Re: Maximum resolution in Fourier
- From: "John Barrett" <john.ae5et@xxxxxxxxxxx>
- Date: Sun, 31 Aug 2008 02:02:06 -0500
|
| Resolution of a DFT is fs/N where fs is sample rate and N is number of
| samples.
| Easy - work it out from that. ie you need a longer time to get more
| resolution.
| You could also reduce the sampling rate if you don't need a large
| bandwidth.
|
I was thinking about this -- if he does a multirate multi stage LPF and
decimate to get down to 64 samples per second, he can end up with a 0-28hz
window, and pull 1024 bins for every 16 seconds of data (all this assuming
2^N FFTs) and achieve 0.0625hz resolution....Wrap up the data through a
circular buffer and you can still have updates as often as you want, but you
wont see some of the peaks rising out of the noise until long after they
actually occured -- which may be a problem if he is trying to correlate with
real world events.
But the requirements stated that the signals to be detected are transient in
the 0-4 second range, so he is either going to have give up resolution, or
give up response time.. there just isnt a choice other than that with the
window of interest specified.
.
- References:
- Maximum resolution in Fourier
- From: Marc Kroeks
- Re: Maximum resolution in Fourier
- From: falderals
- Maximum resolution in Fourier
- Prev by Date: Re: Maximum resolution in Fourier
- Next by Date: Re: Maximum resolution in Fourier
- Previous by thread: Re: Maximum resolution in Fourier
- Next by thread: Macro Assembler or C compiler for the TMS320C10
- Index(es):
Relevant Pages
|
