Path: news.mathworks.com!newsfeed-00.mathworks.com!news.kjsl.com!newsfeed.stanford.edu!newsfeed.news.ucla.edu!nrc-news.nrc.ca!newsflash.concordia.ca!canopus.cc.umanitoba.ca!not-for-mail
From: roberson@ibd.nrc-cnrc.gc.ca (Walter Roberson)
Newsgroups: comp.soft-sys.matlab
Subject: Re: fft, frequecy of a signal
Date: Tue, 17 Jun 2008 13:25:17 +0000 (UTC)
Organization: National Research Council Canada - Conseil national de rechereches Canada
Lines: 43
Message-ID: <g38dvt$cmh$1@canopus.cc.umanitoba.ca>
References: <g355te$q0r$1@fred.mathworks.com> <g37quk$et6$1@fred.mathworks.com> <g387sh$48v$1@canopus.cc.umanitoba.ca> <g38daq$671$1@fred.mathworks.com>
NNTP-Posting-Host: origin.ibd.nrc.ca
X-Trace: canopus.cc.umanitoba.ca 1213709117 13009 192.70.172.160 (17 Jun 2008 13:25:17 GMT)
X-Complaints-To: abuse@cc.umanitoba.ca
NNTP-Posting-Date: Tue, 17 Jun 2008 13:25:17 +0000 (UTC)
Originator: roberson@ibd.nrc-cnrc.gc.ca (Walter Roberson)
Xref: news.mathworks.com comp.soft-sys.matlab:474226



In article <g38daq$671$1@fred.mathworks.com>,
Bryan  <bdg146.removeThis@gmail.com> wrote:
>roberson@ibd.nrc-cnrc.gc.ca (Walter Roberson) wrote in 
>message <g387sh$48v$1@canopus.cc.umanitoba.ca>...
>> Standard definition of frequency: samples per second.

>Wouldn't this depend on the 'type' of frequency they are 
>looking for?  Taking the number of samples and dividing by 
>the total time would give them the sampling frequency, but 
>not the frequency of the analog signal that was sampled.  

True.

>For instance, you can have a 1 Hz signal that is sampled 
>at 1 MHz.  (# of samples) / (total time) would equal 1 
>MHz, but that isn't the frequency of the signal, which is 
>1 Hz.

>Does that make any sense or am I totally off-base here?

No, you are correct, but it should be kept in mind that
when an analog signal is sampled, there will *always* be
multiple frequencies involved, so one would not talk about
"the frequency" (which presumes there is only a single frequency.)

No matter how pure the analog frequency is, sampling it at any
finite sample rate with any finite number of bits per sample
will -necessarily- quantize the signal, converting it into
a finite sequence of square waves. When you fft the signal,
you can always see the effect of the square wave approximation.
The purer the original tone and the lower the noise and the more
accurate (bits per sample) you get the readings, the less noticable
will be the affect on the fft compared to the ideal fft.

Usually the primary visible artifact in the fft of the discretized
signal will be a phase shift, but there will also be infinite
harmonics (as high up as the sampling frequency can distinguish
between them.)
-- 
  "Ignorance has been our king... he sits unchallenged on the throne of
  Man. His dynasty is age-old. His right to rule is now considered
  legitimate. Past sages have affirmed it. They did nothing to unseat
  him."                                       -- Walter M Miller, Jr