waveform phase modify

[nigeltorque]

Hi everybody...

let's suppose i have a waveform formed with a lot of frequencies (i.e. a square waveform). It pass through a device and i detect the phase response of my device.. now i have to implement the phase response of my device to the input signal (i.e. the square waveform) For example: the phase response of my device is zero from 0 to 10000 Hz and minus 90 degrees from 10000 Hz to 20000 Hz.. how is it possible to shift the phase of the 10000 - 20000 frequencies of my input signal of 90 degrees.. i don't know how to extrapolate the phase of alla frequencyes of the signal. is there a .vi or a way to do that?? I find the extract tone .vi but it works just with a single tone..

Any help?? 

Thank you

[johnsold]

Nigeltorque,

I think the phase response you described is physically impossible.  You require a step phase response at 10000 Hz.

An approximation to what you want is called an "all pass filter."  The magnitude response is constant for all frequencies while the phase varies with frequency.  I has been a while since I worked with all phase filters, so I would need to dig out the reference books to see what might be practical.

Another approach to simulating the behavior might work with the square waveform because it has a discrete spectrum.  You could do an FFT.  For the components above 10000 you shift the phase.  Then do an inverse FFT.  I have never tried this and have not though carefully about whether it would work and what the limitations of the method might be.

Lynn

[nigeltorque]

Hi and thank you for your reply

What do you think about the idea below in the picture?? You have a time-domain signal, make an FFT, take the compleax array and slplit real and imaginary part, find module and phase for each element of the array, modify the module (in the picture i multiply every component with a 0.5 gain but it's just an exemple), modify the phase (in the picture i add 90 degrees to every component but it's just an exemple), freturn to real and imaginay components, make an anti-FFT, return in the time-domain ?? Do you thing this can works?? I am testing this, i have just to understand the FFT real.vi because it produces a double spectrum in the frequency domain as it is explain in the .vi help.. Thake a look and let me know what you think about that.. i will post my results..

[johnsold]

Rather than Complex to Real and Imaginary, use Complex to Polar.  The outputs are magnitude and angle.  I tried some of this with mixed results.  I think it has possibilities but did not have time to dig through all the details.

Lynn