作者：桂。

時間：2018-02-06??12:10:14

鏈接：http://www.cnblogs.com/xingshansi/p/8421001.html?

---

前言

本文主要記錄基本的FIR實現，以及相關的知識點。

?一、基本型實現

首先從最基本的FIR入手：

?對應module：

`default_nettype none // module smplfir(i_clk, i_ce, i_val, o_val);parameter IW=15;localparam OW=IW+1;input wire i_clk, i_ce;input wire [(IW-1):0] i_val;output reg [(OW-1):0] o_val;reg [(IW-1):0] delayed;initial delayed = 0;always @(posedge i_clk)if (i_ce)delayed <= i_val;always @(posedge i_clk)if (i_ce)o_val <= i_val + delayed;endmodule

　　

二、通用版FIR

　　前文里最多涉及階數為5的FIR，這里給出適用任意階、給定位寬的FIR。

　　A-參數轉化

vivado仿真用到浮點->定點，需要將給定數據轉為定點補碼、或通過補碼讀取數據。

1）浮點轉定點補碼：

clc;clear all;close all; %=============產生輸入信號==============% N=12; %數據位寬 load fir128.mat; y_n = fir128; y_n=round(y_n*(2^(N-3)-1)); %N比特量化;如果有n個信號相加，則設置（N-n） %=============設置系統參數==============% L=length(y_n); %數據長度 %=================畫圖==================% stem(1:L,y_n); %=============寫入外部文件==============% fid=fopen('win.txt','w'); %把數據寫入sin_data.txt文件中，如果沒有就創建該文件 for k=1:length(y_n)B_s=dec2bin(y_n(k)+((y_n(k))<0)*2^N,N);for j=1:Nif B_s(j)=='1'tb=1;elsetb=0;endfprintf(fid,'%d',tb);endfprintf(fid,'\r\n'); endfprintf(fid,';'); fclose(fid);

　　原型濾波器fir128為128階的FIR濾波器。

生成的txt調用：$readmemb("*.txt",data);

2）給定補碼，讀取原數據：

clc;clear all;close all; filename = 'win.txt'; fid = fopen(filename); data_cell = textscan(fid,'%s','HeaderLines',0); data = data_cell{1,1};Nbit = 12;%number of bits len = length(data)-1;%length of filter wins = zeros(1,len); for i = 1:lenstr_win = data{i};if (str_win(1) == '0')wins(i) = bin2dec(str_win(2:end));endif (str_win(1) == '1')wins(i) = -bin2dec(num2str(ones(1,Nbit-1)))+bin2dec(str_win(2:end));end end wvtool(wins)

　　得到濾波器特性如下圖所示，當然也可以hex2dec轉為16進制，思路一致。

　　B-仿真模型

?testbench:

`timescale 1ns / 1ps module tb;// Inputsreg Clk;reg rst;// Outputsparameter datawidth = 12;wire signed [2*datawidth-1:0] Yout;//Generate a clock with 10 ns clock period. initial Clk <= 0;always #5 Clk = ~Clk;//Initialize and apply the inputs. //-------------------------------------// parameter data_num = 32'd1024; integer i = 0; reg [datawidth-1:0] Xin[1:data_num]; reg [datawidth-1:0] data_out;initial beginrst = 1; #20rst = 0;$readmemb("D:/PRJ/vivado/simulation_ding/009_lpf6tap/matlab/sin_data.txt",Xin); endalways @(posedge Clk) beginif(rst)begindata_out <= 0;endelse begindata_out <= Xin[i];i <= i + 8'd1;end end fastfir firinst( .i_clk(Clk), .i_reset(rst), .i_ce(1'b1), .i_sample(data_out), .o_result(Yout) ); endmodule

　　fast.v:

// `default_nettype none // module fastfir(i_clk, i_reset, i_ce, i_sample, o_result);parameter NTAPS=127, IW=12, TW=IW, OW=2*IW+7;input wire i_clk, i_reset;//input wire i_ce;input wire [(IW-1):0] i_sample;output wire signed [(2*IW-1):0] o_result;reg [(TW-1):0] tap [0:NTAPS];wire [(TW-1):0] tapout [NTAPS:0];wire [(IW-1):0] sample [NTAPS:0];wire [(OW-1):0] result [NTAPS:0];wire tap_wr;// The first sample in our sample chain is the sample we are givenassign sample[0] = i_sample;// Initialize the partial summing accumulator with zeroassign result[0] = 0;//observe filterreg [IW-1:0] fir_coef;integer i = 0;always @(posedge i_clk)beginif(i_reset) fir_coef <= 0;elsebeginfir_coef <= tap[i];i <= i+ 8'd1;end endgenvar k;generatebegininitial $readmemb("D:/PRJ/vivado/simulation_ding/009_lpf6tap/matlab/win.txt", tap);assign tap_wr = 1'b1;endfor(k=0; k<NTAPS; k=k+1)begin: FILTERfirtap #(.FIXED_TAPS(1'b1),.IW(IW), .OW(OW), .TW(TW),.INITIAL_VALUE(0))tapk(.i_clk(i_clk), .i_reset(i_reset), .i_tap_wr(tap_wr), .i_tap( tap[k]), .o_tap(tapout[k+1]),.i_ce(i_ce), .i_sample(sample[0]), .o_sample(sample[k+1]),.i_partial_acc(result[k]), .o_acc( result[k+1]));end endgenerateassign o_result = result[NTAPS][2*IW-1:0];endmodule

　　firtap.v:

// `default_nettype none // module firtap(i_clk, i_reset, i_tap_wr, i_tap, o_tap,i_ce, i_sample, o_sample,i_partial_acc, o_acc);parameter IW=12, TW=IW, OW=IW+TW+8;parameter [0:0] FIXED_TAPS=1;parameter [(TW-1):0] INITIAL_VALUE=0;//input wire i_clk, i_reset;//input wire i_tap_wr;input wire [(TW-1):0] i_tap;output wire signed [(TW-1):0] o_tap;//input wire i_ce;input wire signed [(IW-1):0] i_sample;output reg [(IW-1):0] o_sample;//input wire [(OW-1):0] i_partial_acc;output reg [(OW-1):0] o_acc;//reg [(IW-1):0] delayed_sample;reg signed [(TW+IW-1):0] product;// Determine the tap we are usinggenerateif (FIXED_TAPS != 0)// If our taps are fixed, the tap is given by the i_tap// external input. This allows the parent module to be// able to use readmemh to set all of the taps in a filterassign o_tap = i_tap;else begin// If the taps are adjustable, then use the i_tap_wr signal// to know when to adjust the tap. In this case, taps are// strung together through the filter structure--our output// tap becomes the input tap of the next tap module, and// i_tap_wr causes all of them to shift forward by one.reg [(TW-1):0] tap;initial tap = INITIAL_VALUE;always @(posedge i_clk)if (i_tap_wr)tap <= i_tap;assign o_tap = tap;end endgenerate// Forward the sample on down the line, to be the input sample for the// next componentalways @(posedge i_clk)if (i_reset)begindelayed_sample <= 0;o_sample <= 0;end else if (i_ce)begin// Note the two sample delay in this forwarding// structure. This aligns the inputs up so that the// accumulator structure (below) works.delayed_sample <= i_sample;o_sample <= delayed_sample;end// Multiply the filter tap by the incoming samplealways @(posedge i_clk)if (i_reset)product <= 0;else if (i_ce)product <= o_tap * i_sample;// Continue summing together the output components of the FIR filteralways @(posedge i_clk)if (i_reset)o_acc <= 0;else if (i_ce)o_acc <= i_partial_acc+ { {(OW-(TW+IW)){product[(TW+IW-1)]}},product };// Make verilator happy// verilate lint_on UNUSEDwire unused;assign unused = i_tap_wr;// verilate lint_off UNUSED endmodule

?仿真結果：

轉載于:https://www.cnblogs.com/xingshansi/p/8421001.html

總結

以上是生活随笔為你收集整理的FIR仿真module_04的全部內容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網站內容還不錯，歡迎將生活随笔推薦給好友。