Foreseeing the scattered light investigations on SDB1 we devised an alternative lock (mainly for MICH) without using any sensor from SDB*; we recall:

• MICH is defined as the fringe signal B1p_DC/(B1p_DC + x*B4_DC) up to dark fringe;
• PRCL is defined as B2_8MHz_I/(B1p_DC + x*B4_DC) up until LOW_NOISE_2 included,

where x = 10.5, constant. The best strategy would be to still use two sensors and intercalibrate them, and in such a way to have MICH still on a fringe signal, spanning the [0,1] range; after a few offline tests, we decided to give it a try using only B4_DC:

• the MICH fringe signal (LSC_B4_DC_FRINGE) is defined as (1 - y(s)*B4_DC), where y(s) (i.e. LSC_B4_DC_CALI) depends on the MICH setpoint;
• PRCL (LSC_B2_8MHz_I_NORM) is defined as B2_8MHz_I / ( z(s)*B4_DC ), where z(s) (i.e. LSC_B2_DC_CALI) depends also on the MICH setpoint;

The two DC_CALI parameters have been calibrated during a standard lock by looking at B1p_DC at each step from Recombined to 0.2 dark fringe; the caveat for the MICH one is that it must be changed also on the fly as soon as we start recycling light after closing PRCL, or the fringe definition would be lost; this is the main disadvantage w.r.t. having two intercalibrated sensors as we usually do.

The scheme seems a bit convoluted, but it has advantages from the Metatron point of view:

• we could use different state paths for MICH_LOCK and PRCL_LOCK without modifying the usual ones;
• the MICH fringe definition switch for the Recycled configuration is done inside PRCL_LOCK, and it uses B2_DC as a trigger; the 1 Hz speed of Metatron was enough to make the switch reliable;
• inside ITF_LOCK we only changed the state requests for MICH_LOCK and PRCL_LOCK, and we set the CALI parameters at the same time as the standard gains; when we will switch back to the usual lock acquisition they would be ineffective even if we forget to remove them;
• no configuration parameters were harmed in the process.

We did several tests during the afternoon and the procedure is quite reliable! We already left it as default, and we also made a test closing the B1p_PD2 shutter in order to be sure not to have forgotten some trigger or threshold check of sorts. We should be ready to give up the SDB* sensors from now on.

A side note about the communication between processes: after modifying LSC_Acl, mostly by adding elements to ACL_SUM_CH or ACL_MATRIX_CH objects, sometimes a standard Reload of a Metatron node was not enough to load the new version of the Acl server, resulting in connection errors in Metatron but also the crash of LSC_Acl itself; this would not be dangerous on its own, but in this case a channel sent via Tolm to another Acl process (and probably a DSP as well) would freeze to the last value, which is not safe. This happened for instance to the SSFS flag sent by LSC_Acl to ISYS_Acl, resulting in crazy corrections sent to the mirror which opened the controls.

We conditioned such flag with the IMC_TRA power using the same RELAY recently added for the alignment of the IMC, but the change will be applied tomorrow.