How to get a scalle factor to compute an IFFT of size 2048 without any overflow
Hi everyone, I allow myself to contact you if anyone among you can help me. I need to simulate xilinx IFFT 9.0 of size 2048. My problem is indeed I can not understand how to find the scaling factor in order not to fall on an overflow and to have incorrect results. I can configure my the xilinx IP IFFT as unscalled but I need that my output results in 32 bits(16 bits real and 16 imaginary). I am very grateful if any person can help. The documentation of xilinx is not clear in this subject.
Scaling values are to be obtained by System simulation. Other than unscaled and scaled modes, the XFFT IP has a mode called Block floating point (BFP) mode. Run the XFFT in this mode, and the scaling factor will be determined by the core itself. The output width will also be same as input width.
The scaling factor used by the IP is output as an exponent on the status interface which indicates total no.of bits shifted to right in the FFT result. The output will not overflow in this case.