This morning we investigated on the RF pollution seen by the signal INJ_IMC_TRA (cf entry https://logbook.virgo-gw.eu/virgo/?r=57073)
For the RF frequency band one should look at the signal DAQ_FMOD_Out0_sample, which is the one used for the digital Fmoderr.

In the previous entry we noticed that even when there is no beam arriving on the photodiode, there were anyway lines at 6 8 56 MHz.

We did some tests on the cables bringing the signal from the EIB towards the demodualtion box.
Here below is the tests that we did, times are in local time :

10:55:00: 50 Ohm at the level of the lemo/BNC connection on EIB 
10:58:00 same but without the 50 Ohm
11:08:50 replace the 15m-long LEMO cable from EIB by a BNC cable
11:14:00 BNC cable and 50 ohm
11:16:20 Ferrite block on the BNC cable at the EIB level
11:19:00 Ferrite block on BNC cable in piscina
11:20:15 remove the ferrite block at the EIB level
11:22:15 put a second ferrite block on the BNC cable in piscina
11:xx:xx switch back to the LEMO with now a magnet on the piscina side
11:35:00 With LEMO unplugged from the BNC/LEMO box in piscina (ie only box and few meters of BNC towards the demodulation box)
11:38:40 BNC unplugged from the box
11:41:00 BNC plugged to box without other splitter
11:44:40 BNC plugged without LEMO
11:48:20 we try to connect LEMO to BNC with another LEMO/BNC box
11:52:50 we now plug the 15m long BNC directly to BNC in piscina (with an I)
12:00:30 we change the position of the RF splitter in order to separate the interesting signal befrore arriving on the T splitter
12:02:00 we come back to LEMO-BNC box with the RF splitter

Some conclusion from thoses tests :

- the 8 MHz disappeared. To be noted however that amplitude is set to 0 dB while the others are set to 12 dB
- best configuration BNC to BNC for the 56 MHz (11:52:50 configuration), lemo to BNC for the 6 MHz (11:41:00) 
- Strangely using a dedicated lemo to BNC box worsen the situation
- the "signal distribution", ie 2 BNC splitters at the output of the lemo to BNC box, are worsening the situation
- the 2f lines that can be seen in DAQ_FMOD seem to come from the optics. The 112 MHz line was reduced by two order of magnitude while unplugging the first cable (10:55:00) and the 12 MHz one disappear. 

The origine of the pollution is now better understood. We should speak with some expert to try to get rid of it.