Page 218 - DSP Integrated Circuits
P. 218

5.5 Scaling of Signal Levels                                         203


             ues in the node of interest. This method of computing the L2-norm will be used in
             Example 5.6.

             Loo-Norm
             The Loo-norm corresponds to the maximum value of the magnitude function.
             Hence, we have





                 The values of Lp-norms are illustrated in Figure 5.15 for the second-order sec-
             tion having the transfer function




























                             Figure 5.15 Lp-norms for a second-order section



             5.5.4 Scaling of Wide-Band Signals

             The first step in scaling a filter is to determine the appropriate L^-norm that char-
             acterizes the input signal. Generally, we distinguish between wide-band and nar-
             row-band input signals. However, it is not obvious which type of input signal
             should be used for scaling; the "normal" class of input signals, including noise, or
             some infrequently occurring disturbance. The answer to this question depends, of
             course, on the application.
                 Jackson [17] has derived the following bound on the variance of the signal in
             the critical node v:




                       a)T                                                  0
             where F v(eJ ) is the frequency response to the critical node v and S^eJ ^) is the
             power spectrum of the input signal.
   213   214   215   216   217   218   219   220   221   222   223