Using a script called histogram.py which calls statecounter2.0, I was able to determine how many times ISC_LOCK spent more than 60 seconds in ACQUIRE_DRMI_1F using Minute trends.
I broke this up into Pre Vent and Post vent:
Pre Vent:
Jan1st to April 1st 2025
Length of data 601
max Duration 19 Min
Average 3.5640
Post Vent:
June 1st 2025 to now
170 data points
Longest: 24 minutes.
Average: 5.1823 Min
Post vent break down.... break down:
Jun 1st to Jun 16th
Length of data 100
max Duration: 24min
Average 5.04 min
Jun 16th - now
Length of data 70
max Duration 13 Min
Average 5.38571
Link to a google sheet with all the exported data, and GPS times.
I copied Tony's awesome spreadsheet, and replotted the data sets while thinking about what they mean.
I have the same 4 data sets that Tony has (Jan-April, All of June, and then broken into Early June and Late June, with the divider being the time that I enabled the 'slow let go' of the BS pitch control). However, I've got all the x-axes fixed to be 0-25 minutes. I've also set the y-axes to be (0, number lock segments), so that they are roughly normalized. In the subtitle of the plot I note the percentage of the segments who are 10 mins or longer (actually, from the data set, the percent that have a value of 9 mins or greater). Since we have a 10 minute timer in the guardian that will flip over to trying PRMI or MICH locking, this percentage should help capture the number of locks that take a long time to acquire DRMI.
Notably, the number of segments that take a long time is about 2x larger after the BS slow let go was enabled, if we look at the percent in late June (48% take a long time) versus the percent in early June (28% take a long time) :( But, both of these are much higher than the 18% that took a long time before the vent.
This may mean that the slow letting go of the BS, as currently enabled, is not helpful.
If the statecounter.py code is able to, it could be interesting to get similar statistics, but have the durations start when we leave state 18 (Arms_off_resonance) and the duration end when we get to state 102 (DRMI_locked_check_ASC). That would enable us to more accurately see the total length of time it takes during an acquisition sequence. If we do this, we'd want to count and then exclude from the statistics the number of times we 'give up' and lose lock or do an initial alignment.
