TF crashed (watchdog trip) last night at 6pm local. Very weird. The attached plots show the details. First plot shows all the excitation channels for a 3 hour period around the watchdog trip, you can clearly see where we were driving at ~3000cts on each exc channel (this wasn't happening at the exact moment of the trip. The second plot shows the V4 excitation channel and the V4 L4C that tripped over a 100 second window. There is a very low drive, but high frequency excitation that does not look like the other parts of the transfer function being run at the time. But, this excitation is also roughly at the same time as when we were driving on a different (H1) channel.
Fabrice was able to copy the signal from the V4 excitation channel at the time of the trip and send it again with some matlab trickery, but we saw nothing on any of the chamber's (HEPI, ISI, SUS) sensors. So we still don't know what caused the trip. Weird.
The injection log file shows that a Matlab process on opsws1 started an awgstream excitation on channel H1:HPI-ETMX_OUTF_V4_EXC at GPS time 1071453844, for 1011 seconds. The monitoring of H1:HPI-ETMX_OUTF_V4_EXC_DQ was started 7 seconds earlier at 1071453837.
Following the DAQ restart to service the h1sushtts model change, I noticed that h1lsc DAQ status changed to 0x2000 (new ini file). I pressed the "DAQ Reload" button on h1lsc which cleared the error, and then realized that there is most probably a mismatch between the lsc model and its INI file. True enough, the model was started on Tue 17th but the model was changed and a "make install" issued Wed 18th. At this point the lsc data is most probably corrupted in the frame.
I restarted the h1lsc model to bring the software and the INI file back into synchronization.
Mea culpa. The LSC model was rebuilt to fix the channel names of its antiwhitening switches, but I postponed the restart since the model was in use last night. Thanks for taking care of it.
J. Kissel, J. Rollins, A. Staley In order to align the in-vac REFL path, as well as address proper spacing of the RM's eddy current damping copper block / magnets JAmie and Alexa need the RMs steerable and dampable. However, with the battles I've been having getting the HTTS model up and running, I hadn't noticed the obvious -- if we connect the OMs (whose models are down and out, and screens are not-yet functional) to the USERDACKILL, which blocks all HTTS drives (including the RMs) -- one cannot operate the RMs. I have subsequently, temporarily removed the OM1, OM2, and OM2 inputs to the USERDACKILL, recompiled, reinstalled, restarted, and restored the h1sushtts on the h1asc0 front end. Now, the RMs can damp and actuate as desired. Note -- I'm still not finished with the HTTS and HAUX model modifications; please bare with me in this time of flux. I hope to get through with this by the end of the week.
I lowered the locking threshold this mornig because IMC wasn't locking well. The followings are the current values:
H1:LSC-MC_FM_TRIG_THRESH_ON = 2000
H1:LSC-MC_FM_TRIG_THRESH_OFF = 2000
H1:IMC-TRIG_THRESH_ON = 2000
H1:IMC-TRIG_THRESH_OFF = 2000
H1:IMC-DOF_FM_TRIG_THRESH_ON = 2000
H1:IMC-DOF_FM_TRIG_THRESH_OFF = 2000
We will watch for changes in the alignment onto the ALS refl PD.
- IO_AB_PD2 and IO_AB_PD3 are now wired from the PSL table to the PSL rack. IO_AB_PD2 has light, IO_AB_PD3 is not yet in place.
- IO_MB_M2 was swapped from the 2" 10% trans. optic to the 1" 2% trans. optic, BUT the 1" optic alignment and mount and optic all have issues, so the 2" optic and mount were reinstalled at the end of the day
--- optic mount (U100) needs 3 actuators to adjust the plane of the mirror - actuators interere with the blue mount
--- blue mount needs to be altered
--- 1" optic may have 2 small dings on the back side - I need to reclean it and verify
--- spare optic is completely screwed up with First Contact - has the old clear FC and would not come off - peek was pulling out o the FC - I applied red FC and a new peek tab, and still the clear FC is stuck to the optic, and of course it's stuck to the front side that will see high power - not sure how to save this optic right now
Alignment Irises I have on the table - with beam changes they are no longer centered, so here's the "code" for the good alignment
- High and left on my alignment iris after the bottom periscope mirror
- Centered on my alignment iris on NW corner of the PSL table
I'll center the irises after the new components are installed in the main beam.
To do list:
- install motorized rotation stage
- fix IO_MB_M2 mount and install 1" 2% trans. optic
- install IO_AB_PD3 - power before rotation stage
PRMI is finally aligned and it is now flashing. This should help us to find the POP beam because of higher intra-cavity power and reasonable alignment.
After PRMI was coarsely aligned, I went to ISCT1 and quickly looked for the POP beam, but no beam was found. I also toggeled the HAM1 POP beam diverter, but it didn't help.
The above plot is an exmaple of time series where you can see that ASC_POP_A goes up to more than 100 counts because of the recycling.
Alignment:
Sadly, we're finding that aLIGO OLs are not as reliable as OSEMs as a long term reference, just like in iLIGO. According to Kiwamu, all previous PR-X alignment efforts were with the BS angle restored using OLs, not M0 BOSEMs.
Richard says that OLs were fine in iLIGO as far as nobody touched anything.
Chris, Aaron, Filiberto, Daniel
Thanks to a very motivated crew we have fabricated 2 PD amplifiers for the ITMX and ETMX baffled diodes. The amplifiers are installed, powered up and connected to the slow controls system. The software for ETMX is ready. We still need to activate and test ITMX. The channel names are H1:AOS-ETMX_BAFFLEPD_1/2/3/4 and H1:AOS-ITMX-BAFFLEPD_1/2/3/4.
The ITMX channels are now up and running. The AOS channels are currently running on PLC3. These diodes in principle have remote gain adjust but we are waiting for some backordered parts to make this happen. For now one has to jumper a breakout board on the DB37.
I fixed the fast readbacks for the common mode boards and the power monitor PDs at the X end. (Their AA chassis was plugged into the wrong ADC.)
The "Red Path" PDs on the End X overview screen are not working yet. But they should be, as soon as Cable 86 (which carries them from the field rack to the AA chassis) is installed. It's on Fil's to-do list.
I remeasured the powers in the ALS fiber path, this time without a filter on the power meter.
10.15mW out of ref cav
95uW into path to fss_TPD and camera (TPD currently reads 0.98V)
9.8mW into ALS Faraday
9.08mW out of Farady (HWP is set to 51 degrees, QWP set to 101.5 degrees)
1.77 mW into fiber external PD (5.54V on fiber external PS readbacks) (before I had measured this with this filter on, and got 1 mW, that is why Alexa found a high responsivity for this PD)
7.2mW into fiber
FIBER internal (which monitors the power received from the fiber in the distribution box) reads 1.71, I just measured 4.78mW returning to the PSL when the two fiber ends by the distribution box are connected to each other (there is at lest 4.78mW getting ot the fiber distribution box).
Per Dave's recommendation, we have removed the cds_ezcaread "disconnected" at the top levels of our real time models (they were causing troubles to record the slow channels under the latest RCG version). Those were used in the early days to monitor the suspensions watchdog states. This feature is now done through IPC_CPIE but the old ezcaread were still in there. I found ezcaread blocks in all models except isiitmx and hpiitmx. Did not find any in the ham-isi and ham-hepi models. Dave said there's no rush to recompile for now (work permit will be issued when the models will be recompiled and installed) Side note: we still have to implement the suspension watchdogs monitoring through IPC in the hpiitmx model.
Today I "burped" GV7 and GV20's unpumped gate annulus volumes into aux. pump carts -> I also valved-in IP12 and IP5 (I had originally tried IP6 before IP5 but it didn't step its HV and had minimal net pump speed with its fixed HV value - I'll need to change IP6's controller settings to operate in "step" mode) -> If we do end up opening up the CS and X-end to the BT, I'll valve-out the respective turbos such that no rotating pumps are valved-in while the BT is exposed.
TF running on ETMX HEPI. Should complete sometime in the morning. Running from opsws1.
TF crashed (watchdog trip) last night at 6pm local. Very weird. The attached plots show the details. First plot shows all the excitation channels for a 3 hour period around the watchdog trip, you can clearly see where we were driving at ~3000cts on each exc channel. The second plot shows the V1 excitation channel and the V1 L4C that tripped over a 100 second window. There is a very low drive, but high frequency excitation that does not look like the other parts of the transfer function being run at the time. This excitation is also roughly at the same time as when we were driving on a different (H1) channel. It's unclear whether this is the cause of the trip or not.
Fabrice was able to copy the signal from the excitation channel and send it again with some matlab trickery, but we saw nothing on any of the chambers (HEPI, ISI, SUS) sensors. So we still don't know what caused the trip. Weird.
Today's Activities:
The 2 Thorlabs photodetectors on the PSL table monitoring the DC power after the EOM (IO_AB_PD3) and after the rotation stage (IO_AB_PD2) had their readback cables connected to the EtherCAT auxiliary signals concentartor 3 in rack ISC-R4. IO_AB_PD2 is connected to Extra1 (rear) and IO_AB_PD3 is connected to Extra2 (rear). Slow controls channels have been added to H1ECATC1 PLC1. They are available as H1:PSL-EOM_A_DC and H1:PSL-PERISCOPE_A_DC.
The two diodes were swapped (H1:PSL-PERISCOPE_A_DC had the signal from PD3, and vice versa) I swapped the cables for now. I calibrated this based on my measurements in alog 8963,
The power before pick off (POWERMON) channel is calibrated in mW entering HAM1. I added this channel to the MC screen, now we can use this to monitor how much light is going into the vacuum.