With clues from Tomura-san and Miyakawa-san, I figured out the grounding cuase for RFPDs and WFSs.

Note that this is related to all RFPDs and WFSs.

In the design (LIGO-D1101177), the PD box and the base are insulated with an aluminum plate. However, as noted in the main post, WFS2 box and the base was electrically connected in a somewhat sattle way. It was because there were not insulate washers for the screws attaching the base and the box (see the yellow circle of the attached screenshot). If the black coat (anodized surface) of the base is fade away, the base and the box are electrically connected through the screws. So, depending on the condition of the base surface, in the unfortunate case, the box are connected to the optical table, and it possibly make large ground loops.

To make sure the box doesn't electrically connected, we must use the insulated washer for those screws. Once the fiscal year of 2019 start, I will order bunch of the insulated washers and we will do the PD insulating campaign.

For WFS2, as a temporary treatment, I made two insulated washers with kapton tapes. Now the box is disconnected from the optical table (but they should be replaced in near future). The PDs on the IMC REFL table should be fine now.

Also, I quickly checked the PDs on the IFO REFL. The WFSs seemed fine but the RFPDs are electrically connected to the optical table.