After calibration of the timing of the PCals, the relay of NE was not properly switched off (from PCal_NE_Relay process). Hence the LED placed close to PCAL_NE_PD2 was still flashing the IRIG-B on the photodiode polluting the signal, which is the signal used to subtract the PCal lines in HRec. As a consequence this unexpected noise from IRIG-B was added in hrec instead.
The figure 1 shows the NE and WE PD1 and PD2 FFTs before and after I opened the NE relay to have it in its correct state. The PCal laser power loops was running correctly since they use PD1 power, but NE_PD2 was polluted.
On figure 2, before the fix, there was no coherence between PCal_NEB_PD2 and DARM, while coherence between PCal_NEB_PD2_power and Hrec.
We will add a flag in the DMS to check that the relays are off and the PCal PD2 channels are not polluted.
I have been acting on the master laser. First by resetting the noise heater (off then on again) and then by decreasing the current of the pumping diodes from 1.792 A to 1.772 A. The output power decreased from about 530 mW to 500 mW (see the attached plot). This does not impact the slave loop so much but reduces the number of possible glitches from the master laser.