1 // designs a complex tap fir filter akin to the hilbert transformer.
3 // The hilbert transformer is classified as a linear phase fir
4 // with allpass magnitude response and 90 degree phase response for
5 // positive frequencies and -90 degrees phase for negative frequencies.
6 // Or, if you prefer, normalized frequencies between .5 and 1 since
7 // negative frequencies don't really have much meaning outside the complex
10 // Normally one would use the hilbert transformer in one leg of a complex
11 // processing block and a compensating delay in the other.
13 // This one differs in the following respects:
14 // It is low pass with a cutoff of .078125
15 // The filter is a lowpass kaiser windowed filter with parameter 3
16 // The phase response is 45 degrees for positive frequencies and -45
17 // for negative frequencies.
18 // The coefficent set is used in one path and the same coefficients
19 // are used time reversed in the other. This results in the net effect
20 // of +/- 90 degrees as in the usual hilbert application.
22 // The coefficient set can be used in the gnuradio frequency translating
23 // fir filter for ssb demodulation.
25 // This isn't as computationally efficient as using the hilbert transformer
26 // and compensating delay but fascinating none the less.
28 // This program is for the scilab language a very powerful free math
29 // package similar to Matlab with infinitely better price/performace.
31 // compute the prototype lowpass fir
32 // length is 255 (odd) for the same symmetry reasons as the hilbert transformer
39 h = wfir( 'lp', len, [10.0/256 0], 'kr', [3 0] );
43 H(1:l2)=H(1:l2)*exp(%i*%pi/4);
45 H(l3:len)=H(l3:len)*exp(-%i*%pi/4);
53 f = file('open','taps')
55 fprintf( f, '%f%+fj', j(i), k(i) )