Modifying the pumping current of the diode may actually modify the mode hop temperature mid range.
This morning we went in laser to tune the temperature of the crystal of the ML.

We first tried to change the temperature of the crystal looking at the PDH of the slave but it had a very strange behavior (cf plot 1) and we could not get any information from it.
We then tried to keep the SL locked and to look at the correction (PSL_SL_PZT_SLOW) to see when there were unlocked that were not due to the end of the PZT range (about -0.5 V for the lower limit and 6 V for the upper one). See the attached plot and the different temperatures hereafter:

1297424074  15 Feb 2021 11:34:16 UTC  (15 Feb 2021 12:34:16 CET)
temp display 30
pzt 0.3

1297424167  15 Feb 2021 11:35:49 UTC  (15 Feb 2021 12:35:49 CET)
temp display 28.63
pzt before jump 3V

1297424247  15 Feb 2021 11:37:09 UTC  (15 Feb 2021 12:37:09 CET)
temp display 27.26
pzt before jump 5V

1297424330  15 Feb 2021 11:38:32 UTC  (15 Feb 2021 12:38:32 CET)
temp display 25.9
pzt before jump 6V

1297424415  15 Feb 2021 11:39:57 UTC  (15 Feb 2021 12:39:57 CET)
temp display 24.7
pzt before jump 6V

1297424492  15 Feb 2021 11:41:14 UTC  (15 Feb 2021 12:41:14 CET)
temp display 23.4
pzt before jump 6V

1297424583  15 Feb 2021 11:42:45 UTC  (15 Feb 2021 12:42:45 CET)
temp display 22.2
pzt before jump 5.5V

1297425244  15 Feb 2021 11:53:46 UTC  (15 Feb 2021 12:53:46 CET)
temp display 24.1 (going up with temp)
pzt before jump -0.5V

1297425305  15 Feb 2021 11:54:47 UTC  (15 Feb 2021 12:54:47 CET)
temp display 25.5.
pzt before jump -0.5V

1297425305  15 Feb 2021 11:54:47 UTC  (15 Feb 2021 12:54:47 CET)
temp display 25.9 (!!!)
pzt before jump 1.5V

1297425415  15 Feb 2021 11:56:37 UTC  (15 Feb 2021 12:56:37 CET)
temp display 27.5
1297425461  15 Feb 2021 11:57:23 UTC  (15 Feb 2021 12:57:23 CET)
temp display 29.1

We did it in one way and the other, noticed two unlocks that were not due to the end of range of the PZT:
28.6 while going down
25.9 while going up

This is strange, we were expecting more mode hops (cf figure 3)

At the end of the day we took 25.9 as a mode hop and decided to stay 1.3 degrees away from it :
1297425816  15 Feb 2021 12:03:18 UTC  (15 Feb 2021 13:03:18 CET)
set to 24.6 degrees.

In the meantime Jean-Pierre explained us how to get a clean scan from the SL rampeauto. We may try again the first method (ie looking at the PDH) to better see the mode hops.

We took advantage of working on the ML to change the monitoring signals that we are acquiring.
The channel PSL_ML_NOISE_EATER_FS is now monitoring the "crystal temperature error signal"
The channel PSL_ML_D2_POWER_FS is now monitoring the temperature of diode 1. 
We will update the names soon.

The name of the channels have been updated by Alain.

PSL_ML_CRISTAL_TEMP_ERR_FS is monitoring the error signal of the temperature control of the crystal

PSL_ML_D1_TEMP_ERR_FS is monitoring the error signal of the temperature control of the pumping diode

I took a look at the data of the past weeks.
It was difficult to get any conlusion from the data between the 12/02 (increase of the pumping current) and the 18/02. The weekend of the 13/02 14/02, we had some trouble with the lock of the RFC (log entry https://logbook.virgo-gw.eu/virgo/?r=50762 ). It was responsible for most of the unlocks of the weekend. Then there were several intervention in laser lab at the beginning of last week (plot 1).
The data from the 18/02 to the 23/02 were more interesting (plot 2).  Between the 18/02 and the 22/02, there were 10 fast unlocks. Only one unlock was due to the RFC.
I attached an example of those unlock (plot 3). Nothing noticeable, INJ_EOM_CORR seems to be the first one to have a glitch.
Nothing visible in the slow signal (ENV, HVAC, MAG...) for those unlocks.
We still have to properly check that we are in middle of a mode hop (log entry https://logbook.virgo-gw.eu/virgo/?r=50770), but for now it seems that passing fron 2.0 A to 2.1 A of pumping current has deteriorated the situation ; while passing from 1.8 A to 2 A seemed to have improve it (see https://logbook.virgo-gw.eu/virgo/?r=50535).