Re: scaling fixed point fft

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

You didn't mention the radix of each stage.  Divide by 4 is too much if
it is a radix2 butterfly.  If it is radix 2, the gain is at most 2, not
4.  The max output from an FFT occurs when all the input energy falls
into one output bin, which implies the input is a pure sinusoid with a
frequency that is an integer multiple of the sample time.  If the input
is white noise, the output signal energy is going to be spread evenly
among all the output bins, in effect dividing the input signal by N.  As
you can see, this results in a fairly wide dynamic range.  If you know
the nature of your input, you can scale accordingly.  If not, you need
to either carry extra bits or you can use dynamic scaling such as a
block floating point scheme.

Bob wrote:

Quoted text here. Click to load it

--Ray Andraka, P.E.
President, the Andraka Consulting Group, Inc.
We've slightly trimmed the long signature. Click to see the full one.

Site Timeline