When I compute the cross spectral densitiy function Pxy using cpsd.m, and divide by Pxx, the angle is the wrong sign when compared to tfestimate.m Why is that.

I am computing the transfer function of a first order butterworth filter with random noise as the input. I compute Pxx, Pyy, Pxy with cpsd.m. When I compute the transfer function, Hxy, the angle is not the correct sign. Thus the phase is not correct. Why is this? When I use Tfestimate.m, it is correct.

Answer 1

[Honglei Chen]

This is because in tfestimate, the transfer function is defined as Pyx/Pxx, and Pyx is conjugate of Pxy, that's why you see the sign difference in angle.

You can compare the algorithm section in the two functions

https://www.mathworks.com/help/signal/ref/cpsd.html

https://www.mathworks.com/help/signal/ref/tfestimate.html

HTH