器系数，截止频率50HZ，通带50100HZh=-0.0019,-0.0018,-0.0008,0.0010,0.0036,0.0062,0.0081,0.0082,0.0060,0.0011,-0.0061,-0.0142,-0.

7、0212,-0.0249,-0.0230,-0.0137,0.0034,0.0276,0.0566,0.0871,0.1149,0.1362,0.1477,0.1477,0.1362,0.1149,0.0871,0.0566,0.0276,0.0034,-0.0137,-0.0230,-0.0249,-0.0212,-0.0142,-0.0061,0.0011,0.0060,0.0082,0.0081,0.0062,0.0036,0.0010,-0.0008,-0.0018,-0.0019;n2=length(h); x2=zeros(1,n2);x=zeros(1,n2+n1);x=x2,x

8、1; %在信号前面插入n个0，便于计算y=zeros(1,n1);%定义输出信号数组 for m=0:1000-n2 for k=1:n2; y(m+1)=h(k)*x(m+k)+y(m+1);%滤波 endend figure(2);plot(t,y);title(滤波后信号);MATLAB中验证滤波效果图：三、DSP硬件实测1. 将MATLAB中的信号数据生成CCS可以识别的.DAT文件。生成.DAT文件代码：%将数组ime1：n*1,十进制整数，如果不是整数就要扩大，乘以32768%中的数据存储为TI dat文件固定格式 最终数据也位16进制（8位）%sfm1 =int32(x); %需

9、要导入sfm数据 round:取整，四色五入sfm1(find(sfm10)=sfm1(find(sfm10)+65536; %对小于0的数求补 215=32768sfm2 = dec2hex(sfm1,4); %10进制转换成16进制 %sfm2(find(sfm40)=sfm2(find(sfm40)+; %对小于0的数求补 sfm3 = sfm2; %做转置 leng=length(sfm2);fid=fopen(re3.dat,wt);fprintf(fid,1651 1 1 800n); %在文件头写上：1651 1 1 1000。1651：固定格式；：代表地址（0x）；800：表示

10、数据深度，0x800=1024（32位数据）for i=1:8:leng*4 fprintf(fid,0x); fprintf(fid,%c,sfm3(i+4); %排列顺序：4567 0123 是为了让CCS中显示时是是顺序的，因为导入CCS中时低16位先导入 fprintf(fid,%c,sfm3(i+5); fprintf(fid,%c,sfm3(i+6); fprintf(fid,%c,sfm3(i+7); fprintf(fid,%c,sfm3(i); fprintf(fid,%c,sfm3(i+1); fprintf(fid,%c,sfm3(i+2); fprintf(fid,%c

11、,sfm3(i+3); fprintf(fid,n);endfclose(fid);2. 将滤波器的浮点系数转化成定点数，在CCS上编写FIR滤波器代码。编译平台：CCS3.3软件仿真。内核：DM6437。FIR滤波器代码：#include void FIR();#define FIR_Order 46 /fir滤波器阶数#define FIR_Num 1001 /计算的点数/*float HFIR_Order=-0.0019,-0.0018,-0.0008,0.0010,0.0036,0.0062,0.0081,0.0082,0.0060,0.0011,-0.0061,-0.0142,-0.

12、0212,-0.0249,-0.0230,-0.0137,0.0034,0.0276,0.0566,0.0871,0.1149,0.1362,0.1477,0.1477,0.1362,0.1149,0.0871,0.0566,0.0276,0.0034,-0.0137,-0.0230,-0.0249,-0.0212,-0.0142,-0.0061,0.0011,0.0060,0.0082,0.0081,0.0062,0.0036,0.0010,-0.0008,-0.0018,-0.0019; /fir滤波器系数，浮点*/ short int HFIR_Order=-62,-59,-26,33,

13、118,203,265,269,197,36,-200,-465,-695,-816,-754,-449,111,904,1855,2854,3765,4463,4840,4840,4463,3765,2854,1855,904,111,-449,-754,-816,-695,-465,-200,36,197,269,265,203,118,33,-26,-59,-62; /fir滤波器系数，定点short int X46+1001; /输入信号，由matlab仿真的信号生成.dat文件，导入到CCS中。/long int Y1001; /输出信号int Y1001; /输出信号main() /加入输入信号数据FIR(); /加断点,观察滤波效果void FIR()unsigned int j;unsigned char i;for(j=0;jFIR_Num;j+)for (i=0;iFIR_Order;i+ )/Yj+=(float)(Xj+i)*Hi);Yj+=(int)(Xj+i*Hi);3. 导入.DAT文件，观察滤波前后效果。MATLAB中验证滤波效果图：