| Topic: |
Science > Physics |
| User: |
"Pete" |
| Date: |
10 Apr 2006 06:16:59 AM |
| Object: |
Experiments that measure frequency domain |
I read there are experiments that measure the frequency response of a system
Y(w) and the researcher must use the inverse Fourier transform to get the
time response y(t). Unfortunately, it didn't give an example.
It seems odd that the researcher can't hook up his measuring device to the
apparatus and simply measure the time responses.
I'm curious -- what are some examples of experiments where Y(w) is more
easily measured than y(t)?
Thanks,
Pete
.
|
|
| User: "Richard Herring" |
|
| Title: Re: Experiments that measure frequency domain |
10 Apr 2006 10:51:41 AM |
|
|
In message <443a3eab@usenet.zapto.org>, Pete <nospam@nospam.org> writes
I read there are experiments that measure the frequency response of a system
Y(w) and the researcher must use the inverse Fourier transform to get the
time response y(t). Unfortunately, it didn't give an example.
It seems odd that the researcher can't hook up his measuring device to the
apparatus and simply measure the time responses.
I'm curious -- what are some examples of experiments where Y(w) is more
easily measured than y(t)?
Bragg cells. A mechanical wave travelling along a suitable medium alters
its optical properties, turning it into a diffraction grating. Shine a
light on it from the side, and the resulting diffraction pattern is the
frequency spectrum of the mechanical wave.
--
Richard Herring
.
|
|
|
|
| User: "rge11x" |
|
| Title: Re: Experiments that measure frequency domain |
10 Apr 2006 10:05:34 AM |
|
|
It depends on the bandwidth of the system you want to measure. If the
BW is only a few kHz there is usually no problem to characterize it
direclty in time domain but imagine that you want to characterize a
system whose bandwidth is several GHz. Now to see anything you will
need to generate and measure a pulse whose rise time that is a few
picoseconds. How would you do something like that? It is much easier to
set up a slowly chirped (frequency varying) oscillator and coherently
detect both its amplitude and phase as its signal passes through the
system of your interest.
.
|
|
|
|
| User: "rge11x" |
|
| Title: Re: Experiments that measure frequency domain |
10 Apr 2006 10:06:11 AM |
|
|
It depends on the bandwidth of the system you want to measure. If the
BW is only a few kHz there is usually no problem to characterize it
direclty in time domain but imagine that you want to characterize a
system whose bandwidth is several GHz. Now to see anything you will
need to generate and measure a pulse whose rise time that is a few
picoseconds. How would you do something like that? It is much easier to
set up a slowly chirped (frequency varying) oscillator and coherently
detect both its amplitude and phase as its signal passes through the
system of your interest.
.
|
|
|
|
| User: "rge11x" |
|
| Title: Re: Experiments that measure frequency domain |
10 Apr 2006 10:06:20 AM |
|
|
It depends on the bandwidth of the system you want to measure. If the
BW is only a few kHz there is usually no problem to characterize it
direclty in time domain but imagine that you want to characterize a
system whose bandwidth is several GHz. Now to see anything you will
need to generate and measure a pulse whose rise time that is a few
picoseconds. How would you do something like that? It is much easier to
set up a slowly chirped (frequency varying) oscillator and coherently
detect both its amplitude and phase as its signal passes through the
system of your interest.
.
|
|
|
| User: "Jan Panteltje" |
|
| Title: Re: Experiments that measure frequency domain |
10 Apr 2006 10:17:32 AM |
|
|
On a sunny day (10 Apr 2006 08:06:20 -0700) it happened "rge11x"
<rge11x@netscape.net> wrote in
<1144681580.298407.229140@u72g2000cwu.googlegroups.com>:
Now to see anything you will
need to generate and measure a pulse whose rise time that is a few
picoseconds. How would you do something like that?
http://www.npl.co.uk/electromagnetic/ufe-opto/calibration/ufe-calibration.html
It is much easier to
set up a slowly chirped (frequency varying) oscillator and coherently
detect both its amplitude and phase as its signal passes through the
system of your interest.
Yes, easier......
hehe
.
|
|
|
|
|
| User: "Timo Nieminen" |
|
| Title: Re: Experiments that measure frequency domain |
10 Apr 2006 01:30:00 PM |
|
|
On Mon, 10 Apr 2006, Pete wrote:
I read there are experiments that measure the frequency response of a system
Y(w) and the researcher must use the inverse Fourier transform to get the
time response y(t). Unfortunately, it didn't give an example.
It seems odd that the researcher can't hook up his measuring device to the
apparatus and simply measure the time responses.
I'm curious -- what are some examples of experiments where Y(w) is more
easily measured than y(t)?
This can be the case when
(a) the time response happens so quickly that the resolution and response
time of the instrument can't measure y(t). This tends to happen with very
high frequencies eg in optics, where spectrometers are standard issue.
(b) the system has a very narrow resonance, so you get all the important
information with a small number of frequency domain measurements. Narrow
features in the frequency domain translate into very broad features in the
time domain - think about how many samples in the time domain over what
time are required to resolve a narrow resonance.
--
Timo Nieminen - Home page: http://www.physics.uq.edu.au/people/nieminen/
E-prints: http://eprint.uq.edu.au/view/person/Nieminen,_Timo_A..html
Shrine to Spirits: http://www.users.bigpond.com/timo_nieminen/spirits.html
.
|
|
|
| User: "Jan Panteltje" |
|
| Title: Re: Experiments that measure frequency domain |
10 Apr 2006 01:47:51 PM |
|
|
On a sunny day (Tue, 11 Apr 2006 04:30:00 +1000) it happened Timo Nieminen
<uqtniemi@mailbox.uq.edu.au> wrote in <20060411042548.O12714@emu.uq.edu.au>:
(b) the system has a very narrow resonance, so you get all the important
information with a small number of frequency domain measurements. Narrow
features in the frequency domain translate into very broad features in the
time domain - think about how many samples in the time domain over what
time are required to resolve a narrow resonance.
Actually that is a bit 'shady'.
Suppose you have a metal plate, and it resonance (or one of its resonances)
is 100Hz.
First it is not only 100Hz, there is such a thing as bandwidth (and Q).
me, I would tap the plate, and then see a 100Hz damped oscillation on
the scope (measure 10mS period time), done that++ times.
Shortcut hey :-)
.
|
|
|
| User: "Timo Nieminen" |
|
| Title: Re: Experiments that measure frequency domain |
10 Apr 2006 02:56:56 PM |
|
|
On Mon, 10 Apr 2006, Jan Panteltje wrote:
On a sunny day (Tue, 11 Apr 2006 04:30:00 +1000) it happened Timo Nieminen
<uqtniemi@mailbox.uq.edu.au> wrote in <20060411042548.O12714@emu.uq.edu.au>:
(b) the system has a very narrow resonance, so you get all the important
information with a small number of frequency domain measurements. Narrow
features in the frequency domain translate into very broad features in the
time domain - think about how many samples in the time domain over what
time are required to resolve a narrow resonance.
Actually that is a bit 'shady'.
Suppose you have a metal plate, and it resonance (or one of its resonances)
is 100Hz.
First it is not only 100Hz, there is such a thing as bandwidth (and Q).
me, I would tap the plate, and then see a 100Hz damped oscillation on
the scope (measure 10mS period time), done that++ times.
A handful of frequency domain measurements will give you both f0 and Q, if
you know approximately what f0 and Q are. Suppose that f0 is equivalent to
a wavelength of 1 micron, and Q is equivalent to a time constant of
10^{-8} seconds. Difficult to measure both frequency and damping with a
single set of time domain measurements. One could do it with a couple of
measurements, but if the _shape_ of the frequency response curve is of
interest, then the demands that the Fourier transform places on the time
domain data is severe. The cool thing is that it's still easy to go from
frequency domain to time domain, since only the region near the resonance
does anything interesting, so you can get away with very sparse data away
from the resonance.
I met this case in theory, and it was a case of calculating frequency
response rather than measuring, but it still came down to about 2 hours in
frequency domain, or a minimum of months in time domain.
If you're only interested in f0, then the time domain measurement is
excellent; there's a reason why bells work the way that they work. If the
shape of the frequency response curve matters, OTOH ...
--
Timo Nieminen - Home page: http://www.physics.uq.edu.au/people/nieminen/
E-prints: http://eprint.uq.edu.au/view/person/Nieminen,_Timo_A..html
Shrine to Spirits: http://www.users.bigpond.com/timo_nieminen/spirits.html
.
|
|
|
|
|
|
| User: "Jan Panteltje" |
|
| Title: Re: Experiments that measure frequency domain |
10 Apr 2006 06:35:36 AM |
|
|
On a sunny day (10 Apr 2006 13:16:59 +0200) it happened Pete
<nospam@nospam.org> wrote in <443a3eab@usenet.zapto.org>:
I read there are experiments that measure the frequency response of a system
Y(w) and the researcher must use the inverse Fourier transform to get the
time response y(t). Unfortunately, it didn't give an example.
It seems odd that the researcher can't hook up his measuring device to the
apparatus and simply measure the time responses.
I'm curious -- what are some examples of experiments where Y(w) is more
easily measured than y(t)?
Thanks,
Pete
As you do not say what sort of stuff you measure......
Anyways, the most simple 'time response' is in digital logic,
you use an oscilloscope to measure the in- and output of a gate [for example],
and apply a fast step function to the input of that gate.
This will give you rise time, delay, and oscillations in the time domain.
Also works with an operational amplifier, so any linear circuits.
Then there are special formed pulses you can apply (sin square, sinc, etc),
to get more info.
Here is a nice tutorial on using sinc shaped pulses to speed up transmission
system (make better use of the bandwidth).
http://zone.ni.com/devzone/conceptd.nsf/webmain/1D3FB0D4CDB070AC86256E12007802E3
In case you wanted a frequency response, just use a frequency sweep.... measure
amplitude (same equipment), phase change (same equipment)...
so its easy.
Now what was it you actually wanted to measure?
.
|
|
|
| User: "Peter Jay Salzman" |
|
| Title: Re: Experiments that measure frequency domain |
10 Apr 2006 06:48:49 AM |
|
|
Jan Panteltje <pNaonStpealmtje@yahoo.com> wrote:
On a sunny day (10 Apr 2006 13:16:59 +0200) it happened Pete
<nospam@nospam.org> wrote in <443a3eab@usenet.zapto.org>:
I read there are experiments that measure the frequency response of a system
Y(w) and the researcher must use the inverse Fourier transform to get the
time response y(t). Unfortunately, it didn't give an example.
It seems odd that the researcher can't hook up his measuring device to the
apparatus and simply measure the time responses.
I'm curious -- what are some examples of experiments where Y(w) is more
easily measured than y(t)?
Thanks,
Pete
As you do not say what sort of stuff you measure...... Anyways, the most
simple 'time response' is in digital logic, you use an oscilloscope to
measure the in- and output of a gate [for example], and apply a fast step
function to the input of that gate. This will give you rise time, delay,
and oscillations in the time domain. Also works with an operational
amplifier, so any linear circuits. Then there are special formed pulses
you can apply (sin square, sinc, etc), to get more info. Here is a nice
tutorial on using sinc shaped pulses to speed up transmission system (make
better use of the bandwidth).
http://zone.ni.com/devzone/conceptd.nsf/webmain/1D3FB0D4CDB070AC86256E12007802E3
In case you wanted a frequency response, just use a frequency sweep....
measure amplitude (same equipment), phase change (same equipment)... so
its easy. Now what was it you actually wanted to measure?
Hi Jan,
Nothing, really. I have trouble hooking up a voltmeter. :) I don't really
know much about experimental anything. It was something in a book that
seemed odd to me because I couldn't think of an example.
Wouldn't matter if the experiment were measuring electrical signal, heat
transfer, diffusion, etc. I just wanted to know an example of (any)
experiment in which the frequency response was easier to measure than the
time response. :)
Thanks!
Pete
.
|
|
|
| User: "CWatters" |
|
| Title: Re: Experiments that measure frequency domain |
10 Apr 2006 05:38:13 PM |
|
|
"Peter Jay Salzman" <p@dirac.org> wrote in message
I just wanted to know an example of (any)
experiment in which the frequency response was easier to measure than the
time response. :)
The time response is frequently called the impulse response.
The frequency response of a HiFi system is probably easier test with a
frequency generator that can be swept from 0-25KHz. Hitting the input with a
full power impulse might not do the speakers much good.
.
|
|
|
|
|
|
| User: "ABarlow" |
|
| Title: Re: Experiments that measure frequency domain |
11 Apr 2006 12:37:39 AM |
|
|
Pete wrote:
I read there are experiments that measure the frequency response of a system
Y(w) and the researcher must use the inverse Fourier transform to get the
time response y(t). Unfortunately, it didn't give an example.
It seems odd that the researcher can't hook up his measuring device to the
apparatus and simply measure the time responses.
I'm curious -- what are some examples of experiments where Y(w) is more
easily measured than y(t)?
Thanks,
Pete
Radio telescopes do all of their measurements in frequency space from
what I recall.
A.
.
|
|
|
|
| User: "Andy Resnick" |
|
| Title: Re: Experiments that measure frequency domain |
10 Apr 2006 07:55:42 AM |
|
|
Pete wrote:
I read there are experiments that measure the frequency response of a system
Y(w) and the researcher must use the inverse Fourier transform to get the
time response y(t). Unfortunately, it didn't give an example.
It seems odd that the researcher can't hook up his measuring device to the
apparatus and simply measure the time responses.
I'm curious -- what are some examples of experiments where Y(w) is more
easily measured than y(t)?
There's lots of experimental situations where this occurs: dynamic light
scattering, spectroscopic measurements, crystallography, vibration
measurements, etc.
Note that coverting between the two is not always simple: the phase
components are generally lost, making typical appoximations like
"time-avaged" necessary.
--
Andrew Resnick, Ph.D.
Department of Physiology and Biophysics
Case Western Reserve University
.
|
|
|
|
|
Related Articles |
|
|
