>Actually, I am doing a baseband processing of using two antenna, from a
>frequency range of 200M Hz to 400 MHz and capturing the complex data
(1601
>points) from a VNA. Currently, I am trying to use Matlab IFFT to compute
>the time delay of the system response.
>
>1) I wonder if it is necessary place zero pladding on 0 Hz to 200MHz to
>capture the accuracy of the time delay? or just use IFFT to plot out the
>1601 points of data from 200M to 400MHz?
>
>2) From the IFFT, I have discovered that I have plotted out the x axis
in
>number of points (N). I wonder how can I achieve in ploting in seconds
from
>the earlier example of 200M to 400MHz?
>
Hi Keng,
So, your English is OK! I don't work with Matlab or VNAs, but I'll try to
help. As I understand it, the Vector Network Analyzer (VNA) is giving you
1601 points. Further, each point contains the Magnitude and Phase (or real
and imaginary parts) of equally spaced frequencies between 200 MHz and 400
MHz. Given this frequency response, you want to calculate the
corresponding impluse response by using the IFFT.
Here's the cookbook explanation. Set up a complex array of 8,192 entries,
running from 0 to 8,191. The values between 0 and 4,096 correspond to the
frequencies between DC and 512 Mhz. This places the 1601 values you know
at points 1600 to 3200, corresponding to the frequencies of 200Mhz to 400
MHz. The values you don't know (DC-200MHz and 400Mhz-512MHz) may affect
your final outcome, but for the first try just set them to zero. The
values between 4097 and 8191 need to be a mirror image of the other points.
See here:
http://www.dspguide.com/ch12/1.htm
After taking the IFFT, you will have an impluse response of 8,192 points,
with each point representing 1/(2 x 512Mhz) = 0.9765 nsec.
This may work as you expect. However, it may also turn into a mess
because of the unknown values in the frequency response. Good luck.
Steve
Reply by kengkc●May 22, 20082008-05-22
>Sorry, but I think your English is not good enough to explain your
problem.
> If we don't understand the question, we can't give you an answer. You
>might try having someone help you write the question. Also, here's some
>reading material on the topic that might help. Good luck.
>Steve
>
>http://www.dspguide.com/ch8.htm
>http://www.dspguide.com/ch12.htm
>
Hi Steve,
I apologise for my earlier post. I believed I have a weak
fundamentals of FFT.
Actually, I am doing a baseband processing of using two antenna, from a
frequency range of 200M Hz to 400 MHz and capturing the complex data (1601
points) from a VNA. Currently, I am trying to use Matlab IFFT to compute
the time delay of the system response.
I have doubt for my processing technique:
1) I wonder if it is necessary place zero pladding on 0 Hz to 200MHz to
capture the accuracy of the time delay? or just use IFFT to plot out the
1601 points of data from 200M to 400MHz?
2) From the IFFT, I have discovered that I have plotted out the x axis in
number of points (N). I wonder how can I achieve in ploting in seconds from
the earlier example of 200M to 400MHz?
Thanks for advice.
Rdgs
keng
Reply by SteveSmith●May 21, 20082008-05-21
Sorry, but I think your English is not good enough to explain your problem.
If we don't understand the question, we can't give you an answer. You
might try having someone help you write the question. Also, here's some
reading material on the topic that might help. Good luck.
Steve
http://www.dspguide.com/ch8.htmhttp://www.dspguide.com/ch12.htm
Reply by kengkc●May 21, 20082008-05-21
Hi,
I have some problems regarding IFFT used in Matlab. I have a frequency
range from 2M to 4M Hz where I have obtained the datas into 1601 data
points. Currently, I am trying to get the real values of time domain.
Currently, I have IFFT the data obtained (1601 points) and plotted out
the time domain waveform. Currently, I am having problem in determine the
real time axis of waveform for 1600 points. I wonder example like whta is
the real N=800 point in time domain after IFFT?