The results of the AM/PM measurements:
Both the AM and PM TFs were scaled to make them have the same average value. Manually adjusting the delay line offset for each measurement using the oscilloscope was probably not accurate enough and therefore resulted in different scaling which this should somewhat compensate.
The new calibration factor used: 5 MHz/V at the output of the mixer to obtain the frequency modulation and then division by the mod. freq. to obtain PM.
5 MHz/V because changing the PZT voltage by 0.01 V=> change in beat frequency by 0.1 MHz, which was seen as a 20 mV change in the delay line mixer output.
Again, the calibration is not very precise and I will probably repeat this experiment at some point more precisely.