Use: Analysis of systems, simple convolution, shorthand for e jw, stability. Motivation easier to write. Or X(z) = Z {x(n)}

Similar documents
ELEG 305: Digital Signal Processing

Discrete-Time Signals and Systems. The z-transform and Its Application. The Direct z-transform. Region of Convergence. Reference: Sections

ECE503: Digital Signal Processing Lecture 4

Discrete-time signals and systems

Need for transformation?

y[n] = = h[k]x[n k] h[k]z n k k= 0 h[k]z k ) = H(z)z n h[k]z h (7.1)

UNIT-II Z-TRANSFORM. This expression is also called a one sided z-transform. This non causal sequence produces positive powers of z in X (z).

Z Transform (Part - II)

z-transform Chapter 6

6.003: Signals and Systems

Z-Transform. The Z-transform is the Discrete-Time counterpart of the Laplace Transform. Laplace : G(s) = g(t)e st dt. Z : G(z) =

(i) Represent discrete-time signals using transform. (ii) Understand the relationship between transform and discrete-time Fourier transform

Lecture 04: Discrete Frequency Domain Analysis (z-transform)

Chapter Intended Learning Outcomes: (i) Understanding the relationship between transform and the Fourier transform for discrete-time signals

ECE-S Introduction to Digital Signal Processing Lecture 4 Part A The Z-Transform and LTI Systems

ESE 531: Digital Signal Processing

Let H(z) = P(z)/Q(z) be the system function of a rational form. Let us represent both P(z) and Q(z) as polynomials of z (not z -1 )

8. z-domain Analysis of Discrete-Time Signals and Systems

Discrete Time Systems

DSP-I DSP-I DSP-I DSP-I

III. Time Domain Analysis of systems

ESE 531: Digital Signal Processing

SIGNALS AND SYSTEMS. Unit IV. Analysis of DT signals

EE 521: Instrumentation and Measurements

Review of Discrete-Time System

Lecture 19 IIR Filters

Digital Signal Processing Lecture 10 - Discrete Fourier Transform

Module 4 : Laplace and Z Transform Problem Set 4

Very useful for designing and analyzing signal processing systems

Digital Signal Processing. Midterm 1 Solution

Signals and Systems. Spring Room 324, Geology Palace, ,

Lecture 7 Discrete Systems

! z-transform. " Tie up loose ends. " Regions of convergence properties. ! Inverse z-transform. " Inspection. " Partial fraction

The Z-Transform. Fall 2012, EE123 Digital Signal Processing. Eigen Functions of LTI System. Eigen Functions of LTI System

The Z transform (2) 1

Z-Transform. 清大電機系林嘉文 Original PowerPoint slides prepared by S. K. Mitra 4-1-1

Z-Transform. x (n) Sampler

EE102B Signal Processing and Linear Systems II. Solutions to Problem Set Nine Spring Quarter

Chapter 7: The z-transform

DIGITAL SIGNAL PROCESSING. Chapter 3 z-transform

Topic 4: The Z Transform

Digital Signal Processing:

Z - Transform. It offers the techniques for digital filter design and frequency analysis of digital signals.

How to manipulate Frequencies in Discrete-time Domain? Two Main Approaches

Lecture 2 OKAN UNIVERSITY FACULTY OF ENGINEERING AND ARCHITECTURE

Solutions: Homework Set # 5

EE 225D LECTURE ON DIGITAL FILTERS. University of California Berkeley

# FIR. [ ] = b k. # [ ]x[ n " k] [ ] = h k. x[ n] = Ae j" e j# ˆ n Complex exponential input. [ ]Ae j" e j ˆ. ˆ )Ae j# e j ˆ. y n. y n.

LECTURE NOTES DIGITAL SIGNAL PROCESSING III B.TECH II SEMESTER (JNTUK R 13)

ELEN E4810: Digital Signal Processing Topic 4: The Z Transform. 1. The Z Transform. 2. Inverse Z Transform

X (z) = n= 1. Ã! X (z) x [n] X (z) = Z fx [n]g x [n] = Z 1 fx (z)g. r n x [n] ª e jnω

EEL3135: Homework #4

GATE EE Topic wise Questions SIGNALS & SYSTEMS

VI. Z Transform and DT System Analysis

Signals and Systems Lecture 8: Z Transform

Discrete-time first-order systems

UNIVERSITY OF OSLO. Please make sure that your copy of the problem set is complete before you attempt to answer anything.

Digital Signal Processing Lecture 4

Discrete-Time David Johns and Ken Martin University of Toronto

Signals and Systems. Problem Set: The z-transform and DT Fourier Transform

Discrete Time Systems

Lecture 18: Stability

EC Signals and Systems

2. Typical Discrete-Time Systems All-Pass Systems (5.5) 2.2. Minimum-Phase Systems (5.6) 2.3. Generalized Linear-Phase Systems (5.

Digital Signal Processing Lecture 3 - Discrete-Time Systems

Like bilateral Laplace transforms, ROC must be used to determine a unique inverse z-transform.

2.161 Signal Processing: Continuous and Discrete Fall 2008

EE Homework 5 - Solutions

QUESTION BANK SIGNALS AND SYSTEMS (4 th SEM ECE)

The z-transform Part 2

Digital Signal Processing Lecture 9 - Design of Digital Filters - FIR

Ch. 7: Z-transform Reading

Your solutions for time-domain waveforms should all be expressed as real-valued functions.

Digital Signal Processing Lecture 8 - Filter Design - IIR

The z-transform and Discrete-Time LTI Systems

ECE503: Digital Signal Processing Lecture 6

z-transforms Definition of the z-transform Chapter

Chap 2. Discrete-Time Signals and Systems

Responses of Digital Filters Chapter Intended Learning Outcomes:

Lecture 8 - IIR Filters (II)

Z-TRANSFORMS. Solution: Using the definition (5.1.2), we find: for case (b). y(n)= h(n) x(n) Y(z)= H(z)X(z) (convolution) (5.1.

Analog LTI system Digital LTI system

VU Signal and Image Processing

Module 4. Related web links and videos. 1. FT and ZT

INFINITE-IMPULSE RESPONSE DIGITAL FILTERS Classical analog filters and their conversion to digital filters 4. THE BUTTERWORTH ANALOG FILTER

Transform Analysis of Linear Time-Invariant Systems

/ (2π) X(e jω ) dω. 4. An 8 point sequence is given by x(n) = {2,2,2,2,1,1,1,1}. Compute 8 point DFT of x(n) by

Recursive, Infinite Impulse Response (IIR) Digital Filters:

Signals & Systems Handout #4

EE482: Digital Signal Processing Applications

Voiced Speech. Unvoiced Speech

Signal Analysis, Systems, Transforms

! Introduction. ! Discrete Time Signals & Systems. ! Z-Transform. ! Inverse Z-Transform. ! Sampling of Continuous Time Signals

Advanced Training Course on FPGA Design and VHDL for Hardware Simulation and Synthesis

Digital Signal Processing, Homework 1, Spring 2013, Prof. C.D. Chung

Digital Filter Structures. Basic IIR Digital Filter Structures. of an LTI digital filter is given by the convolution sum or, by the linear constant

Lecture 7 - IIR Filters

1. Z-transform: Initial value theorem for causal signal. = u(0) + u(1)z 1 + u(2)z 2 +

3.1. Determine the z-transform, including the region of convergence, for each of the following sequences: N, N::: n.

Lecture 8 - IIR Filters (II)

Transcription:

1 VI. Z Transform Ch 24 Use: Analysis of systems, simple convolution, shorthand for e jw, stability. A. Definition: X(z) = x(n) z z - transforms Motivation easier to write Or Note if X(z) = Z {x(n)} z = e +jw, X(z) = x(n) e -jwn = F { x(n) } Can think of z as shorthand symbol for e +jw. Remember that z can equal r e +jw where r is any real number.

2 B. Convergence of z transform. Can converge even if Fourier transform doesn t. Convergence related to stability (absolute summability) Ex F{u(n)} = u(n) e -jwn is useless not absolutely summable, u(n) e -jwn = 1 = But if r >1 z transform u(n) (re +jw ) -n r -n = 1 ( 1- (1/r) )

3 1/r < 1 so converges For z = r > 1 Region of convergence (R.O.C.) drawn below. (1) Rational Functions in z common (usually closed form for transform of infinite length sequence)

4 Z( a n u(n) ) = a n u (n) z -n = 1 1- az -1 = z z - a if az -1 < 1 a < z Rational function poles and zeroes:

5 b(n) z -n B(z) H(z) = = = a(n) z -n A(z) K (1-z 1 z -1 ) (1-z 2 z -1 ) (1-z 3 z -1 ).. (1-z M z -1 ) (1-p 1 z -1 ) (1-p 2 z -1 ).. (1-p N z -1 ) = K Π (1-z i z -1 ) Π (1-p i z -1 ) H(z i ) = 0 for zeroes z i H(p i ) = for poles p i Theorem: If h(n) is real, the finite poles and zeroes of H(z) must be real or come in complex conjugate pairs. Partial Proof: (1) Note that: b( n ) = K n i= 1 z k ( i, n, j) where k(i,n,j) = index of the ith root in the jth term of b(n)

(2) h(n) real means b(n) and a(n) are real (3) Considering b(n), b(0) = K, so K is real (4) b(1) = -K( z 1 +z 2 + z 3 +.. z M ) = real so the imaginary parts of the zeroes sum to 0. One way this can happen is when complex roots come in conjugate pairs, and the other roots, if any, are real. (5) b(m) = ±K( z 1 z 2 z 3.. z M ) = real. One way this can happen is when complex roots come in conjugate pairs, and the other roots, if any, are real. 6

7 (2) Finite length sequences X(z) = x(n) z -n Will converge in some region if x(n) < Case1 n 1 = n 2 = 0 x(0)z -0 converges for entire z-plane. Case 2 n 2 > 0 x(n) z -n = x(n)/ z n In trouble if z = 0 so z 0 if n 2 > 0. Case3 n 1 < 0 have terms x(-1)z, x(-2)z 2 z if n 1 < 0.

8 (3) Right Sided Sequences X(z) = x(n)z -n + x(n)z -n x(n) = 0 for n < n 1 X(z) = x(n)z -n Ex x(n) = a n u(n) Converges for exterior of circle.

9 Large z makes z -1 small. If converges for z 1, x(n)z -n < Converges for z > R x _ except for z = if n 1 < 0. Right sided sequences have z-transforms which converge for exterior of a circle. Iff z-transform converges for exterior of a circle, it is right sided in time domain. (4) Left Sided Sequences x(n) = 0 for n > n 2

10 X(z) = x(n) z -n = x(n) z -n + x(n) z -n Converges if z < some value. Therefore converges for interior of circle except z=0 if n 2 > 0. x(n) z -n Ex x(n) = -(b n )u(-n-1) X(z) = - b n z -n = 1 - b -n z n z < b Draw region of convergence. Note no poles in the R.O.C.

11 (5) Two sided Sequences Extends from - to X(z) = x(n)z -n = x(n)z -n + x(n)z -n First converges for z large (exterior of circle z = R x _,) Second converges for z small (interior for circle R x + ) Common region of convergence is an annulus: R x _ < z < R x + and and z- transform exists.

12 Ex Find X(z) and its R.O.C. for x(n) = a n u(n) + b n u(-n) C. Inverse z -Transforms (1) Cauchy integral theorem leads to counterclockwise countour in region of convergence. x(n) = (1/2πj) X(z) z n-1 dz

13 We won t use this. Note; looks complicated but leads to partial fraction expansion often. (2) Power Series If we have X(z) = x(n)z -n and we have x(n), inverse transform by inspection. H(z) = 0.2z + 1 + 0.3 z -1 h(-1) = 0.2, h(0) = 1, h(1) = 0.3 (3) Partial Fraction Expansion for Rational Case H(z) = B(z) A(z)

14 = b(n)z -n a(n)z -n If we know poles p(n) of H(z), rewrite H(z) as K b(n)z -n Π (1- p(n)z -1 ) A 1 A 2 A N = + + (1- p(1)z -1 ) (1- p(2)z -1 ) (1- p(n)z -1 ) H(z) = A k (p(k)) n z -n h(n) = A k (p(k)) n u(n)

15 Ex Expansion of right sided sequence. x(n) = (1/2) n u(n) + (1/3) n u(n) 1 1 X(z) = ----------- + ------------ 1-1/2 z -1 1 1/3 z -1 2 5/6 z -1 = ------------------------- 1 5/6 z -1 + 1/6 z -2 A B = ----------- + ------------ 1-1/2 z -1 1 1/3 z -1 A = B =

16 (4) Long Division H(z) = B(z)/A(z) Rules (1) For a causal solution, put powers of z in descending order for both B(z) and A(z). (2) For an anti-causal solution, put powers of z in ascending order for both B(z) and A(z).

17 Comments (1) Solutions from the long division approach are identical to those from the partial fraction approach (2) The long division approach leads to a difference equation. Ex. H(z) = B(z)/A(z) = Y(z)/X(z). If x(n) = δ(n), X(z) = 1 and Y(z) = H(z). Normalize the coefficients so that a(0) = 1. Let A(z) = 1 + A (z). Then B(z) Y(z) = 1 + A (z) X(z) Y(z) = B(z)X(z) A (z)y(z). With x(n) = δ(n) and h(n) = y(n), h(n) = b(k) δ(n-k) - a(k)h(n-k)

D. Properties and Uses of Z-Transforms (1) Shift of a Sequence (most important basic use of z-transform). X(z) = x(0) + x(1)z -1 +x(2) z -2 z X(z) = x(0) z + x(1) +x(2) z -1 = Y(z) Z -1 (z X(z)) = y(n) = x(n+1) all n or Z -1 (z k X(z)) = x(n+k) all n, shift to left. Z -1 (z -k X(z)) = x(n-k) all n, shift to right. (2) Multiplication by Exponential Sequence y(n) = a n x(n) 18 Y(z) = a n x(n)z -n = x(n) (a -1 z) -n

19 = X(z) = X(a -1 z) If X(z 1 ) is special in original transform (poles, zero, etc) X(a -1 z 2 ) is special now. z 1 = a -1 z 2 or z 2 = a z 1 Expand or show z plane. (3) Convolution of Sequences (Most important Property) y(n) = h(k) x(n-k)

20 Z-transform both sides Y(z) = h(k) x(n-k) z -n z -n = z -k z -(n-k) Y(z) = h(k) z -k x(n-k) z -(n-k) Y(z) = H(z) X(z) ( Note: h(f(n)) g(z) -f(n) = X(g(z)) ) This implies that convolution is equivalent to multiplying polynomials in z -1 Ex (3z + 1 + z -1 ) (2 + 3z -1 + z -2 ) h(n) = 3 δ(n+1) + δ(n) + δ(n-1) x(n) = 2 δ(n) + 3 δ(n-1) + δ(n-2)

21 Y(z) = 6z + 11 + 8z -1 + 4z -2 + z -3 y(n) = 6δ(n+1) + 11δ(n) + 8δ(n-1) + 4δ(n-2) + δ(n-3) may be easier than straight convolution method. Ex h(n) = a n u(n), x(n) = b n u(n) H(z) = 1 1- az -1 X(z) = 1 1- bz -1

22 Y(z) = 1 (1-az -1 ) (1-bz -1 ) A B = + (1-az -1 ) (1-bz -1 ) 1 A = 1-bz -1 = 1 1-(b/a)

23 = a (a-b) 1 B = 1-az -1 = 1 1-(a/b) = b (b-a)

a b y(n) = [ ( ) a n + ( ) b n ]u(n) a - b b - a 24 Sometimes easier than time domain method. (4) System Function or Transfer Function. In Laplace Transforms, transfer function is Output(s)/intput(s) Also C(s) = M(s) R(s) Given M and R get C Also c(t) = m(t) r(t-t) dt Same in z transform and Fourier transform.

25 For Linear Shift invariant system, System function or transfer function is z transform of impulse response. y(n) = h(k) x(n-k) Y(z) = H(z) X(z) If z = 1 or z = e jw, we get frequency response of system. Ex If H(z) = 2z 1 + 1 + 2z -1 H(e +jw ) = 2e jw + 1 + 2 e -jw = 1 + 4cos(w)

26 Ex : FIR Filters h(n) = 0 n < M or n > N H(z) = h(n)z -n is transfer function. Ex : IIR Filters a(k) y(n-k) = b(r) x(n-r) for all n z-transform eqn. A(z) Y(z) = B(z) X(z) Y(z) X(z) = B(z) A(z)

27 = b(r) z -r a(k) z -k = K Π (1 z r z -1 ) Π (1 p k z -1 ) Pole Zero pattern in z plane specifies the type of causality and stability for the sequence. If poles are inside unit circle, H(z) converges for z outside unit circle and have stable causal filter. 1 1- az -1 az -1 = 1 z = a

28 1 = a n z -n iff az -1 < 1 1- az -1 for z > a Z-Transform Examples Ex Find the transform of n x(n) / z ( x (n)z -n ) = - n x(n) z -(n+1) X'(z) = - z -1 n x(n) z -n = - z -1 n x(n) z -n

29 Therefore, Z { n x(n) } = -z X'(z) Ex Find the z transform of n a n u(n) 1 z z { a n u(n) } = = = X(z) (1-az -1 ) z - a X'(z) = [ ((z-a) z) / (z-a) 2 ] = -a (z-a) 2 -z X'(z) = a z (z-a) 2 Region of convergence? Same as for X(z)

30 Ex Find z-transform of sequence x(n) = n for 0 n N-1 N for n N Straightforward Way x(n) = n u(n) (n-n) u(n-n) r(n) r(n-n) z(n y(n) ) = -z d/dz Y(z) y(n) = u(n) Here, U(z) = z -n = 1 1- z -1

31 Y'(z) = - z -2 (1-z -1 ) 2 -zy'(z) = z -1 (1-z -1 ) 2 = R(z) X(z) = R(z) [ 1 z -N ] = z -1 z -N-1 (1-z -1 ) 2 Other way: x(n) = [u(n) u(n-n)] * u(n-1) X(z) = U(z) [ U(z) z -N U(z) ] z -1

32 = z -1 (1 z -N ) (1-z -1 ) 2 Ex h(n) = 2 n u(n) + (1/3) n u(n) (a) Find H(z) and its R.O.C. (b) Find a stable version of h(n) that has the same H(z), but with a different R.O.C. (c) Find a stable set of difference equations for the filter. (a) (b)

(c) 33

34 Ex Generalize the previous example K h(n) = c i a in u(n) i=1 (a) Find H(z) and its R.O.C. (b) If the 3 rd term, c 3 a 3n u(n) is unstable, find a stable version of it that leads to the same H(z), but with a different R.O.C. (c) How many versions of H(z) are there, which have the same form but with a different R.O.C.? (d) Out of these, how many are stable? (e) Which H(z) corresponds to H(e jw )? (f) Give a parallel form block diagram for this filter. (a)

35 (b) (c) (d) (e) (f)

36 Ex Z { cos(w 1 n) u(n) } Ex Given X(z) = e a z, find x(n)

Ex. An IIR digital filter has the impulse response (.5 n n h(n) = ) u( n) + (.2 ) u(n) (a) Find a stable H(z) in closed form, and its region of convergence. (b) Give a stable set of recursive difference equations for the filter. Use the parallel form. (c) Add the two H(z) terms of part (b ) together to get H(z) with one numerator and one denominator. Give the difference equation corresponding to this new direct form H(z). (d) In the pseudocode below, which is based upon H(z) in part (c), give correct expressions for A, B, C, and D, assuming that x(n) = y(n) =0 for negative n. 37

38 y(0) = A y(1) = B For n = C to N y(n) = D End 1.5z 1 H ( z ) = + (a) 1.5z 1.2 (b) z 1 1 (c) 1 z +.1z H ( z ) = 1.7 z +.1z 1 2 1 2 (d)

E. Three Standard Implementations of H(z) 39 1. Parallel Form H ( z ) = h + i z N /2 1 k k 1 2 k = 1 1+ fk z + gk z y ( n) = h x( n) + i x( n 1) k k k f y ( n 1) g y ( n 2) k k k k N /2 y( n) = y ( n) k = 1 k

40 2. Cascade Form H ( z ) = c + d z + e z N /2 1 2 k k k 1 2 k = 1 1+ fk z + gk z y ( n) = c x ( n) + d x ( n 1) + e x ( n 2) k k k k k k k f y ( n 1) g y ( n 2) k k k k x ( n) = x( n), 1 x ( n) = y ( n) k k 1 y( n) = y ( n) N /2

41 3. Direct Form Multiplying out the Cascade form or adding up the parallel form, we get a single transfer function: H ( z ) = a(0) = 1 M r= 0 N k = 0 M b( r) z a( k) z r k = Y ( z) X ( z) y( n) = b( r) x( n r) a( k) y( n k) r= 0 k = 1 N