Why do I get so many phase jumps in the FRF of an impact hammer test?

Hello to the community!

I am performing an experimental modal analysis and the following problem occured: I use NI Labview to get my measurement signals from an impact hammer and our laser vibrometer; so for those well familiar with modal analysis - its a mobility Frequency Response Function (FRF = Velocity/Force).

Afer loading the data into Matlab I am doing software triggering and use just the spike of the hammer impact; afterwards I am performing a Fast Fourier Transform of the truncated and windowed hammer and vibrometer signals and I calculate the FRF by dividing the two transformed signals.

Now my problem arises: If I plot the phase of the FRF I get a highly oscillating phase plot - the phase is rapidly jumping between +180° and -180°, and this over a wide frequency range:

If I use unwrap(...) I get the following result, which does not really look better either:

So has anyone an idea? Both signals look quite sensible when plotted over time; but I really have no idea right now what causes these jumps in the phase of the FRF.

Maybe someone can help? Thanks!

[TM Med]

Hello, I have the same problem for a different application of fft2,did you find a solution? thank you in advance for your help.

[Greg Dionne]

Have you tried the modalfrf function with the Signal Processing Toolbox?

see Example