Today the new version of the INJ_MAIN Metatron node was put in operation; in the Figure there is the new layout of the node; a few remarks:

• as usual, the DOWN state is the one to be requested when one want to do things by hand in the Injection system;
• the INJ_FAULT state will be used in case of severe misbehaviour of the Injection, and Metatron will not exit by itself from this state; it is different in scope w.r.t. the old END state (hence the name change) and it is not used by now, it will be discussed with INJ people how to use it effectively;
• the GALVO_CLOSE and RFC_CLOSE states are largely unaffected by the update, if not for some polishing; the only novelty is that they are now non-requestable states, since it made little sense in the first place as they are transient states;
• IMC_RESTORED is the usual good state for the INJ system, but only when in standalone mode, i.e. outside the main Lock Acquisition sequence: here we will implement the checks on the Master Laser thermal correction and possibly some checks on the drift of the IMC top stage position; the skeleton is already there (you can see the relevant states in the graph), but nothing happens here just yet; in case of a IMC unlock when in this state, the machine will go to DOWN, then it will find its way back to here;
• the FmodErr sequence: the actual tuning is identical to the most recent one, which is Paolo's loop using INJ_FmodErr_ACp as error signal, but the logic around it has been revised:
  • in FMODERR_TUNING, the RFC is opened, Drift Control is enabled and the thermal loop is opened (BPC.ML_TH_ENB = 0); then, the IMC_perturb line is switched on, the loop is engaged and the error signal is checked against a strict tolerance; when the criteria is met, the loop is opened and the IMC is lock-cycled; the MC top-stage Z position is saved for reference; all of this is done by default, without any prior check on the goodness of the previous tuning;
  • in RFC_SCAN we close the thermal loop and the RFC, and we wait for it to acquire the lock; in this phase, if the IMC unlocks by itself, the FmodErr procedure starts again from scratch; when all is good, we disable the drift control and move to full bandwidth alignment;
  • in FMODERR_CHECK, we see if the RFC locked in the good place, i.e. if the INJ_FmodErr_ACp error signal is still below a tolerance, which is a bit less strict than the first one; if it is not, we start again from scratch;
  • in FMODERR_TUNED we do nothing, except checking with the usual decorator if the IMC is locked; differently from IMC_RESTORED, an unlock while in here will make the machine to ask directly for IMC_RESTORED, which means putting the INJ system in standalone mode without automatically re-do the FmodErr tuning after an unlock, as it was the case up until now;
  • all the FmodErr states are non-requestable (aside from the final state, obviously) and non-redirectable from;
• RESET_ML_TH_CORR and MOVE_MC_F0_Z will be used respectively to reset the thermal correction loop if the correction is too high and to move the MC top-stage position in case of need; they are not active for the time being, and they will be reachable only from the standalone path;
• the entire Noise Budget path suffered no change at all;
• I kept in an offline backup the old states relative to FmodErr done mechanically and via modulation frequency change, since they are either obsolete or ahead of time.

Since now we want to do the FmodErr procedure before starting the Lock Acquisition, and not anymore in case of an unlock, all of the above made some changes in ITF_LOCK necessary:

• in DOWN we don't do anything anymore, we just request IMC_RESTORED (INJ good and standalone);
• the secondary configurations (CITF, SHORT_MICH, PR*I) were left unaffected;
• there is the new FMODERR_TUNED state (index=39) between DOWN and LOCKING_ARMS, which does the obvious; however, since for commissioning / recovery / laziness purposes one does not need FmodErr tuned every time, there is the new 'fmoderr_skip' configuration parameter at the top of the ITF_LOCK.ini file; when this is set to 1 / true / yes, ITF_LOCK will request IMC_RESTORED instead of FMODERR_TUNED, effectively skipping the tuning and moving on locking the arms.

The ITF_STATUS node has been updated to reflect these changes and reloaded (going to STOP to ensure a correct reload); since they are not used by now, I disabled the automatic saving of the thermal correction reference and the MC top-stage position reference which were done by this node (after I found out that it did...), since I'm not sure that these values should be saved at the end of the Lock Acquisition, but rather by INJ_MAIN itself after a successfull FmodErr tuning.

I witnessed about half a dozen unlocks so far with the new node, and everything seems to work fine, and no FmodErr mistuning has been found while in LOW_NOISE_2, when the slow FmodErr loop is engaged and its signal provides a good information about the mistuning.

EDIT: I will monitor this update along the weekend, but if there is the dire need of going back to the previous situation, there is a backup of the three nodes involved in /virgoDev/Automation/userapps/backup/INJ_MAIN_elog_38756; please do not copy over the files, but open both the old and the new one for each file, and copy-paste the content; this is in order to avoid permission issues on the files like it happened some time ago.

Then, for ITF_STATUS it is sufficient to "LOAD" again the node and select the STOP state; for ITF_LOCK and INJ_MAIN it is better to stop/restart the processes from the VPM, and close/reopen the GUI client afterwards.