How to produce a Logarithmic Sweep (Chirp) over a Frequency Band

[Daniel Akeju]

Good morning,

We followed your instructions and came close but the system crashed and our work wasn't saved. But I think we understand how to get it. My partner was the one tackling that and he successfully got data into controls although we had difficulty with the indicators (I have to ask him, I can't remember exactly). This was on Friday; we should return and be able to uploaded an updated VI.

Notwithstanding, I saw some improvements with our results. The graphs now make sense. My major concern is that someone that looks at the graphs will assume a sampling frequency of 16k when it is in fact 32k. I don't know how to solve this; I have played with the clock but can't seem to figure it out.
Another trouble is with some noise-like portion in between the sensible trace of some of our graphs; I assume this is because of the zero part of the chirp (since there is zero input at the last half of the chirp while we are still sampling).
Thank you,

Daniel

[NateBraska]

Hi Daniel,

So it seems you have most of the issues taken care of. What are the parts you are having a problem with?

Regards,

Nathan B

Applications Engineer

National Instruments

[johnsold]

Daniel,

Put an indicator on the front panel which displays the sample rate.

Another hard lesson about saving often.  Hope you did not lose too much work.

Lynn

[Daniel Akeju]

Thank you,

Lynn has been very helpful. This is in fact a wonderful forum.

I wonder if you can give me tips on converting the horizontal axis from frequency values to time values (when needed). The graphs look sensible to some extent but every one of them has frequency as the horizontal axis (including the time signals).

Thanks a lot,

Daniel

[Daniel Akeju]

Good evening,

Sorry for the delay in response. I showed my partner your response, hopefully, he'll organize everything soon so I can post it. I have been working mostly in MATLAB. Going with the reference paper, I hope to synthesize the chirp in the frequency domain using the group delay. I just completed that in MATLAB and it agrees with the paper. 

Another question is about a smart way to implement this spectral-domain synthesis in LV. Ideas that come to mind are:

1) using Mathscript (my partner seems to have done some small work with this on a previous project)

2) if everything fails, transfer the FFT values from MATLAB to LV via a text document

Thank you,

Daniel

[johnsold]

Daniel,

I am not sure what you are asking about the horizontal axis.  Are you displaying both time domain and frequency domain signals on the same graph?

You should be able to implement the algorithm directly in LV.  I am not taking the time to dig out the paper to look at it now.  Can you briefly summarize the algorithm and the difficulties you are ahving with the implementation?

Lynn

[Daniel Akeju]

Hello,

For example, the plot of chirp signal (although it comes out right) does not have time on the x-axis; what you see is frequency (which we don't want since the chirp plot is in the time domain).

I understand the algorithm to a decent extent. I am putting finishing touches to it in MATLAB. First, you find the group delay (tau) and this is used to get the phase. We can then get the X(f) as cos(phase) + j*sin(phase) as every frequency. There are little complications here and there to get this but it isn't too bad. I am more comfortable in MATLAB so I hope to carry over the code if possible.

Thanks a lot,

Daniel

[johnsold]

Daniel,

The axis labels are entered manually.  The graph does not "know" whether the data represents time, frequency, length, or temperature.  Just type in the label you want.

Lynn

[Daniel Akeju]

Hello,

I wonder if there are any nuances I might not know about FFT/IFFT in LV. I ask because I had to learn how MATLAB interpretes results for the FFT inorder to get desired results. I wonder if LV has a different way of looking at the data. I tried reading stuff up online but I would be glad to be referred to a link where I can read more.

Thank you,

Daniel

[johnsold]

Daniel,

MATLAB anf LabVIEW d some things differently.  Array indexes start at zero in LV and one in MATLAB. Newer versions of LV have a Shift? input to the FFT functions which changes the order of the output array. Look at the detailed help file for the FFT in LV.  That gives lots of information about the function and the arrangement of the outputs.

Search the LV help for FFT and look in the help index.  In the index under fast Fourier transform are several entries which may be useful,

Lynn