In particular, Bas already found that the comb at 10.2782 Hz comes from the central building. Therefore, further investigation here reported has been focused on the actuation at level of the NI and WI RefMasses.
Summary of the main tests (time in UTC):
#1. 13oct11, ~10:30, unplugged WI and NI RefMass coils (cable V at level of the tower flange, reconnected the morning after) --> comb disappears (this can be seen in Fig1 for the fundamental and the 1st harmonic). It has also seen that by unplugging only the WI coils, the comb is still present, although reduced.
#2. 17oct11, 17:05, turned OFF the EtherCAN of the CoilRelay system (this is the system that drives the relay for the HP/LN1/LN2 transitions) --> comb still present on the DF.
#3. 18oct11, 18:20, turned OFF the EtherCAN and unplugged the bus connector of all the CAN digital I/O modules (on WI, BS, NI). To be mentioned that the digital I/O modules cannot be turned off or unplugged from the coil driver, otherwise the relays return in their rest position, corresponding to HP. --> comb disappears (in Fig2 this is shown for the 1st harmonic, in Fig3 there is a comparison before/after, Ca_WI_RM_xx is the channel monitoring the coil voltage).
Conclusions:
-The comb enters through the NI and WI RefMass coils (test#1).
-The source is the CAN network of the CoilRelay system. Turning OFF only the etherCAN (test #2) is not enough, since the I/O modules continue to communicate trough the ~10m-long cable, connecting WI-BS-NI. Indeed, unplugging also the bus connector of the I/O modules (test #3), no CAN communication is possible along that cable.
To be mentioned that the CAN networks of NE and WE don't contribute to this disturbance because the local bus there is quite short (below 0.5m).
