0% found this document useful (0 votes)
110 views22 pages

FIR - Kaiser Design PDF

Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
110 views22 pages

FIR - Kaiser Design PDF

Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 22
Section 7.5 Design of FIR Filters by Windowing 533 7.5 DESIGN OF FIR FILTERS BY WINDOWING As discussed in Section 7.2, commonly used techniques for the design of IIR filters have evolved from applying transformations of continuous-time IIR systems into discrete- time IIR systems. In contrast, the design techniques for FIR filters are based on directly approximating the desired frequency response or impulse response of the discrete-time system, ‘The simplest method of FIR filter design is called the window method. Thismethod generally begins with an ideal desired frequency response that can be represented as Hale!” DY hatnte", (7.53) where haln] is the corresponding impulse response sequence, which can be expressed in terms of Hu(e!) as Hale!*)el" dw. (7.54) hain} Many idealized systems are defined by piecewise-constant or piecewise-smooth fre- quency responses with discontinuities at the boundaries between bands. As a result, these systems have impulse responses that are noncausal and infinitely long. The most, straightforward approach to obtaining an FIR approximation to such systems is to trun- cate the ideal impulse response through the process referred to as windowing. Equa- tion (7.53) can be thought of as a Fourier series representation of the periodic frequency response H,(e!®), with the sequence hal] playing the role of the Fourier coefficients. Thus, the approximation of an ideal filter by truncation of the ideal impulse response is identical to the issue of the convergence of Fourier series, a subject that has received a great deal of study. A particularly important concept from this theory is the Gibbs phenomenon, which was discussed in Example 2.18. In the following discussion, we will see how this effect of nonuniform convergence manifests itself in the design of FIR filters. A particularly simple way to obtain a causal FIR filter from fg{n] is to truncate h(n}, i.c., to define a new system with impulse response A{7] given by4 hain], O 1 — 8, The stopband cutoff frequency a is defined to be the lowest frequency such that | (e!)| <6. Therefore, the transition region has width Aw = 0, — wp (7.73) for the lowpass filter approximation. Defining A= —2Wlogio 5, (7.74) Kaiser determined empirically that the value of 8 needed to achieve a specified value of A is given by 0.1102(A ~8.7), A> 50, B= 4 0.5842(A — 21) + 0.07886(A — 21), 21< A < 50, (7.75) 0.0, A<2L (Recall that the case 8 = 0 is the rectangular window for which A = 21.) Furthermore, Kaiser found that to achieve prescribed values of A and Aw, M must satisfy __A-8 ~ 2285Aw" Equation (7.76) predicts M to within +2 over a wide range of values of Aw and A. Thus, with these formulas, the Kaiser window design method requires almost no iteration or trial and error. The examples in Section 7.6 outline and illustrate the procedure. M (7.76) Amplitude Samples 0 02m O4z 0.6 08 = Radian frequency () (by M=10 Radian frequency (w) to) Figure 7.32 (a) Kaiser windows for 6 = 0, 3, and 6 and M = 20. (b) Fourier transforms corresponding to windows in (a). (c) Fourier transforms of Kaiser win- dows with 6 = 6 and M = 10, 20, and 40, Chapter7 Filter Design Techniques Relationship of the Kaiser Window to Other Windows The basic principle of the window design method is to truncate the ideal impulse re sponse with a finite-length window such as one of those discussed in this section. The corresponding effect in the frequency domain is that the ideal frequency response is convolved with the Fourier transform of the window. If the ideal filter is a lowpass filter, the discontinuity in its frequency response is smeared as the main lobe of the Fourier transform of the window moves across the discontinuity in the convolution process. To a first approximation, the width of the resulting transition band is determined by the width of the main lobe of the Fourier transform of the window, and the passband and stopband ripples are determined by the side lobes of the Fourier transform of the window. Because the passband and stopband ripples are produced by integration of the symmetric window side lobes, the ripples in the passband and the stopband are approx- imately the same. Furthermore, to a very good approximation, the maximum passband and stopband deviations are not dependent on M and can be changed only by changing the shape of the window used. This is illustrated by Kaiser's formula, Eq. (7.75). for the window shape parameter, which is independent of M. The last two columns of Table 7.2 compare the Kaiser window with the windows of Eqs. (7.60). The fifth column gives the Kaiser window shape parameter (8) that yields the same peak approximation error (8) as the window indicated in the first column. The sixth column shows the corresponding transition width [from Eq. (7.76)] for filters designed with the Kaiser window. This for- mula would be a much better predictor of the transition width for the other windows than would the main-lobe width given in the third column of the table. In Figure 7.33 is shown a comparison of maximum approximation error versus transition width for the various fixed windows and the Kaiser window for different Approximation error vs. Transition width [* = fixed windows, o = Kaiser (B = integer)] (I -20+ * Baniet 30+ | 40+ 2D, Hanning 50 t . KaiserS"O¢ Hamming 0 Approximation error (4B) OSe Figure 7.33 Comparison of fixed windows with Kaiser windows ina lowpass filter design application (M = 32 and ae = 7/2). (Note thatthe designation "Kaiser 6” means Kaiser window with 6 = 6, etc.) Section 7.6 Examples of FIR Filter Design by the Kaiser Window Method 545, values of . The dashed line obtained from Eq. (7.76), shows that Kaiser's formula is an accurate representation of approximation error as a function of transition width for the Kaiser window. 7.6 EXAMPLES OF FIR FILTER DESIGN BY THE KAISER WINDOW METHOD In this section, we give several examples that illustrate the use of the Kaiser window to obtain FIR approximations to several filter types including lowpass filters. These examples also serve to point out some important properties of FIR systems. 7.6.1 Lowpass Filter With the use of the design formulas for the Kaiscr window, it is straightforward to design an FIR lowpass filter to meet prescribed specifications. The procedure is as follows: 1. First, the specifications must be established. This means selecting the desired w» and w, and the maximum tolerable approximation error. For window design, the resulting filter will have the same peak error 4 in both the passband and the stopband. For this example, we use the same specifications as in Example 7.5, wp = 0.47, ws = 0.6, 5; = 0.01, and 62 = 0.001. Since filters designed by the window method inherently have 5; = 5, we must set = 0.001. 2. The cutoff frequency of the underlying ideal lowpass filter must be found. Owing to the symmetry of the approximation at the discontinuity of Hy(e/”), we would set og = OP TM on 2 3. To determine the parameters of the Kaiser window, we first compute Aw = 0; wp =0.2n, A = —20logig5 = 60. ‘We substitute these two quantities into Eqs (7.75) and (7.76) to obtain the required values of B and M. For this example the formulas predict, B=5.653, M=37. 4, The impulse response of the filter is computed using Eqs, (7.71) and (7.72). We obtain = 0.57. sinwe(n~a) | folB(1 = {en ~a)/aP)*] hin) = |x =a) bP 0, otherwise, » O

You might also like