EG1110 SIGNALS AND SYSTEMS – A class of these inputs are sinusoidal inputs Frequency response function Intro • Transfer functions give information about a system’s – stability - very important – transient behaviour • Given a system with transfer function G(s), output can be computed using Y (s) = G(s)U (s) where U (s) is (almost) any input • However, often interested in a system’s output due to a particular input. 1 2 Basic idea of frequency response function The maths (optional) • Frequency response function determines output of system if input is sinusoidal. u(t) y(t) G • It can be proved, via convolution that Given a sinusoidal input an LTI system will produce a sinusoidal output, possible shifted in amplitude and phase • Let G be a stable system with transfer function G(s) (poles in left half s-plane) (in steady state) • This means after initial conditions and transients have died away • Let Y (s) and U (s) be Laplace Transforms of y(t) and u(t). • Let u(t) = αsin(ωt) the output is a sine wave appropriately scaled and shifted in phase ⇒ U (s) = αω s2 + ω 2 (1) (From Laplace Transform tables) • Let G(s) be G(s) = 3 (s + z1)(s + z2) . . . (s + p1)(s + p2) . . . 4 (2) • We calculate the output as • Hence taking inverse Laplace transforms... e−p1t + A{z2e−p2t + . . .} y(t) = β1e−jωt + β2ejωt + A | 1 αω Y (s) = G(s)U (s) = G(s) 2 s + ω2 limt→∞ =0 (as G is stable, p1, p2, . . . > 0) • We know that, using partial fractions we can split up G(s) as G(s) = ? ? + + ... s + p1 s + p2 (3) Hence in steady state yss(t) = β1e−jωt + β2ejωt and we can separate s2 αω ? ? = + + ω 2 s + jω s − jω • Thus we can write an expression for Y (s) as Y (s) = To find β1, β2 note that β1 A2 β2 A1 G(s)ωα = + + + s2 + ω 2 s + jω s − jω s + p1 s + p2 β1(s − jω) + β2(s + jω) + (s + jω)(s − jω)[A1/(s + p1) + A2/(s + p2) + . . .] = (s + jω)(s − jω) A2 β2 A1 β1 + + + s + jω s − jω s + p1 s + p2 6 5 This implies that Thus, yss = G(s)ωα = β1(s − jω) + β2(s + jω) + (s + jω)(s − jω)(A1/(s + p1) + A2/(s + p2) + . . .) αG(−jω) −jωt αG(jω) jωt + e e −2j 2j However, as G(jω) is a complex number we can write it as Letting s = jω we get G(jω) = |G(jω)|ejφ G(jω)α = β2(2jω) αG(jω) ⇒ β2 = 2j where φ = 6 G(jω) = arctan(=(G(jω))/<(G(jω))) Similarly Letting s = −jω we get G(−jω) = |G(jω)|e−jφ G(−jω)α = β1(−2jω) αG(−jω) ⇒ β1 = −2j 7 Thus −e−jωt−φ + e−jωt+φ 2j = α|G(jω)| sin(ωt + φ) yss = α|G(jω)| 8 (4) Implications of frequency response function • Given a system G(s) and an input α sin(ωt) the output (in steady state) is given by y(t) = Y sin(ωt + φ) 6 G(jω) = 6 where – ω - frequency of input sine (and output sine). – α - amplitude of input sine – Y = α|G(jω)| - amplitude of the output. – φ = 6 G(jω) - phase difference between output and input sine • G(jω) is known as the frequency response function or the sinusoidal transfer function. • Can also calculate the frequency response function from input/output behaviour: 9 Y (jω) |G(jω)| = U (jω) 10 Y (jω) U (jω)