Forum Discussion
Altera_Forum
Honored Contributor
16 years agoIgnore my above remarks since your fft output is 16 bits.
I understand from your equations that you are converting the 16 bit output into 32 bit bus then doing the two stage shift(by exponent then by 9). Your work-around of 9 bits shift then taking 16 LSBs doesn't make sense to me. To avoid calculation problems of software Let us go back to your simple example: You said for your input sine of 30146 amplitude you get fft output as: Real: 3384 (0x0D38) Imaginary: -14391 (0xC7C9) fair enough. what is your exponent value in above example? It should be -1 so that output scales to 2^ -(-1) = 2 otherwise there is something wrong.