fourier transform of a voltage output in real time from a fiber optic sensor

[Dhiraj]

Hi,
I am totally new to LabView and had some queries regarding my project (I am a research student)
I am making a fiber optic pressure sensor and need to perform the fourier transform of the output of the sensor in real time. The output of the sensor is given to a photodetector and the output is available as a voltage. By carrying out the fourier transform of this voltage signal I hope to get the frequency makeup of the signal. This frequency is further needed for calculating the wavelength and the wavelength is further required in modelling the progress of the wave. This is for tsunami detection.
Can somebody please help me?

[LV_Pro]

A few questions to help get us started. What frequency range are you planning for? What means were you planning to get the output of your photodetector into your computer? What is your definition of "real time"? A FFT requires a number of samples (2^n) to perform its calculations. Depending on the answer to the first question above determines how long between the start of data capture and the output from the FFT. On a reasonably fast machine (anything realistically capable of running the "modern" versions of LabVIEW i.e.6.x and above) the FFT calculation is pretty fast. If you are looking at marine wave data I would presume fairly low frequency signals in the sub Hz to 100's of Hz (vs KHz, MHz, GHz) in which case the longest time element will be gathering the 256+ data samples to perform the FFT on. Let us know more details so that we can help. As to its "doability" this is the type of project that LabVIEW excels at.

P.M.

Message Edited by LV_Pro on 07-07-2005 08:26 AM

[Dhiraj]

Thanks for the response.
Well let me give you some more details. The sesor that I have is a diaphragm based sensor. The diaphragm is made of perspex sheet as it reflects laser light. The sensor is moored at a depth of 200m. Now when a tsunami passes over the sensor, the sensor measures the dynamic pressure caused by the tsunami. The diaphragm bends and the received light intensity increases. The problem is that the diaphragm would bend  under the influence of a huge surface wave too, say with a wave height of 15 to 20m. So the photodetector that I am using will give the same output for both the waves. The only distinguishing factor is the wavelength of the wave. For a tsunami, it is typically 200-500Kms while for a surface wave, it is 200-500m max. So what I was thinking was, that if I can carry out a FT of the photodetector out put, I can perhaps get the frequency signature of the wave. Then with a staggered arrangement of the sensors, I can calculate the wave velocity and subsequently the wavelength as, I assume. I already have the frequency signature of the wave from the FT. The photodetector output is in terms of voltage. So I have the photodetector output voltage as 10V (arbit.) for two waves-one, a tsunami and the other a surface wave with a very high amplitude. And on the basis of their wavelength, I want to distinguish between them. Is it possible?

[johnsold]

With km wavelengths the period of each wave must be in the 10s of seconds to minutes range. To get enough data for FFT based methods to be accurate would take much longer than you want to wait for the results. Time domain analysis is probably more appropriate. Look for peaks and valleys and measure the time between them to get the period. With this approach you could begin to get results after 1/2 period of the wave if the data is not too noisy. Of course eliminating all the noises which will likely exist in such a system will be a significant challenge. Read up on 1/f noise if you are not already familiar with it.

Lynn