Wanted to post some info on the Injection situation.
5:23utc: Verbal Alarm says, "CW injections have stopped"
This put a red Excitation light on H1CALINJ on the CDS Overview. (Nothing is red on H1CALINJ when you look at its medm window.)
On the CAL_INJ_CONTROL2.adl medm (via sitemap/CAL/HWINJ CTRL), there's nothing RED or alarming for CW. When opening the CW button you get its filter bank, and on this only the OUTPUT is ON and all three OUTPUT channels are zero. A Time Machine of this CW filter bank from 24hrs ago shows same state for OUTPUT only being on, but you can see that the OUTMON does have a signal....this signal (and also the Calibration Injection Master Output signal) also go to zero at 5:23utc. Since there is nothing obvious I can do via the filter bank, I'm assuming this is an issue off-site and so I will leave this for others to address.
Attached is a screenshot of relevant screens to this:
- CDS Overview w/ RED light for H1CALINJ (and its green window below)
- H1 CAL INJ overview screen + CW INJ filter bank (time machined to 24hrs ago)
- H1 CAL INJ overview screen + CW INJ filter bank (current state)
- NDS Scope showing CW OUTPUT go to zero at 5:23utc
- INJ_TRANS guardian node + log
A check at LLO shows CW injections still going on (attached). Unlike the Burst/CBC injections the CW injection use local resources. Now perhaps they have run out of pre-generated data. Yes we need to update our screen to use FEC 42 instead of FEC 117 for injection monitoring.
I restarted psinject (CW) at May 02 2019 15:01:58 UTC. H1 was out-of-lock at the time.
To clarify, CW injections are computed on the fly, in 20-second buffers. There are no off-site file transfers or long-term storage of strain data involved. I believe Dave will be implementing an automatic restart of crashed injections, as exists at LLO.
Dave and I changed ours to CAL INJ TRAMP to 1sec.
LHO should add ramping to the CW excitation. At LLO, we are using 1 second Ramp Time for L1CAL-INJ_CW (first attachment). However LHO, has a 0-second Ramp Time for H1CAL-INJ_CW (second attachment). This is why the CW injection cut off so abruptly. Both sites have a 2-second Ramp Time on the TRANSIENT. This likely got missed in the upgrade to dedicated CALINJ model.
