it seems the swap of the PD used to lock the PMC f(see 40736) has induced some extra length nois at the PMC level.
We swapped from a 1 mm PD to a 3 mm PD to "reliabilize" the PMC reflected power survey. At that time we 've noticed some ~ 2 MHz noise on the PDH signal and a somewhat modified PDH shape wrt to the old situation.
See the attached slide.
Some clues were connected with the way the PD is powered (no RF feedthrough on the case), this is under investigation, we are preparing a spare PD with "old" fashion power supply cable/connector.