DSPRelated.com
Forums

How can I decide the blackman window's points ?

Started by stephenduan4513 June 1, 2006
Hi:all
I am trying to measure the SNR of a second order sigma-delta ADC. I want
to just do a FFT analysis for the output of the quantizer. So, I collect
65536 points for FFT. I was told that I should use a Blackman window
here.But how many points of the blackman window should I use?
Here are some specs:
sample frequency:12.5MHz
Baseband frequency:20KHz
Input signal freq:1000Hz
Is there any idea how I can decide the points of that window?
Or could anyone tell me the relationship between the input freq,sampling
freq, FFT points and windows length?
Thanks

stephenduan4513 wrote:

> Hi:all > I am trying to measure the SNR of a second order sigma-delta ADC. I want > to just do a FFT analysis for the output of the quantizer. So, I collect > 65536 points for FFT. I was told that I should use a Blackman window > here.But how many points of the blackman window should I use? > Here are some specs: > sample frequency:12.5MHz > Baseband frequency:20KHz > Input signal freq:1000Hz > Is there any idea how I can decide the points of that window? > Or could anyone tell me the relationship between the input freq,sampling > freq, FFT points and windows length? > Thanks
You want to window the entire sample sequence fed to the DFT. The Blackman window will taper the signal to zero at the ends, effectively reducing the number of available samples by two. That doesn't matter much, but you can center a 65538-point window on your 65536 samples. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������