-1

for Rectangular window if data length is 1K then FFT Length Should be 1K FFT only because ENBW of rectangualr window is 1.0 but for Minimum 4-Sanple BlackmannHarris window is 2.0. so for improving accuracy in terms of either frequency or amplitude which method is preferrded in below 1. for 1K Data i will append 1K zeros and Caluclate 2K FFT or 2. instead of 1K data take 2K data and calclate 2K FFT

or else any other method please calrify ..

kranthi kumar
  • 1
  • As a favor to you Mr. Kumar, I strongly suggest that you use correct capitalization, correct punctuation, and correct spelling in your questions here. That will make your questions much easier for us to understand and make you appear more professional. – Richard Lyons Aug 01 '15 at 01:55

1 Answers1

0

Whether to use a rectangular or another window depends on what you're trying measure in your FFT results. The use of non-rectangular windows is only necessary if you're trying to reduce spectral leakage. Is that your goal? What is your goal?

In general, more "accurate" spectral measurements are made by performing multiple FFTs on multiple (independent) blocks of input time samples. Then you average the magnitudes of those multiple FFT output sequences. For example, let's say your multiple FFT magnitude sequences are |X1(m)|, |X2(m)|, |X3(m)|, and |X4(m)|. Then the averaged spec mag for, say, the output sample m = 7 is [|X1(7)|+|X2(7)|+|X3(7)|+|X4(7)|]/4.

Richard Lyons
  • 4,305
  • 11
  • 24