I still have to understand how this algoritm will work in situations where the computed coefficients can be out of the range of my fixed point format. Some scaling will be required, yes. I would like to avoid using a 32-bit format for fft coefficient storage.I still have to understand how this algoritm will work in situations where the computed coefficients can be out of the range of my fixed point format. Some scaling will be required, yes. I would like to avoid using a 32-bit format for fft coefficient storag