Yes, there is often a constant scaling factor, such as a 1/(2*Pi), or square root of that, as I vaguely recall. Each FFT package seems to use its own favorite scaling.

You must look up the detailed description of the inverse FFT package that you are using. It should place the equations for the forward and the inverse FFT side by side in the description, from which you can see exactly how the scaling is being handled. You generally have to check for every different FFT routine you use (MatLab, MathCad, R, Numerical Recipes, etc...)

Dear Kapil,

Use a normalization sqrt(N) before IDFT, where N is the IDFT size. An inverse normalization 1/sqrt(N) should be used when doing DFT.

Best Regards,

Vinay Trivedi

IIT Patna