Frequency Response of Discrete-time LTI Systems Prof. Siripong Potisuk Transfer Functions Let x[n] be a nonzero input to an LTI discrete-time system, and y[n] be the resulting output assuming a zero initial condition. The transfer function, denoted by H(z), is defined: Z{ y[n]} Y ( z ) H ( z) Z{ x[n]} X ( z ) Can be determined by taking the Z-transform of the governing LCCDE and applying the delay property N M Z a0 y[n] ak y[n k ] bk x[n k ] k 1 k 0 Y ( z ) b0 b1 z 1 bM z M H ( z) X ( z ) a0 a1 z 1 a N z N The system’s impulse response: h[n] Z 1 H ( z) BIBO Stability • BIBO = Bounded-input-bounded-output • A linear time-invariant (LTI) discrete-time system with transfer function H(z) is BIBO stable if and only if the poles of H(z) satisfy | pi | 1, 1 i N • That is, the poles of a stable system, whether simple or multiple, must all lie strictly within the unit circle in the complex z-plane the unit circle • Marginally unstable one or more simple Ex. Consider a 2nd order discrete-time LTI system with y[n] 1.2 y[n 1] 0.32y[n 2] 10x[n 1] 6 x[n 2] (a) Determine the transfer function of the system and comment on the stability of the system. (b) Determine the zero-state response due to a unit-step input and the DC gain of the system. Frequency Response For a discrete-time LTI system, the frequency response is defined as j Y ( e ) j H (e ) X (e j ) In terms of transfer function, j H ( e ) H ( z ) z e j , The frequency response is just the transfer function evaluated along the unit circle in the complex z-plane. Im(z) periodic in with period 2 H(ej) 1 Re(z) H (e j ) H (e j 2f ) H ( z ) z e j 2 f , 2 f H ( F ) H (e j 2FTs Fs Fs ) H ( z ) z e j 2 FT s , F 2 2 For H(z) generated by a difference eq. with real coefficients, Fs H ( F ) H ( F ), 0 F 2 A( F ) | H ( F ) | (Evenfunction) Im{H ( F )} ( F ) tan (Odd function) Re{H ( F )} 1 Ex. Consider a 2nd order discrete-time system with z 1 H ( z) 2 z 0.64 Plot the magnitude and phase responses of the system. Determine also the DC and the high-frequency gain. Effects of Pole & Zero Locations j1 • A zero at z z1 1e indicates that the filter will fully reject spectral component of input at 1 • Effects of a zero located off the unit circle depends on its distance from the unit circle. • A zero at origin has no effect. • A pole on the unit circle means infinite gain at that frequency. • The closer the poles to the unit circle, the higher the magnitude response. Ex. Roughly sketch the magnitude response of the system with z 2 ( z 1) H ( z) ( z 0.89)(z 0.5 j 0.8)(z 0.5 j 0.8) Ex. Roughly sketch the magnitude response of the system with 0.05634(1 z 1 )(1 1.0166z 1 z 2 ) H ( z) (1 0.683z 1 )(1 1.4461z 1 0.7957z 2 ) For a given choice of H(ej) as a function of , the frequency composition of the output can be shaped: - preferential amplification - selective filtering of some frequencies Ex. Consider a 1st order IIR digital filter with 0.5(1 c)( z 1) H ( z) z c (a) Determine c such that the system is BIBO stable. (b) Without plotting the magnitude response of the system, determine the type of this filter. (c) Verify the answer in (b) using MATLAB.