FREQUENCY RESPONSE ANALYSIS Introduction The system characteristics are described using a steady-state sinusoidal input signal and considering the response of the system as the frequency of the sinusoid is varied. The frequency response is defined as the steady-state response of the system to a sinusoidal input signal. s = jω G( s ) ⎯ ⎯⎯→ G( jω ) Forms of plotting G(jω) -- Bode, Nyquist, Nichols(gain- phase) plot. Experimental information can be used for design purposes. Correlation of the sinusoidal and time response F ( s ) = L [ f ( t )] = ∫ ∞ 0 f ( t )e − st dt 1 σ + j∞ st f ( t ) = L [ F ( s )] = F ( s ) e ds ∫ − j ∞ σ 2πj −1 Laplace Transform where s = σ + jω F ( jω ) = F[ f ( t )] = ∞ ∫−∞ f ( t )e − jωt dt 1 j∞ jω t f ( t ) = F [F ( jω )] = F ( j ω ) e dω Fourier Transform ∫ 2πj j∞ −1 Laplace Transform : Fourier Transform : poles and zeros on s-plane amplitude and phase information corresponding to a sinusoidal input. ∞ ∫ | f ( t )| dt < ∞ → Fourier transform exists −∞ Let y(s) = T(s)r(s), 1 then y ( t ) = 2π ∞ jωt ω ω T ( j ) r ( j ) e dω ∫ −∞ Complex variable (a). s = σ + jω Cartesian coordinate (b). s = r ( cosθ + jsinθ ) Polar coordinate (c). s = r ejθ Euler form ω r σ θ Bode Plot H.W. Bode (Bell Lab. 1939-42) Frequency response G(jω): the steady-state response of a system (forced response of an unstable system) with transfer function G(s) to a sinusoid of unit amplitude and frequency ω is a sinusoid of amplitude A(ω) and phase Φ(ω), where G(jω) = A(ω) e jΦ(ω) . Example Consider the system response described by ω Y (s) = G(s) 2 , a partial-fraction expansion 2 s +ω corresponding to Y(s) is b1 b2 b0 b0* Y ( s) = + + LL + + s + a1 s + a2 s + jω s − jω and − jφ ⎡ G ( s )ω ⎤ G ( − jω ) G ( jω ) e b0 = ⎢ = = ⎥ − j2 −2j ⎣ s − jω ⎦ s = − jω jφ ⎡ G ( s )ω ⎤ G ( jω ) G ( jω ) e b0 * = ⎢ = = ⎥ j2 2j ⎣ s + j ω ⎦ s = jω e j (ω t +φ ) − e − j (ω t +φ ) = sin(ω t + φ ) 2j ⎧⎪ Re( G ( jω ) ) ⎫⎪ φ = tan ⎨ ⎬ Im( G ( j ω ) ) ⎪⎩ ⎪⎭ −1 y (t ) = b1e − a1t + b2 e − a2t + LL + G ( jω ) sin(ω t + φ ) y (∞) = G ( jω ) sin(ω t + φ ) for a stable system ∴ A(ω ) = G (ω ) and φ (ω ) = ∠(G (ω ) ) 1 Input signal 0.5 gain A(ω) time (sec) 0 phase Φ(ω) Output signal -0.5 -1 0 2 4 6 8 The Bode plotting technique allows plotting frequency response in an efficient way and it is sufficient for control system design. 1) magnitude curve (log. Scale dB) vs. frequency (log. Scale radians/second) 2) phase curve (linear scale degree or radian) vs. frequency (log. scale) This allows the plotting of a high-order G(jω) by simple graphical addition of the separate terms. Example v ⎛ r1 ⎞ j (φ1 −φ2 −φ3 ) s1 G ( jω ) = v v = ⎜ ⎟e s2 ⋅ s3 ⎝ r2 ⋅ r3 ⎠ r1 G ( jω ) = ⇒ log10 G ( jω ) = log10 r1 − log10 r2 − log10 ⋅ r3 r2 ⋅ r3 arg{G ( jω )} = φ1 − φ2 − φ3 log10 Ae jφ = log10 A +j { φ log10 e 123 phase plot magnitute plot ln G ( jω ) = ln | G (ω ) | + jφ (ω ) The unit of the magnitude plot P2 G dB = 10 log10 P1 V2 = 20 log10 V1 dB (decibels): 20log10|G| (Alexander Graham Bell) The unit of the phase plot Degree or radian 1 0 Example s+8 G( s) = 2 s + 2 s + 10 s+8 = ( s + 1 + 3 j )( s + 1 − 3 j ) jω 5 0 -5 jω + 8 G ( jω ) = ( jω + 1 + 3 j )( jω + 1 − 3 j ) -10 -10 ( jω ) − ( −8) = (( jω ) − ( −1 − 3 j ))(( jω ) − ( −1 + 3 j )) π dis ( zi , jω ) G ( jω ) = i π dis ( pk , jω ) k -8 -6 -4 -2 0 2 , arg{G ( jω )} = ∑ arg( zi , jω ) − ∑ arg( pk , jω ) i k Example Let s = jω , then the frequency response can be obtained as 2 G ( jω ) = jω ( jω + 1)( jω + 2) Gain and phase of the frequency response are 1 1 1 2 G ( jω ) = 2 = jω jω + 1 jω + 2 ω (1 + ω 2 )( 4 + ω 2 ) 1 1 1 ∠G ( jω ) = ∠ +∠ +∠ jω jω + 1 jω + 2 −1 = −90 − tan ω − tan o −1 ω 2 20 log G ( jω ) (dB) ∠ G ( jω ) (度 ) ω(rad/sec) G ( jω ) 0.01 100 40 -90.9 0.05 20 26 -94.3 0.1 10 20 -98.6 0.5 1.7 4.8 -130.6 1 0.63 -4 -161.6 5 1 .5 × 10 −2 -36.7 -236.9 10 2 × 10 −3 -54.2 -253 50 1 .6 × 10 −5 -96 -266.6 100 2 × 10 −6 -114 -268.3 G ( jω ) ∠G ( jω ) 100 G ( jω ) 50 0 -50 -100 -150 degree -200 -250 -300 0 20 40 60 frequency 80 (rad/sec) 100 G ( jω ) ∠G ( jω ) 50 G ( jω ) in dB 0 -50 degree -100 -150 -200 -250 0.01 0.1 1 frequency 10 (rad/sec) 100 The advantages of using Bode technique Bode plots of systems in series simply add. A much wider range of system behaviour can be displayed on a signal plot. Bode plot can be determined experimentally. Integration of system analysis and design. Bode form of the transfer function (b1s + 1)(b2 s + 1)LL kG ( s ) = k0 l s ( a1s + 1)( a2 s + 1)LL For type 0 system and s=jω , k0 is the gain at ω = 0 and k0 is called the dc gain of the system. Classes of terms of transfer functions (1). k0 ω (3). ( j + 1) ±1 ωc (2). ( jω ) ± n ⎡⎛ jω ⎞ ⎤ jω (4). ⎢⎜ + 1⎥ ⎟ + 2ζ ωn ⎥ ⎢⎝ ω n ⎠ ⎣ ⎦ 2 ±1 k = -0.1 (rad/sec) constant gain : k -20 dB (degree) 180 magnitude: 20 log |k| (rad/sec) phase: ⎧⎪ 0o ⎨ o ⎪⎩180 if k is positive if k is negative Im jω s-plane Re ( jω ) ±1 magnitude: 20 log ω ±1 phase: ±90o 1 s 20 dB 0.1 0.1 1 -90o rad/sec 1 rad/sec magnitude plot phase plot Im ωc φc jωc Re ω (j + 1) ±1 ωc 2⎤ ⎡ ⎛ω ⎞ ⎛ω ⎞ ±20 log 1 + ⎜ ⎟ = ±10 log ⎢1 + ⎜ ⎟ ⎥ ⎝ωc ⎠ ⎢ ⎝ωc ⎠ ⎥ ⎣ ⎦ 2 magnitude: phase: −1 ⎡ ω ⎤ tan ⎢± ⎥ ω c⎦ ⎣ Asymptotes 0 dB -10 |G(jω)| -20 -30 -40 -2 10 -1 10 0 10 frequency 1 10 2 10 (rad/sec) Corner frequency: ωc ± 3dB correction at breakpoint Asymptotes 0 (degree) -20 -40 arg{G(jω)} -60 -80 -100 -2 10 -1 10 10 0 freuency 10 1 10 2 (rad/sec) ⎡⎛ jω ⎞ jω ⎤ + 1⎥ ⎟ + 2ζ ⎢⎜ ωn ⎥ ⎢⎣⎝ ω n ⎠ ⎦ 2 magnitude: phase: ±1 ⎧⎡ 2 ⎤2 2⎫ ⎡ 2ζω ⎤ ⎪ ⎛ ω ⎞ ⎪ ⎢ ⎥ +⎢ ±10 log ⎨ 1 − ⎜ ⎟ ⎥ ⎬ ⎣ ωn ⎦ ⎪ ⎪⎢⎣ ⎝ ω n ⎠ ⎥⎦ ⎩ ⎭ ⎡ ⎛ ω ⎞⎤ ⎢ ±2ζ ⎜ ⎟⎥ ⎝ωn ⎠ ⎥ −1 ⎢ tan ⎢ 2⎥ ⎢1 − ⎛ ω ⎞ ⎥ ⎢ ⎜⎝ ω ⎟⎠ ⎥ n ⎣ ⎦ jω jω n − ζω n + j 1 − ζ 2 ωn − ζω n − j 1 − ζ 2 50 1 G( s) = 2 s + 0.4 s + 1 0 -50 -1 10 10 0 10 1 0 -100 -200 -1 10 10 0 10 1 ζ = 0 .2 ω n = 1 rad / sec Frequency response of 2nd order system ωr = ωn 1 − 2ζ 2 , ζ < 0.707 −1 M pω = G ( ωr ) = ⎛⎜ 2ζ 1 − ζ 2 ⎞⎟ , ζ < 0.707 ⎝ ⎠ Frequency response measurements The frequency response of a stable system can be obtained using frequency response analyzer. The system poles and zeros then can be determined by the experimental data. DC gain = 24.5 dB ≈ 16 poles: 15, 150 rad/sec 40 dB 20 0 -20 -40 -60 -2 10 10 0 10 2 4 10 rad/sec 0 degree -50 16 × 15 × 150 G( s) = ( s + 15)( s + 150) -100 -150 -200 -2 10 10 0 10 15 2 150 10 4 rad/sec 散戲