Richard,

Utter nonsense.

73,
Gene
W4SZ

Richard Clark wrote:
On Sun, 6 Feb 2005 19:35:55 -0000, "Dave" wrote:


you will of course note that i quoted the term 'static' to denote that it
was indeed not static in the infinite sense of mathematics, but in the real
sense of it being a constant value over some measured time period.


Hi Dave,

This is fine as an elaboration that grows beyond the horizon of the
question. But it really serves no purpose but to embroider a glaring
lack of facts.


the other obvious thing you are missing is that DC is a frequency... and it
does come out in the Fourier transform as a frequency of zero with some
finite magnitude for any waveform who's average value is not zero. it does
not even require that the waveform stay at some value forever, a short pulse
going from 0v to 1v and back to 0v will have a DC component in its spectrum
both mathematically and on any meter that can properly respond to it.


After having done a quite rigorous contract with HP deeply involved in
both the strict math and the mechanics of Fourier, I am well versed to
know how not all Fourier Transforms offer equal outcomes - especially
with, or without D.C. as a product.

Pure Fourier mandates a waveform constant throughout all time (from
its inception to its end) and is strictly one of those "thought
experiments" when it comes to transients - obviously. This has been,
and is the first issue that comes with its classic requirements
clashing with utility.

Practical Fourier mandates that you remove all D.C. components through
what is called "windowing" your data. You can choose not to and enjoy
all the artifacts of spurious behavior (like frequency folding,
frequency blurring, and false baselines) and translate those results
into blighted proofs too. With transients, practical Fourier
techniques mandate the presumption that they are periodic through all
time by literally forcing the data set (through extensive reworking)
to fit this requirement (which returns us to classic expectation).
Such reworking of original data demands that you make a choice: you
either discard amplitude accuracy for the sake of frequency accuracy,
or discard frequency accuracy for the sake of amplitude accuracy. The
irony of this choice is that you have already discard D.C. and its
appearance within the product of results is imbued with error. Need I
ask if you know how much error? Ignorance to what accuracy your
transforms bring reduce arguments to platitudes.

Simple, day to day, practical applications of Fourier discard these
issues as trivialities immediately and offer no pretense of "accuracy"
because you couldn't appreciate it anyway (unless the designer of an
IIR or FIR was so incompetent as to be that obvious).

However, this goes beyond the pale to justify a glaring presumption
"average signal magnitude not zero" with the trappings of Fourier.
For one, absolutely no one knows what is meant by "impulsive
wideband signal (500 - 2.5 GHz)." This is so extremely vague as to be
supportable by a multitude of waveforms that have wildly different
Fourier results. For instance, there is the classic impulse of
physics, lightning, which is notoriously low in frequency content over
the interval cited. Clearly such impulsions suffered are not from
nature. But this begs forcing the nomenclature to fit the problem.
There are other impulses like a gated sine wave (which obviously
renders a zero average), or a swept sine wave (same obvious average),
or a ramp wave, or a saw tooth wave, or a triangle wave, or chirp
wave. The list of impulsive wideband signals is exhausting, but
hardly accommodating to the same universal expectation.

If our correspondent, Galilea, cannot come to terms with responding to
technical issues when seeking help, the growth of forced speculation
creates its own illusory discussion that is best limited to idle
conversation overheard in a ticket line waiting for a movie box office
to open.

73's
Richard Clark, KB7QHC