# 頻譜分析那些事 ## 1.FFT Basic SL-廣義 > 級數 > complex > transform ### 廣義 用正交函數表示f(x) ![](https://i.imgur.com/j5ZsKmi.png) Cn為f(x)與$\varphi$單位內積結果,可理解成當f(x)用$\varphi$(x)表示時,base $\varphi$的比重,再更簡易可以理解成單位轉換。 以下舉例f(x)=t, 0<t<1 ![](https://i.imgur.com/g9bQcXo.png) 注意f(x)為非奇非偶函數,只有在0~1之間有定義 稱此定義域為0<x<L, L=1 ### 級數解 接著將$\varphi$換成sin/cos,亦即將函數用sin/cos表示,注意sin/cos正交特性(想想偏光鏡)。 - #### sin series ![](https://i.imgur.com/udIEd4r.png) sin級數合成之f(x)為奇函數,週期為2L ![](https://i.imgur.com/txWIiqq.png) 此為以[$sin\pi x$ $sin2\pi x$ $sin3\pi x$ ....]為basis展開f(x)。 - #### cos series ![](https://i.imgur.com/jCHrzmu.png) cos級數合成之f(x)為偶函數,週期為2L - #### 全幅 series ![](https://i.imgur.com/yDwF8Xc.png) 使用sin及cos合成之f(x)為非奇非偶函數,週期為2L,可視為sin及cos級數相加合成。 - #### complex ![](https://i.imgur.com/mDJIqw9.png) 使用Euler equation,複係數合成之f(x)為全幅,非奇非偶函數,週期L=無窮,即為fourier transform。 注意此合成結果在0<x<L外亦有定義。 ![](https://i.imgur.com/EKM33lf.png) --- 假設輸入改成 $f(t)=0.7*sin(2*pi*50*t)$,此為一弦波訊號,使用Fourier transform,即complex series form 展開。因只有sin成分,cos與此正交,f(t)之Fourier transform可以只用sin series表示,以下用圖解說。 ## 2.FFT 模擬 with Matlab ### 單頻 $0.7*sin(2*pi*50*t)$ #### Time Domain ![](https://i.imgur.com/dAEa5LL.png) #### Spectrum ![](https://i.imgur.com/HonLHwa.png) 其中bn即為amplitude specturm之peak,而n=50即為頻率。 ![](https://i.imgur.com/roIjNeZ.png) 注意正交性帶來這個結果: ![](https://i.imgur.com/oYwmsO1.png) ### 多頻 $0.7*sin(2*pi*50*t) + sin(2*pi*120*t)$ #### Time Domain ![](https://i.imgur.com/js7Kmez.png) #### Spectrum ![](https://i.imgur.com/hrUS5k7.png) ### 有雜訊 $0.7*sin(2*pi*50*t) + sin(2*pi*120*t) + 2*randn(size(t))$ #### Time Domain ![](https://i.imgur.com/dNH3ABY.png) #### Spectrum ![](https://i.imgur.com/PY2ce5i.png) ## 3.實際頻譜/頻譜儀操作 ### 3.1 What can Spectrum do? + Frequency Responce -> Bode Plot ![](https://i.imgur.com/jWy3Qey.png) + Frequency Spectrum ![](https://i.imgur.com/4yASiRn.png) ### 3.2 Matlab上的FFT 詳情見 [Signal Processing - Fourier Transform](/OePXfsiAQ0iHYvd4wjG9tw) 以下是很好用於理解跟測試時間及頻率關係的matlab code ```shell=Matlab Fs = 1000; % Sampling frequency T = 1/Fs; % Sampling period L = 1; % Length of signal (sec) t = (0:T:L); % Time vector (sec) figure(1) S = sin(2*pi*100*t) + sin(2*pi*1*t); plot(t,S) title('Signal Corrupted with Zero-Mean Random Noise') xlabel('t (milliseconds)') ylabel('X(t)') Y = fft(S); P2 = abs(Y/L); P1 = P2(1:L*Fs/2+1); %foldering P1(2:end-1) = 2*P1(2:end-1)/Fs; f = Fs*(0:(L*Fs/2))/(L*Fs); figure(2) plot(f,P1) title('Single-Sided Amplitude Spectrum of X(t)') xlabel('f (Hz)') ylabel('|P1(f)|') ``` ### 3.3 常見術語 + Central Freq. + Span + Resolution Bandwidth ![](https://i.imgur.com/tXFV2mW.png) + SNR ![](https://i.imgur.com/npRgQKt.png) ## 4.USRP打訊號表現 + 調變(AM/FM) ![](https://i.imgur.com/d89qUY2.png) AM: y(t) = [A + m(t)].c(t) FM: y(t) = A sin [ wct + (Δf / fm) sin wmt ] + 頻寬 + Fc + Filter ## 5.頻譜分析該注意的事 + 天線頻段 + 阻抗匹配 + 天線指向 + 衰減 ## Reference + Spectrum Analyzer Fundamentals – Theory and Operation of Modern Spectrum Analyzers, Rohde-Schwarz + https://www.strongpilab.com/trace-operation-spectrum/ ###### tags:`Signal Processing`, `FFT`,`Spectrum` >jessest94106@g.ncu.edu.tw Department of Space Science & Engineering Center for Astronautical Physics & Engineering National Central University, Taiwan [name=PieappleJ] [time=Sun, Mar 27, 2022]
{"metaMigratedAt":"2023-06-16T22:37:44.826Z","metaMigratedFrom":"Content","title":"頻譜分析那些事","breaks":true,"contributors":"[{\"id\":\"81efaec0-484d-4044-9e62-fd943b5b9ca6\",\"add\":3913,\"del\":601},{\"id\":\"63168405-c16e-4b6d-9540-8f62d6980a57\",\"add\":12,\"del\":12}]"}
    936 views