563489578934
74 Signals and Spectra Chap. 2 Polar Fourier Series The quadrature Fourier series, Eq. (2–95), may be rearranged and written in a polar (amplitudephase) form. The polar form is where w(t) is real and w(t) = D 0 + n=q a n=1 D n cos(nv 0 t + w n ) (2–103) a n = e D 0, n = 0 (2–104) D n cos w n , n Ú 0 b n = -D n sin w n , n Ú 1 (2–105) The latter two equations may be inverted, and we obtain D n = e a 0, n = 0 (2–106) 2a 2 n + b 2 n, n Ú 1 f = e c 0, n = 0 2ƒc n ƒ, n Ú 1 and w n = -tan -1 b n £ ≥ = lc (2–107) a n , n Ú 1 n where the angle operator is defined by l[# ] = tan -1 a Im[ # ] b (2–108) Re[# ] It should be clear from the context whether l denotes the angle operator or the angle itself. For example, l90° denotes an angle of 90°, but l[1 + j2] denotes the angle operator and is equal to 63.4° when evaluated. The equivalence between the Fourier series coefficients is demonstrated geometrically in Fig. 2–11. It is seen that, in general, when a physical (real) waveform w(t) is represented by a Fourier series, c n is a complex number with a real part x n and an imaginary part y n (which are both real numbers), and consequently, a n , b n , D n , and wn are real numbers. In addition, D n is a nonnegative number for n Ú 1. Furthermore, all of these coefficients describe the amount of frequency component contained in the signal at the frequency of nf 0 Hz. In practice, the Fourier series (FS) is often truncated to a finite number of terms. For example, 5 or 10 harmonics might be used to approximate the FS series for a square wave. Thus, an important question arises: For the finite series, are the optimum values for the series coefficients the same as those for the corresponding terms in the infinite series, or should the coefficients for the finite series be adjusted to some new values to give the best finite-series approximation? The answer is that the optimum values for the coefficients of the truncated FS are the same as those for the corresponding terms in the nontruncated FS. † As we have seen, the complex, quadrature, and polar forms of the Fourier series are all equivalent, but the question is, Which is the best form to use? The answer is that it † For a proof of this statement, see [Couch, 1995].
- Page 146: Sec. 2-2 Fourier Transform and Spec
- Page 150: Sec. 2-2 Fourier Transform and Spec
- Page 154: Sec. 2-2 Fourier Transform and Spec
- Page 158: Sec. 2-2 Fourier Transform and Spec
- Page 162: ( ( Sec. 2-2 Fourier Transform and
- Page 166: ` Sec. 2-2 Fourier Transform and Sp
- Page 170: Sec. 2-2 Fourier Transform and Spec
- Page 174: Sec. 2-3 Power Spectral Density and
- Page 178: Sec. 2-3 Power Spectral Density and
- Page 182: Sec. 2-4 Orthogonal Series Represen
- Page 186: Sec. 2-4 Orthogonal Series Represen
- Page 190: Sec. 2-5 Fourier Series 71 2-5 FOUR
- Page 194: Sec. 2-5 Fourier Series 73 where th
- Page 200: 76 Signals and Spectra Chap. 2 PROO
- Page 204: 78 Signals and Spectra Chap. 2 w(t)
- Page 208: 80 Signals and Spectra Chap. 2 PROO
- Page 212: 82 Signals and Spectra Chap. 2 Inpu
- Page 216: 84 Signals and Spectra Chap. 2 is t
- Page 220: 86 Signals and Spectra Chap. 2 From
- Page 224: 88 Signals and Spectra Chap. 2 0 dB
- Page 228: 90 Signals and Spectra Chap. 2 Band
- Page 232: 92 Signals and Spectra Chap. 2 f s
- Page 236: 94 Signals and Spectra Chap. 2 w s
- Page 240: 96 Signals and Spectra Chap. 2 W(f)
- Page 244: 98 Signals and Spectra Chap. 2 MATL
74<br />
Signals and Spectra Chap. 2<br />
Polar Fourier Series<br />
The quadrature Fourier series, Eq. (2–95), may be rearranged and written in a polar (amplitudephase)<br />
form. The polar form is<br />
where w(t) is real and<br />
w(t) = D 0 +<br />
n=q<br />
a<br />
n=1<br />
D n cos(nv 0 t + w n )<br />
(2–103)<br />
a n = e D 0, n = 0<br />
(2–104)<br />
D n cos w n , n Ú 0<br />
b n = -D n sin w n , n Ú 1<br />
(2–105)<br />
The latter two equations may be inverted, and we obtain<br />
D n = e a 0, n = 0<br />
(2–106)<br />
2a 2 n + b 2 n, n Ú 1 f = e c 0, n = 0<br />
2ƒc n ƒ, n Ú 1<br />
and<br />
w n = -tan -1 b n<br />
£ ≥ = lc (2–107)<br />
a n , n Ú 1<br />
n<br />
where the angle operator is defined by<br />
l[# ] = tan -1 a Im[ # ]<br />
b<br />
(2–108)<br />
Re[# ]<br />
It should be clear from the context whether l denotes the angle operator or the angle<br />
itself. For example, l90° denotes an angle of 90°, but l[1 + j2] denotes the angle operator<br />
and is equal to 63.4° when evaluated.<br />
The equivalence between the Fourier series coefficients is demonstrated geometrically<br />
in Fig. 2–11. It is seen that, in general, when a physical (real) waveform w(t) is represented by<br />
a Fourier series, c n is a complex number with a real part x n and an imaginary part y n (which<br />
are both real numbers), and consequently, a n , b n , D n , and wn are real numbers. In addition, D n<br />
is a nonnegative number for n Ú 1. Furthermore, all of these coefficients describe the amount<br />
of frequency component contained in the signal at the frequency of nf 0 Hz.<br />
In practice, the Fourier series (FS) is often truncated to a finite number of terms. For<br />
example, 5 or 10 harmonics might be used to approximate the FS series for a square wave.<br />
Thus, an important question arises: For the finite series, are the optimum values for the series<br />
coefficients the same as those for the corresponding terms in the infinite series, or should the<br />
coefficients for the finite series be adjusted to some new values to give the best finite-series<br />
approximation? The answer is that the optimum values for the coefficients of the truncated<br />
FS are the same as those for the corresponding terms in the nontruncated FS. †<br />
As we have seen, the complex, quadrature, and polar forms of the Fourier series are<br />
all equivalent, but the question is, Which is the best form to use? The answer is that it<br />
† For a proof of this statement, see [Couch, 1995].