当前位置:网站首页>FIR filter of IQ signal after AD phase discrimination
FIR filter of IQ signal after AD phase discrimination
2022-07-08 01:13:00 【wgm1996】
Recently, I encountered a small problem in the project , Current radar systems have multiple waveforms to choose from , The maximum bandwidth is 300M,AD The module is a module made by the board supplier , For convenience , The other party after the identification IQ The signal passes through a fixed 300M Bandwidth half band filter for low-pass filtering , Output baseband signal .
In practical use , The transmission waveform bandwidth is 300M when , It can achieve the best matched filtering , Have the best signal-to-noise ratio , The bandwidth of the transmitted waveform is less than 300M when , for example 50M, The filter is still pressed 300M Filtering , Because the probability density function of noise is a rectangle in the frequency domain , Other 250M The noise power in the spectrum will also pass through the filter , Cause the power of noise to increase , The signal-to-noise ratio is reduced .
In order to improve the signal-to-noise ratio , You can add another layer to the baseband signal 50M Low pass filter of , In my project , There are two ways to achieve , The first is in the baseband signal IQ Two channels are respectively connected with filter coefficients ( The set of real Numbers ) Convolution ; The second is right IQ The complex signal composed of FFT To the frequency domain , Dot multiply with the frequency domain of the filter coefficients , Then reverse FFT that will do .
In the second implementation method, special attention should be paid , Complex coefficient FIR Filter and real coefficient FIR Filters are slightly different , Suppose the real coefficient is a, Then the corresponding complex coefficient is a+ai, And then we can move on FFT To the frequency domain , You can't simply use real coefficients to transform to the frequency domain .
summary :
1、AD The post low-pass filter should be designed according to the waveform bandwidth , The low-pass filter in a system with multiple bandwidths must match the waveform , In my case of fixed filter, it can be solved by cascaded filter .
2、 Pay attention to real coefficient and complex coefficient FIR The difference between filters
边栏推荐
猜你喜欢
Semantic segmentation model base segmentation_ models_ Detailed introduction to pytorch
5.过拟合,dropout,正则化
1.线性回归
[deep learning] AI one click to change the sky
Ag9310meq ag9310mfq angle two USB type C to HDMI audio and video data conversion function chips parameter difference and design circuit reference
Basic implementation of pie chart
利用GPU训练网络模型
7.正则化应用
High quality USB sound card / audio chip sss1700 | sss1700 design 96 kHz 24 bit sampling rate USB headset microphone scheme | sss1700 Chinese design scheme explanation
Chapter VIII integrated learning
随机推荐
Introduction to paddle - using lenet to realize image classification method II in MNIST
Binder core API
完整的模型训练套路
Mathematical modeling -- knowledge map
Ag9310 same function alternative | cs5261 replaces ag9310type-c to HDMI single switch screen alternative | low BOM replaces ag9310 design
8.优化器
String usage in C #
Basic implementation of pie chart
[go record] start go language from scratch -- make an oscilloscope with go language (I) go language foundation
130. 被圍繞的區域
Chapter VIII integrated learning
New library launched | cnopendata China Time-honored enterprise directory
How to write mark down on vscode
High quality USB sound card / audio chip sss1700 | sss1700 design 96 kHz 24 bit sampling rate USB headset microphone scheme | sss1700 Chinese design scheme explanation
Chapter IV decision tree
完整的模型验证(测试,demo)套路
Complete model training routine
Design method and application of ag9311maq and ag9311mcq in USB type-C docking station or converter
新库上线 | CnOpenData中华老字号企业名录
A network composed of three convolution layers completes the image classification task of cifar10 data set