iFFT for Band-pass Measurement

Hi,

I have a outsource data (txt. file)of a band pass measurement which i took from a vector network analyzer. Measurement was made between 700 MHZ-1GHz The data contains 2 columns; 1. 301 measured frequency points 2. real return loss values of points. I have to transform the data from frequency domain to time domain with ifft function in order to calculate distance to fault on a coaxial cable How could i do it? My data is not complex conjugate and not symetric. Shoul i zero padding and mirroring ? What should i do to take ifft?

Thanks in advance!

Answer 1

[David Young]

Solve the problem in two stages:

1. Read the data from the file into an array, containing complex values. See textread, or the import wizard.
2. Call the ifft function. This is general - it does not require any particular symmetries.

[cmd]

Hi David thanks for your answer,

My data is not complex, it is absolute values of complex values. for example;

Frequency Return Loss

7e+008 -4.08136

7.01e+008 -5.84865

7.02e+008 -8.39599

Can i transform the real data to time domain as real?

The code i tried to write is that,but i could't get the right result;

fid = fopen('rlos.txt');

FDomain = fscanf(fid, %f %f\n',[2,inf]);

FDomain = FDomain';fclose(fid);

f = FDomain(:,1);

RL = FDomain (:,2);

N = 301;

fstep = 1e6;

dt = 1/(fstep*(N-1));

t = dt*[0:N-1];

inv_RL = ifft(RL);

plot(t,inv_RL);

Answer 2

[cmd]

I have only positive frequency points and Real RL values. Should i have complex conjugate symetric data for ifft in order to transform to time domain? Is my ifft point 301 or 2^n like 512 or 1024? How could i get it? Please halp me..

Thanks in advance...

Answer 3

[jks]

i am working on the same kind of problem , please let me know if you have solved the case

Answer 4

[Joh Yhan]

Hi cmd,

Try to replace your fstep to the following. B is the measurement bandwidth.

fstep = B / N;

Answer 5

[AK]

Well I am also working on some similar kind of problem in 5-10 GHz range which i obtained from VNA which i need to calculate ifft but assuming zero values in 5-10 GHz range will always make results go wrong so some other technique will be required to be done. please let me know if r able to find solution to this answer

[Walter Roberson]

We *might* be able to find a solution to this, but to do that we need you to answer the questions raised in response to your Question on this topic, http://www.mathworks.com/matlabcentral/answers/25345-ifft-of-bandpass-signal-data

Answer 6

[Dr. Seis]

You cannot reconstruct your timeseries unless you have either:

1. Complex values for both positive and negative frequencies. The real parts of the amplitudes are symmetric about 0 frequency, while the imaginary parts of the amplitudes are anti-symmetric [i.e., imag(G(f)) = -1*imag(G(-f)) ] about 0 frequency.

2. Absolute values for both positive and negative frequencies (which are symmetric about 0 frequency) and the phase-angle information associated with the real and imaginary parts.