02.10.2019 Views

UploadFile_6417

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

358 Chapter 7 FIR FILTER DESIGN<br />

Note that the fir2 does not implement the classic optimum frequency<br />

sampling method. By incorporating window design, fir2 has found an alternative<br />

(and somewhat clever) approach to do away with the optimum<br />

transition band values and the associated tables. By densely sampling<br />

values in the entire band, interpolation errors are reduced (but not minimized),<br />

and stopband attenuation is increased to an acceptable level. However,<br />

the basic design is contaminated by the window operation; hence,<br />

the frequency response does not go through the original sampled values. It<br />

is more suitable for designing FIR filters with arbitrary shaped frequency<br />

responses.<br />

The type of frequency sampling filter that we considered is called a<br />

Type-A filter, in which the sampled frequencies are<br />

ω k = 2π<br />

M k, 0 ≤ k ≤ M − 1<br />

There is a second set of uniformly spaced samples given by<br />

ω k = 2π ( )<br />

k + 1 2<br />

, 0 ≤ k ≤ M − 1<br />

M<br />

This is called a Type-B filter, for which a frequency sampling structure is<br />

also available. The expressions for the magnitude response H(e jω ) and the<br />

impulse response h(n) are somewhat more complicated and are available<br />

in Proakis and Manolakis [23]. Their design can also be done in MATLAB<br />

using the approach discussed in this section.<br />

7.5 OPTIMAL EQUIRIPPLE DESIGN TECHNIQUE<br />

The last two techniques—namely, the window design and the frequency<br />

sampling design—were easy to understand and implement. However, they<br />

have some disadvantages. First, we cannot specify the band frequencies<br />

ω p and ω s precisely in the design; that is, we have to accept whatever<br />

values we obtain after the design. Second, we cannot specify both δ 1 and<br />

δ 2 ripple factors simultaneously. Either we have δ 1 = δ 2 in the window<br />

design method, or we can optimize only δ 2 in the frequency sampling<br />

method. Finally, the approximation error—that is, the difference between<br />

the ideal response and the actual response—is not uniformly distributed<br />

over the band intervals. It is higher near the band edges and smaller in<br />

the regions away from band edges. By distributing the error uniformly,<br />

we can obtain a lower-order filter satisfying the same specifications. Fortunately,<br />

a technique exists that can eliminate these three problems. This<br />

technique is somewhat difficult to understand and requires a computer<br />

for its implementation.<br />

Copyright 2010 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).<br />

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!