Added new ST2 SEI configuration nodes today. These are very similar to the ST1 nodes but perhaps a bit more simple. As of right now they are not managed by the SEI_CONF manager nodes, but will be once Jim W gets them all set up where he wants them.
Screenshot of the Guardian overview with the new nodes.
Daniel, Dave:
new h1psliss code was installed and the DAQ was restarted. Main DAQ change was the addition of 9 fast channels at 2048Hz to the commissioning frame:
H1:PSL-ISS_OUTERLOOP_PDSUMINNER_OUT_DQ
H1:PSL-ISS_OUTERLOOP_PDSUMOUTER_OUT_DQ
H1:PSL-ISS_OUTERLOOP_PD_CAL_OUT_DQ
H1:PSL-ISS_OUTERLOOP_RIN_AC_COUPLING_OUT_DQ
H1:PSL-ISS_OUTERLOOP_RIN_CTRL_OUT_DQ
H1:PSL-ISS_OUTERLOOP_RIN_ERR1_OUT_DQ
H1:PSL-ISS_OUTERLOOP_RIN_ERR2_OUT_DQ
H1:PSL-ISS_OUTERLOOP_RIN_INNER_OUT_DQ
H1:PSL-ISS_OUTERLOOP_RIN_OUTER_OUT_DQ
I have created a new GDS calibration FIR filters file. The included filters are designed to correct output of the front-end CALCS calibration, improving accuracy at high frequencies on ~10% level. This new filters file contains more aggressive high-pass filters with a cutoff of 9 Hz and contains accurate values for EP1-9 for use in the kappa calculations. The new filters file can be found in the calibraiton SVN:
aligocalibration/trunk/Runs/PreER10/GDSFilters/H1GDS_1156183417.npz
This filter was created using calibration SVN version #3295 and following instructions outlined here.
Attached are plots of the corrections applied to DELTAL_RESIDUAL and each of the DELTAL_CTRL_PUM/UIM/TST/ channels (all DELTAL_CTRL_* channels receive the same corrections). Attached are also plots comparing the h(t) spectrum for some ER9 data (1152076416-1152076544) calibrated using this filters file with output of the front-end CALCS model. The agreement is at the expected level given the corrections include in the GDS filters.
The GDS calibration pipeline was restarted at 1157822672 with the above new filters file.
These filters include the use of a tukey window around the final time domain filter. This seemed to fix the spectral leakage issues seen by downstream tools. I believe the previous filters were not falling off smoothly enough and were adding low-frequency features to the h(t) data. This become noticable after switching to double precision channels, because the old dewhitening filters in use with the (whitened) single precision channels included the tukey window. Since these dewhitening filters were no longer in use when we switched to double precision channels without whitening, the issues with the residual and control correction filters became obvious.
Tonight we saw ETMX 15007 Hz MODE 18 ring up almost immediately upon increasing power to 50 W. It damped with no problems with -1k damping gain, no phase. Active damping for this mode has been added to the SUS_PI guardian. First attachment shows ring up and subsequent damping.
Not sure why this mode didn't ring up last night; the only thing that's different is that last night Kiwamu babysat the ITMX CO2 laser power. Second attachment shows last night, third attachment shows tonight. Upon arriving at 50 W last night, the CO2 power was ~0.08 W, versus tonight the power was ~0.39 W. However, SRC and PRC gouy phases are both very similar at that time.
Apparently the PEM EY wind sensor EPICS IOC is malfunctioning causing the H1:PEM-EY_WIND_ROOF_WEATHER_MPH channel to be dead. This channel is checked in one of the DIAG_MAIN PEM tests, so the dead channel was causing DIAG_MAIN to go into a connection error state, thereby not executing any of the diagnostic tests.
I temporarily bypassed the test with the faulty channel, replacing it with a notification that the sensor channel is dead. This change should be reverted once the channel is restored.
While I was at it, I did a little maintenance on DIAG_MAIN:
Revomed the PEM_WIND test. After talking with Dave it doesnt seem too useful when we have other tools in the control room to show wind. Plus the connections to the weather stations seem to be unreliable and we dont want to stop the DIAG_MAIN from showing notificaitons just because of a connection error at a weather station, especially in an observing run.
Jenne, Evan, Sheila
Since Saturday (after RH settings were changed and we recovered from the power outage) we have been having sudden locklosses in the Analog CARM transition and the next few states. We measured the CARM open loop this afternoon and saw that the gain was too high leaving us with too little phase margin (blue trace in screenshot). We reduced the gain in the common mode summing junction by 8 dB (from 16dB to 8dB), green trace in the attachment. Both measurements in the attachment were taken at DC readout.
Evan's model of the CARM loop can be found at 22188
Since lowering this gain we haven't had any more of the locklosses shortly after switching to analog carm.
We lost lock a few minutes ago with a BS ISI trip (COIL DRIVER chassis BIO) We aren't sure what that is but everything looks fine on the CDS overview.
This trip has just happened again. But, this time we aren't able to fix it. Sheila went to the CER to look, and all the lights on the coil driver chassis are green. No overtemps are on, and fuse status is green. We're calling Hugh right now, but if anyone sees this and knows how to fix it, please give the control room a call.
We tracked down the BS ISI trip to a single bit flip in the h1seib2 DIO_0 card, corresponding to the BS ISI ST1 V3 coil driver:
H1:ISI-BS_BIO_IN_CD_ST1_V3_STATUS
This bit being true was causing the ST1 watchdog to be permanentaly tripped.
We tried without success to clear the problem with the following:
None of that cleared the bad bit. Eventually Sheila checked the binary output from the coil driver and determined that everything looked good. This gets the coil driver off the hook, narrowing the problem down to the BIO chassis or the BIO card itself in the IO chassis.
Given that we determined there was no actual problem with the coil driver, we decided to just bypass the problem by using the H1:ISI-BS_BIO_IN_BIO_IN_TEST EPICS channel to add in a value to the BIO input that would flip the single bad bit back to the desired value. This solves the problem for the time being
Once the BIO chassis/card is fixed, the H1:ISI-BS_BIO_IN_BIO_IN_TEST channel will need to be set back to zero (0). See the "BIO" screen from the ISI BS overview screen.
Unfortunately this bad bit is actually periodically flipping, and not staying flipped, so our test sum-in hack solution is not working. We've had maybe two more locklosses caused by these bit flips.
1720 -1735 -> To and from Y-mid LN2 @ exhaust after 32 seconds with LLCV bypass valve opened 1/2 turn -> Restored valve to as found state. Next CP3 overfill to be Wednesday, Sept. 14th.
Evan H., Jenne, Matt, Kiwamu,
We locked the interferometer at 50 W with the latest ring heater setting (RHX = 0.5 W, RHY = 2.5 W, 29588). So far the interferometer has been locked at 50 W for roughly 1 hour.
P.S. we have now having difficulty damping a PI mode (mode2, 15520 ITMX) and decreased the PSL power back to 30 W.
[DRMI lock on POP sensors]
As reported yesterday (29601), we had a difficulty in switcing the sensors from the 3fs to 1fs earlier today (29603). In the end, I manually executed the sensor switch process one by one and for some reasons this was successful. I then measured the open loop transfer functions of the DRMI LSC degrees of freedom, but they looked OK. See the first attachjment. This may be due to that we did not wait for long enough time to let the new CO2 setting settle (29603) for lock acquisition ([CO2X, CO2Y] = [500 mW, 1000 mW]). Not sure.
Also, in lock acquisition, I manually kept aligning PR3 when the interferometer was at ANALOG_CARM in order to maintain the lock.
[CO2 tuning at 50 W]
This is not well tuned, but the below is an OK CO2 tuning for 50 W which gave us a 30% imbalance at the AS port OSA.
[CO2X, CO2Y ] = [300 -400 mW, 0W]
This time I did not spend time for tuning the dCO2 at 20 W or 40 W. Maybe I should have done that to collect more data points.
[PIs]
MODEs 17 and 27 needed a sign flip. Mode 27 seems tricky -- every time when it rang up we needed to flip the control sign.
The PI difficulties are probably my fault. The phases may not have been set properly after the power glitch.
After damping PIs (Terra is writing about this) we tried to take the IFO back to 50 W, but lost lock within a few minutes. I'm not sure why we lost lock, it didn't seem to be PI or the usual sidebands tanking problem, there was a glitch in CHARD and DHARD a second before lockloss (1st attachment).
We spent some time relocking and found that we had some problems durring the latter CARM offset reduction steps which were similar to what we had last night. We measured the DARM loop at RF DARM and at the state CARM 15 pm, it seems fine (2nd attachment).
It is possible that our relocking difficulties were due to bad alignment, or the new TCS. As we were about to try to relock the 3rd time, we got hit by an EQ so we stopped for tonight.
Here is some analysis of the sideband imbalance.
Synopsis -- Overall, adjustment of the CO2 lasers I did at 50 W improved the sideband imbalance at the AS port
However, the resulting CO2 setting does not match what we expected from the previous test (29585). We expected the final CO2 settings to be [CO2X, CO2Y] = [100 mW , 0 mW], but we ended up with [400 mW, 0 mW].
The 45 MHz seems to prefer a high CO2 contrast of roughly 400 mW (CO2X minus CO2Y) regardless of the ring heater settings so far, and this prevents us from further reducing the common lens.
[The sideband evolution as seen by the OSA]
The below shows a plot of the OSA raw output. I plotted several scans from different times, each of them is separated almost by 10 minutes in time.
Also, here is a rough time line of what I have done in this lock.
Here is another plot showing how the sideband amplitude evolved as a function of time.
It is evident that increasing CO2X helped reducing the imbalance. Before I started changing CO2X, there was a slow trend in which the imbalance kept decreasing.
Here is another trend plot.
The carrier recycling gain stayed at 29 at 50 W. The sideband build up or ASAIR90 seemed to have reached some kind of equilibrium after approximately 1 hour or so. Also, looking at ASAIR RF 45 Q, I don't see any change in the DARM offset point -- hopefully this is an indication of a stable optical gain. The Pcal line at 331.9 Hz was too small that it was burried below intensity noise. So I could not directly check the DARM optical gain.
The intensity noise coupling changed as a function of time at 50 W. Here is a plot showing several coupling transfer functions from different time.
The measurement time were set to identical to the ones showed in above or 29637. Here is a plot of the transfer coefficient at 400 Hz as a function of time.
As you can see, the coupling became worse at the beginning for about 20 minutes or so. Then it came back to a value as small as the very begining. Although high frequency above 100 Hz seems to have settled to a small coupling, the low frequency part looked worse at the end. See the first attachment.
As for the HWS signals, while the HWSX signal seems reasonable, the HWSY reports a twice large self heating.
Here are some conclusions from comparison of the predicted ITM lensing (outputs from TCS-SIM) and HWS outputs.
The below shows a comparison between the single-pass defocus measured by HWSX and that predicted by simulator (with some coefficient fitting similarly to 27330):
I added a constant offset to the simulated defocus in order to plot it on top of the measured defocus. I removed the ring heater component that was unfortunately in the middle of settling to the equilibrium in the simulator due to the unscheduled model restart (29592).
The below is a same plot, but for the ITMY substrate defocus.
As noted in the legend, the component from the self heat needed to be increased by a factor of two.
The stability of AS90 was apparently better with the combination of the new ring heater setting and dynamically-decreased CO2s.
The below plot shows a comparison of this lock stretch (Sep. 12) and two lock stretches (Sep. 09) from a day before we started changing the ring heater setting.
The two lock stretches from Sep. 09 both showed AS90 decreasing monotonically as a function of time. This was something we have been suffering from (29486) and has been an issue since the begining of this month (29457) at which we considerably changed the interferometer alignment.
In contrast, AS90 settled to a value pf 600 counts on a time scale of 30 minutes or so in the new lock stretch. This lock was stable until PI modes started ringing up. The carrier recycling gain was slightly lower than the previous two stretches. I don't know if this due to misalignment or new ring heater setting.
Here are the evolution of the wavefront gradient as a function of time as seen by the Hartman sensors.
The starting time is set to Sep 12 2016 00:30:30 UTC at which the interferometer was in the middle of powering up to 50 W. The gradients are initilized such that the gradients are zero at dt = 0.
The GDS calibration filters were updated for ER9 on July 7, 2016. These filters are designed to correct the output of the front-end CALCS calibration model.
These filters were generated using the script run_td_filters.m located in the calibration SVN under aligocalibration/trunk/Runs/PreER9/H1/Scripts/TDfilters with SVN version #3176. Information on the exact parameters files used to generate these filters and the version of DARMModel used for these filters can be found in the run file run_td_filters.m.
In addition to the residual and control chain correction filters, the updated filters also include dewhitening filters that are unity since the pipeline is now run by ingesting double precision channels that are no longer whitened. The next generation of the GDS calibration code will not require dewhitening filters to be present in the filters file.
The new filters file can be found in calibration SVN under aligocalibration/trunk/Runs/ER9/GDSFilters/H1GDS_1151960706.npz.
Attached are plots of the residual and control correction filters. These plots compare the frequency response of the GDS FIR filters to the true frequency domain model they are based on.
A few notes I forgot to mention above:
