ECEN3513 Signal Analysis Lecture #23 16 October 2006 Read 6.1, 6.2; Review 6.3 Problems 4.7-2, 6.2-1, 6.3-5 Test #2 on 27 October Quiz 5 results: Hi = 7.8, Lo = 2.9, Ave = 5.35 Standard Deviation = 1.35 Read 6.4, 6.5 (1st section), 6.7 (1st section) Problems: 6.4-3, 6.4-4, 6.5-1 Test #2 on 27 October Quiz 6 Results: Hi =8.0, Lo = 2.0, Ave. = 5.00 Standard Deviation = 2.04 yp(t): Base Periodic Function i i 0 200 t 5 10 j 0 500 f 5.000001 10 i j sin 2 f time varies from -5 to +5 seconds 200 j 500 1 500 i Ts 50 1 sin 2 f 4 4 1 1 2 f 2 f 4 4 envelope ( f) x Frequencies vary from -5 to +5 Hz in 0.02 Hertz increments 500 Reenvelopefjcos 2fjti Imenvelopefjsin2fjti j0 x x i j0 1.5 1 xi 0.5 0 0.5 6 4 2 0 ti 2 4 6 i Ts Fourier Transform of yp(t) 2 envelope f j 1 0 6 4 2 0 2 4 6 2 4 6 fj 4 atan2 Re envelope f j Im envelope f j 2 0 6 4 2 0 fj y(t): Periodic Time Domain Waveform i i 0 200 t 5 10 j 0 20 f 5.00000 10 i j sin 2 f envelope ( f) 20 x i time varies from -5 to +5 seconds 200 j Frequencies vary from -5 to +5 Hz in 0.5 Hertz increments 20 1 1 sin 2 f 4 4 1 1 2 f 2 f 4 4 20 j Re envelope f cos 2 f t j i j0 j Im envelope f sin 2 f t x x j i i j0 1 xi 0.5 0 6 Ts 2 4 2 0 ti 2 4 6 i Ts Fourier Series of y(t) 2 envelope f j 1 0 6 4 2 0 2 4 6 2 4 6 fj 4 atan2 Re envelope f j Im envelope f j 2 0 6 4 2 0 fj Linear Time Invariant? yes δ(t) → h(t)... an impulse response exists convolution works y(t) transfer function exists Y(f) = x(t) * h(t) = X(f) H(f) Energy & Power transfer functions exist |Y(f)|2 = |X(f)|2 |H(f)|2 J/Hz SYY(f) = SXX(f) |H(f)|2 W/Hz Linear Time Invariant? no δ(t) → h(t)... an impulse response exists convolution gives you the wrong answer y(t) transfer function does not exist Y(f) ≠ x(t) * h(t) ≠ X(f) H(f) Energy & Power transfer functions don't exist ≠ |X(f)|2 |H(f)|2 J/Hz SYY(f) ≠ SXX(f) |H(f)|2 W/Hz |Y(f)|2 Autocorrelation (Power Signal) RXX(τ) = lim 1 T→∞ T x(t)x(t+τ)dt T When in doubt, evaluate this one first If = 0, evaluate energy signal autocorrelation SXX(f) = lim |X(f)|2 non-zero at same T→∞ T freqs as X(f)! RXX(τ) ↔ SXX(f) W/Hz Autocorrelation & Power Spectrum form a Fourier Transform pair Autocorrelation (Power Signal) RXX(τ) = lim 1 T→∞ T x(t)x(t+τ)dt = T +∞ SXX(f) e -j2πfτ df -∞ units are W/Hz RXX(0) = ?? Average (normalized) power of x(t) How can you find RXX(0) from SXX(f)? Evaluate at τ = 0 The area under power spectrum = average (normalized) power of x(t) What's SXX(f) for x(t) = 3cos2π100t? What's SXX(f) for x(t) = 3sin2π100t? Autocorrelation (Energy Signal) RXX(τ) = lim T→∞ x(t)x(t+τ)dt T RXX(τ) ↔ |X(f)|2 J/Hz Autocorrelation & Energy Spectrum form a Fourier Transform pair GXX(f) = |X(f)|2 non-zero at same freqs as X(f)! Autocorrelation (Energy Signal) RXX(τ) = lim T→∞ x(t)x(t+τ)dt = T +∞ |X(f)|2 e -j2πfτ df -∞ units are J/Hz RXX(0) = ?? Average (normalized) energy of x(t) How can you find RXX(0) from |X(f)|2? Evaluate at τ = 0 The area under energy spectrum = average (normalized) energy of x(t) What's |X(f)|2 for x(t) = rect[(t-0.5)/1]? What's |X(f)|2 for x(t) = rect[t/1]? Autocorrelation (Real World) ^ RXX(τ) = 1 T x(t)x(t+τ)dt T observation interval ^ ^ RXX(τ) ↔ SXX(f) W/Hz FT of autocorrelation yields estimate of Power Spectrum ^ SXX(f) = ^ |X(f)|2, T ^ where X(f) = x(t) e-jωt dt T Autocorrelation (Real World) ^ RXX(τ) = x(t)x(t+τ)dt T observation interval ^ ^ RXX(τ) ↔ GXX(f) W/Hz FT of autocorrelation yields estimate of Energy Spectrum ^ GXX(f) = ^ |X(f)|2, ^ where X(f) = x(t) ejωt dt T sinusoid + d.c. i i 0 200 t 50 100 j 0 400 f 4.00000 8 i time varies from -50 to +50 seconds 200 j j Ts 50 Frequencies vary from -4 to +4 Hz in 0.02 Hertz increments 400 envelope ( f) 0.5j ( f 0.099) ( f .101) 0.5j ( f .101) ( f .099) ( f .001) ( f .001) 50 .000001 δ(f-0.1) 400 x i j Re envelope f δ(f+0.1) 400 cos 2 f t j i j0 δ(f) j Im envelope f sin 2 f t x x j i i j0 2 xi 1 0 60 40 20 0 ti 20 max( x) 2 40 60 i Ts Magnitude & Phase Plots 60 envelope f j 40 20 0 4 3 2 1 0 1 2 3 4 fj j Imenvelopefj phase atan2 Re envelope f j f 205 0.1 phase 180 205 90 2 phase j 0 2 4 3 2 1 0 fj 1 2 3 4
© Copyright 2026 Paperzz