1. Limited time only! Sign up for a free 30min personal tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Nyquist Frequency

  1. Sep 26, 2013 #1
    1. The problem statement, all variables and given/known data

    There seems to be varying answers on what the definition of "Nyquist Frequency" is, between my professor, myself, and various internet sites.

    One group says:

    "The Nyquist Frequency is 1/2 the sampling rate"

    and the other group says:

    "The Nyquist Frequency is twice the signal's bandwidth"

    Both groups agree that your sampling rate must be twice the signal's bandwidth to avoid aliasing, but when a problem simply says "the sampling rate is x, and the signal's bandwidth is y, what is the Nyquist Frequency?" I get different answers depending on which definition I use.

    Which is the correct definition, and how can I argue this to my professor?

    2. Relevant equations



    3. The attempt at a solution

    Sources, group 1:

    http://en.wikipedia.org/wiki/Nyquist_frequency
    http://www.encyclopedia.com/doc/1O13-Nyquistfrequency.html

    Sources, group 2:

    http://www2.egr.uh.edu/~glover/applets/Sampling/Sampling.html
    Google "Nyquist Frequency Definition" and google shows up as group 2.
     
  2. jcsd
  3. Sep 26, 2013 #2

    D H

    User Avatar
    Staff Emeritus
    Science Advisor

    To me there is no right or wrong. Both are right. Words can have multiple meanings, even in a technical context. What's wrong is arguing that only one sense is right, the other wrong.

    Picture yourself as a designer of some digital system. Here you are worried about what the sampling frequency has to be. You find the highest frequency of concern, double it to yield the minimum sampling frequency, and then set the sampling frequency even higher than that for safety. The Nyquist frequency is based on system design parameters, and it is this dictated by design Nyquist frequency that defines the sampling frequency.

    Now picture yourself as an operator of this system. The system was built and deployed long ago, and it's getting a bit on in years. Things are getting a bit rough on the edges. Can you see some noise signal in the telemetered data? If the frequency of that noise is more than half the sampling frequency, no you can't. Now the sampling frequency is set in the design. You can't change it, so now the sampling frequency defines the Nyquist frequency.
     
  4. Sep 26, 2013 #3

    rude man

    User Avatar
    Homework Helper
    Gold Member



    EDIT: there is confusion between Nyquist frequency and Nyquist rate.

    The Nyquist frequency is by definition half the sampling rate, since that frequency and all frequencies below that frequency will not be aliased when sampled.

    The Nyquist rate is the sampling rate at which a given signal will not be aliased.

    The N. rate is not simply twice the max. signal frequency. For example, if a signal contains W Hz lying between mW and (m+1)W Hz, m an integer, i.e. a pass-band signal, the Nyquist rate is 2W, not 2W(m+1).

    I would advise not paying too much attention to the nomenclature and concentrate instead on the theory.

    The sampling rate can be anything. In fact, deliberate undersampling is often done. The answer is that the Nyquist rate is twice the highest frequency in a baseband signal (extending from 0 to f).

    But if the frequency range is limited to f1 to f2 then the Nyquist rate (min. sampling frequency to avoid aliasing) assumes a more complex formula, almost always being < 2f2.
     
    Last edited: Sep 26, 2013
  5. Sep 26, 2013 #4

    rude man

    User Avatar
    Homework Helper
    Gold Member

    ???
    Maybe want to rewrite that.
     
  6. Sep 27, 2013 #5
    I am more than happy to just accept that as the answer and move on, but how did you determine that your answer was the correct choice?
     
  7. Sep 27, 2013 #6

    rude man

    User Avatar
    Homework Helper
    Gold Member

    Whom are you addressing? If it's I the answer is MSEE.
    You can also look it up in wikipedia.
     
  8. Sep 27, 2013 #7

    marcusl

    User Avatar
    Science Advisor
    Gold Member

    In common usage, the Nyquist frequency is the folding frequency, which is half the sampling rate. People are very sloppy about this, however, as you have noted.
    It is interesting to read Nyquist's papers (1924 and 1928) on the subject of signalling, because he never discussed sampling as is presently done in digital systems. His interest was in the frequency spectrum of telegraph signals, which he examined through their Fourier series. Claude Shannon later referred to "Nyquist interval" as the period at the folding frequency (half the sampling frequency). Other authors coined "Nyquist rate" for the sampling frequency, and on it went. I leave Nyquist's name out of my discussions and talk in terms of the sampling frequency, half the sampling frequency, etc. so there is no confusion.
     
  9. Sep 27, 2013 #8

    rude man

    User Avatar
    Homework Helper
    Gold Member

    That is what I urged also. In general, too much emphasis is placed on nomenclature.
     
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?
Draft saved Draft deleted



Similar Discussions: Nyquist Frequency
  1. Nyquist Sampling (Replies: 0)

  2. Nyquist rate (Replies: 1)

Loading...