Using tfestimate(), I have been able to determine the sensitivity S of a physical system. Now, to do some calculations, I used frd() to create a frequency response object. This neatly recovers the phase of the system by preventing 'wrap-around'. However, at high frequencies, it does this a bit too overzealous for my taste; it will determine that the phase lag and lead is up to 2000 degrees. I believe this is just noise, and don't really care about the actual values; however, the large phase messes up the scaling of my phase plot.
So, how can I prevent frd() from creating ridiculous phases lead/lag from noise? Or is there another way to solve this (perhaps filtering?)
At low frequencies, the phase 'wraps around', and this is corrected by frd(). At high frequencies, the noise is converted to ridiculous phases, which I don't really want.