EEE598D: Analog Filter & Signal Processing Circuits Instructor: Dr. Hongjiang Song Department of Electrical Engineering Arizona State University Dr. Hongjiang Song, Arizona State University Thursday January 17, 2002 Today: Continuous-Time Filter Fundamental • Fourier Transformation • Laplace Transformation • Transfer Function (TF) Dr. Hongjiang Song, Arizona State University RLC Network and Response • The traditional question: For a given input vi(t) signal and a RLC network, what is the output signal vo(t)? Vi L R Vo C d 2 vo (t ) dvo (t ) + RC + vo (t ) = vi (t ) LC 2 dt dt vo (t ) |t =0 = vo (0) (1.1) dvo (t ) dv (0) |t =0 = o dt dt • The solution: It is a differential equation. – Solvable but not trivial!! Dr. Hongjiang Song, Arizona State University Steady State Response • Sometimes we only interest in the steady state responses of the form: vi (t ) = Vi e jωt and vo (t ) = Vo e (1.2) jω t • Equation (1.1) can be simplified as: (1 − LCω 2 + jωRC )Vo = Vi or Vo 1 H ( jω ) = = Vi 1 − LCω 2 + jωRC Dr. Hongjiang Song, Arizona State University (1.3) Frequency Response • The magnitude (gain) response of the network is 1 H ( jω ) = (1 − LCω ) + ( RCω ) 2 2 • We have: H ( jω ) ω → 0 = 1 H ( jω ) ω →∞ = 1 / ω 2 |ω →∞ = 0 H ( jω ) ω →1/ LC Dr. Hongjiang Song, Arizona State University 1 L = ≡Q R C 2 (1.5) Gain Response Q≡ Dr. Hongjiang Song, Arizona State University ω LC 1 L R C Frequency Response • The phase response of the network is RCω θ ( jω ) = tan ( ) 2 (1 − LCω ) −1 • We have: θ ( jω ) |ω →0 = 0 θ ( jω ) |ω →1/ LC = −π / 2 θ ( jω ) |ω →∞ = −π Dr. Hongjiang Song, Arizona State University (1.6) Phase Response ω LC Dr. Hongjiang Song, Arizona State University General Network Responses • Question 1: How is the network response to a general input (non-sinusoidal) signal? • Question 2: What is the network’s transient response (non-steady-state response)? Dr. Hongjiang Song, Arizona State University Answer to Q1: Fourier Transformation • A generic input signal can be expressed by combination of sinusoidal signal using Fourier transformation Dr. Hongjiang Song, Arizona State University The Fourier Transformation • The Fourier Transform pair is defined as ∞ F ( jω ) = ∫ f (t )e − jωt dt −∞ 1 f (t ) = 2π Spectrum of f(t) Dr. Hongjiang Song, Arizona State University ∞ j ωt ( ) ω F j e dt ∫ −∞ (1.7) Useful Fourier Transform Pairs (1.8) Dr. Hongjiang Song, Arizona State University Useful Fourier Transform Pairs (1.9) Dr. Hongjiang Song, Arizona State University Response by Fourier Transformation • Apply Fourier Transform to (1.1), we have − LCω 2Vo ( jω ) + jωRCVo ( jω ) + Vo ( jω ) = Vi ( jω ) Vo ( jω ) 1 ⇒ H ( jω ) ≡ = Vi ( jω ) 1 − LCω 2 + jωRC Dr. Hongjiang Song, Arizona State University Answer to Q2: Laplace Transformation • The transient response of the network can be calculated by using the Lapalce Transformation Dr. Hongjiang Song, Arizona State University Laplace Transformation • The (one side) Laplace Transform is defined as: ∞ F ( s ) ≡ L( f (t )) ≡ ∫ f (t )e − st dt 0 σ + j∞ 1 − st f (t ) = L ( F ( s )) = F ( s )e dt ∫ 2πj σ − j∞ −1 Dr. Hongjiang Song, Arizona State University (1.10) Useful Laplace Transform Pairs (1.11) Dr. Hongjiang Song, Arizona State University Useful Laplace Transform Pairs (1.12) Dr. Hongjiang Song, Arizona State University General Continuous-Time Linear Filter • A general continuous-time (CT) linear filter can be expressed as a differential equation: k n m k d vi (t ) d vo (t ) bk = ∑ ak ∑ k k dt dt k =0 k =0 k k d vo (t ) d vo (0) |t =0 = k k dt dt Dr. Hongjiang Song, Arizona State University (1.13) k = 0,1,...n − 1 Filter Response using Laplace Transform • The response of the above filter can be solved using Laplace Transform: m d k vo (t ) d k vi (t ) L(∑ bk k ) = L(∑ ak k ) dt dt k =0 k =0 n m Vo ( s ) = ∑a s k =0 n (1.14) k k k b s ∑k Vi ( s ) + T ( s ) Initial condition k =0 m k a s ∑ k vo (t ) = L−1[( k =n0 ∑b s k =0 Dr. Hongjiang Song, Arizona State University k k )Vi ( s )] + L−1[T ( s )] (1.15) Example of Laplace Transform • Look at filter response of the above filter for given initial condition Vi L R Vo C dvo (t ) d 2 vo (t ) − jωt + RC + vo (t ) = Ae LC 2 dt dt (1.16) vo (t ) |t =0 = 0 dvo (t ) |t =0 = 0 dt Dr. Hongjiang Song, Arizona State University Example of Laplace Transform • Using (1.5), we have • Let Vm 1 Vo ( s ) = ( LCs 2 + RCs + 1) ( s + jω ) R α≡ 2L R 2 1 β≡ −( ) LC 2 L (1.17) (1.18) • We have Vo ( s ) = Vm 1 ( s + α + jβ )( s + α − jβ ) ( s + jω ) Dr. Hongjiang Song, Arizona State University (1.19) Example of Laplace Transform • With partial-fraction expansion, we have A B C (1.20) + + Vo ( s ) = ( s + jω ) ( s + α + jβ ) ( s + α − jβ ) • where Vm | s = jω A= 2 s LC + sRC + 1 Vm |s =α + jβ B= ( s + α − jβ )( s + jω ) Vm |s =α − jβ C= ( s + α + jβ )( s + jω ) Dr. Hongjiang Song, Arizona State University (1.21) Example of Laplace Transform • Using (1.3) the response can be solved as: vi (t ) = Ae − jωt + Be −αt − jβt + Ce −αt + jβt (1.22) transient response (→ 0 as t → ∞) Vo ( s ) A= |s = jω steady-state response Vi ( s ) Dr. Hongjiang Song, Arizona State University Transfer Function • For steady-state response, the s-domain ratio of the output signal to input signal is known as the transfer function (TF) of the filter: V (s) H ( s) = o Vi ( s ) (1.23) • For the given example above, TF is given as: Vo ( s ) 1 = 2 H ( s) = Vi ( s ) s LC + sRC + 1 Dr. Hongjiang Song, Arizona State University (1.24) S-Domain Representation of Operators • By using the Laplace Transform, we have: Integrator: t 1 y (t ) = ∫ x(τ )dτ ⇒ Y ( s ) = X ( s ) ⇒ s to ∫ dt ⇔ 1 (1.25) s Differentiator: dx(t ) ⇒ Y ( s ) = sX ( s ) ⇒ y (t ) = dt Dr. Hongjiang Song, Arizona State University d dt ⇔ s (1.26) Transfer Function • Generally, for a filter described the following linear differential equation: k n m k d vo (t ) d vi (t ) bk = ∑ ak ∑ k k dt dt k =0 k =0 (1.27) • The transfer function can be calculated using Laplace Transform as: m H (s) = k a s ∑ k k =0 n k b s ∑k k =0 Dr. Hongjiang Song, Arizona State University am s m + am −1s m −1 + ... + a1s + a0 = (1.28) n n −1 bn s + bn −1s + ... + b1s + b0 S-Domain Impedance of Components • By Laplace Transform, we have: Resistor: v(t ) = Ri (t ) ⇒ V ( s ) = RI ( s ) ⇒ ZR = R Capacitor: dv(t ) I (t ) 1 = ⇒ V (s) = I (s) dt C sC ⇒ 1 (1.29) ZC = sC ⇒ Z L = sL Inductor: di (t ) v(t ) = L ⇒ V ( s ) = sLI ( s ) dt Dr. Hongjiang Song, Arizona State University Calculation of Transfer Function • The transfer function of a linear circuit can be calculated by using s-domain impedance through Kirchhoff’s law: Vi sL R Vo 1/sC Vo ( s ) ZC 1 / sC = = Vi ( s ) Z L + Z R + Z C sL + R + 1 / sC 1 = 2 s LC + sRC + 1 Dr. Hongjiang Song, Arizona State University (1.30) System Model of CT Filters • Linear Systems constant constant K (a1 x1 (t ) + a2 x(t )) = a1 K ( x1 (t )) + a2 K ( x2 (t )) • Time- (shift-) Invariant Systems if y (t ) = K ( x(t )) Then y (t − to ) = K ( x(t − to )) Dr. Hongjiang Song, Arizona State University S-Domain Representation of Linear TimeInvariant (LTI) Systems ∞ L{ y (t )} = L{K ( x(t ))} = L{K ( ∫ x(ζ )δ (t − ζ )dζ )} −∞ ∞ = L{ ∫ x(ζ ) K (δ (t − ζ ))dζ } = L{x(t )} ⋅ L{K (δ (t ))} −∞ Y ( s ) ≡ L{ y (t )} Let: X ( s ) ≡ L{x(t )} Then: h(t ) ≡ K (δ (t )) H ( s ) ≡ L{h(t )} Y (s) = H (s) X ( s) Transfer Function (TF) of system Impulse response of system Remark: Continuous-time LTI System can be completely determined by its impulse response. Dr. Hongjiang Song, Arizona State University Steady State Response of ContinuousTime LTI Systems x(t ) = Au (t )e jωt ∞ ⇒ X ( s ) = A∫ e −( s − jω ) t dt = 0 Y (s) = H (s) ⋅ X (s) = A s − jω H ( s ) H ( jω ) = + Other Terms s − j ω s − jω y (t ) t →∞ = Au (t ) H ( jω )e jωt + yh (t ) = H ( jω ) x (t ) Approach zero for stable system Steady State Response of the system Dr. Hongjiang Song, Arizona State University