Hi, I have two questions about the FFTC coprocessor:
1 - Is it possible to use float values instead of integer values? I am working with values between +-1. Which would be the best way of dealing with this? Tranforming this +-1 interval into +-2^15?
2 - About the scaling factor, the User Guide says it can be 0, 1, 2 or 3. However, in some cases I get 65535, 65534, 65533; which seems to be a representation of -1, -2 and -3 in Two's complement. Actually when I get these values of exponent, the FFT result appears to be 2x, 4x or 8x the result obtained in Matlab. I only get "negative" scaling factors when the output results are small (close to 0). Is this done to avoid the loss of precission when working with small values? The User Guide does not say anything regarding this.
Thank you.