Fast Fourier Transforms on Motorola's Digital Signal Processors
Fast Fourier Transforms on Motorola's Digital Signal Processors
Fast Fourier Transforms on Motorola's Digital Signal Processors
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
“Data<br />
acquisiti<strong>on</strong> <strong>on</strong><br />
the DSP96002 is<br />
truly parallel<br />
with CPU<br />
instructi<strong>on</strong><br />
executi<strong>on</strong>.”<br />
A real-valued input FFT is a special case of the complex<br />
FFT where all imaginary comp<strong>on</strong>ents in the input<br />
are zero. Under this c<strong>on</strong>diti<strong>on</strong>, input sequence is real,<br />
and the time sequence has a symmetric <str<strong>on</strong>g>Fourier</str<strong>on</strong>g> transform<br />
in the frequency domain. Only half of the<br />
frequency sequence needs to be computed for realvalued<br />
input FFTs or real FFTs. Recall the definiti<strong>on</strong><br />
of the DFT:<br />
X( k)<br />
x()e r<br />
j2πrk ( ) N ⁄ –<br />
=<br />
N– 1<br />
∑ k<br />
r = 0<br />
= 0, 1,<br />
N– 1<br />
∑<br />
If x(r) is real,<br />
X∗( – k)<br />
x()e r<br />
j2πrk ( ) N ⁄<br />
x()e r<br />
j2πrk ( ) N ⁄ –<br />
= = =<br />
r = 0<br />
N– 1<br />
∑<br />
and<br />
N– 1<br />
∑<br />
r = 0<br />
…N–<br />
1<br />
X( k)<br />
X∗( N– k)<br />
x∗()e∗ r<br />
j – ( ) 2πrN k – ( ( ) ) N ⁄<br />
x()e r<br />
j2πrk ( ) N ⁄ –<br />
= = =<br />
r = 0<br />
SECTION 6<br />
Real-Valued Input<br />
FFT Algorithm<br />
N– 1<br />
r = 0<br />
Eqn. 6-1<br />
Eqn. 6-2<br />
Eqn. 6-3<br />
MOTOROLA 6-1<br />
∑<br />
X( k)