Curve Fitting Of a Frequencie Response

[nigeltorque]

HI Richard..

is possible for you to post the .vi for the 2010 version??

Thanks

[nigeltorque]

Anyway Altenbach.. i try to explane what i am doing with this .vi.

I have to modeling a non-linear dynamic system.. it could be scomposed in a linear component and a non linear component.

To charachterize the linear part i send to my dynamic system (vacuum tube amplifier) a low-level signal with frequencie 10 Hz - 25000 Hz as input and see how it respondes. I take the frequencie response curve, normalize it to 1, make 1/x and the formula i found fittind this curve is the TRANSFER FUNCTION of the linear component..

So i just do all this to find out the TF of the linear component of my amplifier to charachterize it; in the attachment you see how is possible to "divide" a DUT (device under test) with a Black Box approach..

Hope i was clear...

P.S i think to find out the TF it will be more easier to use the System Identification Tool that is full of Bode and TF vi, but i don't have it... and i can't download the evaluation module cause i use Labview Student Edition 2010..

[Richard2950]

Attached is the 2010 version of the vi I posted before.

Richard

[nigeltorque]

Hi Richard

thanks for your .vi it was very useful. But i still have a question.. how is possible that if i just normalize the signal to 1, and do 1/x is impossible for me to fit new curve?? It's just the reciprocal, it's a simple curve but it's impossible to fit.. do you know why?? how is it possible?? i have to use the polynomial fit .vi because i need to specify the polynomial order to use the math equation in the right way..

In the attachment there is the final .vi; if i find the way to fit that curve my vi will be complete! 

I will be very grateful to anyone who could help me..

[altenbach]

---

nigeltorque wrote:

But i still have a question.. how is possible that if i just normalize the signal to 1, and do 1/x is impossible for me to fit new curve?? 

---

You normalize between zero and 1, so at least one point is zero.

Once you do a 1/x, at least one point is NaN because a division by zero is not defined.