Forums

How spectral leakage changes the phase spectrum?

Started by bharat pathak February 21, 2008
Hello All,

      I want to understand how spectral leakage changes the
      phase spectrum of a sine wave?

      Also when I apply window (blackman harris 11 Term or
      chebyshev window with 300db attenuation), I see a
      significant amount of reduction in spectral leakage
      in the amplitude part of the spectrum.

      But this windowing also affects the phase part of the
      spectrum. 

      After performing windowing, how reliable is the phase
      information and what information can be deduced from
      the phase plot (post windowing).

Thanks and Regards
Bharat Pathak

Arithos Designs
www.Arithos.com
On 21 Feb, 17:16, "bharat pathak" <bha...@arithos.com> wrote:

> &#2013266080; &#2013266080; &#2013266080; After performing windowing, how reliable is the phase > &#2013266080; &#2013266080; &#2013266080; information and what information can be deduced from > &#2013266080; &#2013266080; &#2013266080; the phase plot (post windowing).
Phase information is inherently elusive, and messing around with phase spectra is a can of worms at the best of times. Just accept the fact that the phase spectrum changes by windowing, and leave it at that. I don't think there are very many applications where there is a justifiable need to reconstruct the 'true' phase spectrum. Rune
On Feb 21, 9:18 am, Rune Allnor <all...@tele.ntnu.no> wrote:
> On 21 Feb, 17:16, "bharat pathak" <bha...@arithos.com> wrote: > > > After performing windowing, how reliable is the phase > > information and what information can be deduced from > > the phase plot (post windowing). > > Phase information is inherently elusive, and messing around > with phase spectra is a can of worms at the best of times. > Just accept the fact that the phase spectrum changes by > windowing, and leave it at that. I don't think there are > very many applications where there is a justifiable need > to reconstruct the 'true' phase spectrum.
Phase information is useful for analyzing data across successive or overlapped DFT frames (see "phase vocoder"). Phase information also appears important for accurate frequency estimation near DC or Fs/2. Relative phase information is required to correctly reconstruct a time or spatial domain waveform (say edges in a photographic image). Communication using phase modulation requires some form of phase detection, elusive situation or not. Different windows do affect the phase information (or ratio between oddness and evenness) differently for different frequencies. Sometimes this affect can be compensated for. IMHO. YMMV. -- rhn A.T nicholson d.0.t C-o-M
On 21 Feb, 18:50, "Ron N." <rhnlo...@yahoo.com> wrote:
>&#2013266080;Communication using phase modulation requires > some form of phase detection, elusive situation or not.
And the relevance of this wrt the phase spectrum is...? Rune
On Feb 21, 10:25 am, Rune Allnor <all...@tele.ntnu.no> wrote:
> On 21 Feb, 18:50, "Ron N." <rhnlo...@yahoo.com> wrote: > > > Communication using phase modulation requires > > some form of phase detection, elusive situation or not. > > And the relevance of this wrt the phase spectrum is...?
The phase spectrum contains information about a signals phase in relationship to the window and/or data aperture.
On 21 Feb, 19:40, "Ron N." <rhnlo...@yahoo.com> wrote:
> On Feb 21, 10:25 am, Rune Allnor <all...@tele.ntnu.no> wrote: > > > On 21 Feb, 18:50, "Ron N." <rhnlo...@yahoo.com> wrote: > > > > Communication using phase modulation requires > > > some form of phase detection, elusive situation or not. > > > And the relevance of this wrt the phase spectrum is...? > > The phase spectrum contains information about a signals > phase in relationship to the window and/or data aperture.
That's a no-brainer. I was wondering what the connection is between phase modulation and the phase spectrum. Since you mention phase modulation in a thread concerning the phase spectrum I assume you know of a connection? Maybe a demodulation technique which is based on the phase spectrum? Rune
On Feb 21, 11:03 am, Rune Allnor <all...@tele.ntnu.no> wrote:
> On 21 Feb, 19:40, "Ron N." <rhnlo...@yahoo.com> wrote: > > > On Feb 21, 10:25 am, Rune Allnor <all...@tele.ntnu.no> wrote: > > > > On 21 Feb, 18:50, "Ron N." <rhnlo...@yahoo.com> wrote: > > > > > Communication using phase modulation requires > > > > some form of phase detection, elusive situation or not. > > > > And the relevance of this wrt the phase spectrum is...? > > > The phase spectrum contains information about a signals > > phase in relationship to the window and/or data aperture. > > That's a no-brainer. I was wondering what the connection > is between phase modulation and the phase spectrum. Since > you mention phase modulation in a thread concerning the > phase spectrum I assume you know of a connection? > Maybe a demodulation technique which is based on > the phase spectrum?
The one I know of is used in an attempt to recover telemetry data from deep space probes. IMHO. YMMV. -- rhn A.T nicholson d.0.t C-o-M
On Feb 21, 8:16 am, "bharat pathak" <bha...@arithos.com> wrote:
> Hello All, > > I want to understand how spectral leakage changes the > phase spectrum of a sine wave?
A sinusoid which is periodic in a DFT aperture will have the same phase relationship to the beginning and end of the aperture and a very closely related (identical modulo pi) phase in the middle. A sinusoid which is not periodic in the aperture will not have the same phase at the beginning and end of the aperture for any of the DFT bin frequencies. The information about the ratio of odd to even function will be spread out far and wide into many bins (in the shape of the transform of the window, called "leakage"). The frequency transform of non-periodic sinusoid can also be interfered with from its own negative frequency image near DC or Fs/2. Thus the phase content of the nearest magnitude peak bin may not be exactly representative of any reference phase to some fixed point in the window, as would be true in the case of an exactly periodic in the aperture sinusoid. Windowing, with something like a raised-cosine, Gaussian, or Blackman-Nuttall, will chop off most of the discontinuity near the window's ends, and pull most of the "leakage" into a more central location, where information (including magnitude and the oddness/evenness ratio) can be more easily and locally interpolated. IMHO. YMMV. -- rhn A.T nicholson d.0.t C-o-M
On 21 Feb, 20:41, "Ron N." <rhnlo...@yahoo.com> wrote:
> On Feb 21, 11:03 am, Rune Allnor <all...@tele.ntnu.no> wrote: > > > > > > > On 21 Feb, 19:40, "Ron N." <rhnlo...@yahoo.com> wrote: > > > > On Feb 21, 10:25 am, Rune Allnor <all...@tele.ntnu.no> wrote: > > > > > On 21 Feb, 18:50, "Ron N." <rhnlo...@yahoo.com> wrote: > > > > > > Communication using phase modulation requires > > > > > some form of phase detection, elusive situation or not. > > > > > And the relevance of this wrt the phase spectrum is...? > > > > The phase spectrum contains information about a signals > > > phase in relationship to the window and/or data aperture. > > > That's a no-brainer. I was wondering what the connection > > is between phase modulation and the phase spectrum. Since > > you mention phase modulation in a thread concerning the > > phase spectrum I assume you know of a connection? > > Maybe a demodulation technique which is based on > > the phase spectrum? > > The one I know of is used in an attempt to recover telemetry > data from deep space probes.
I did not ask if you know of a phase modulation technique which works. I asked if you know of a phase modulation technique which is based on computing the phase spectrum, as opposed to some time-domain democulation technique. Rune
On Feb 22, 1:42 am, Rune Allnor <all...@tele.ntnu.no> wrote:
> On 21 Feb, 20:41, "Ron N." <rhnlo...@yahoo.com> wrote: > > > > > On Feb 21, 11:03 am, Rune Allnor <all...@tele.ntnu.no> wrote: > > > > On 21 Feb, 19:40, "Ron N." <rhnlo...@yahoo.com> wrote: > > > > > On Feb 21, 10:25 am, Rune Allnor <all...@tele.ntnu.no> wrote: > > > > > > On 21 Feb, 18:50, "Ron N." <rhnlo...@yahoo.com> wrote: > > > > > > > Communication using phase modulation requires > > > > > > some form of phase detection, elusive situation or not. > > > > > > And the relevance of this wrt the phase spectrum is...? > > > > > The phase spectrum contains information about a signals > > > > phase in relationship to the window and/or data aperture. > > > > That's a no-brainer. I was wondering what the connection > > > is between phase modulation and the phase spectrum. Since > > > you mention phase modulation in a thread concerning the > > > phase spectrum I assume you know of a connection? > > > Maybe a demodulation technique which is based on > > > the phase spectrum? > > > The one I know of is used in an attempt to recover telemetry > > data from deep space probes. > > I did not ask if you know of a phase modulation technique which > works. I asked if you know of a phase modulation technique > which is based on computing the phase spectrum, as opposed to > some time-domain democulation technique.
You'll have to ask the deep space communications experts. My limited understanding is that the deepest working space probes were designed before digital frequency domain techniques were even computationally feasible.