3:37UTC
4:09UTC
5:37UTC
.3micron Alarm also at EX
4:31UTC
5:03UTC
Following up on yesterday's restart of CW HW injections with a new actuation scheme, here are comparisons over 24-hour intervals of the excitation channel H1:CAL-PINJX_HARDWARE with what it was previously when a time-domain inverse actuation filter was used. One benefit for transient search groups is that if sporadic CW injection dropouts are seen again in O2, they should not induce nasty glitches in DARM (see figures 11-13 below). The bottom line for CW searches is that things look close to what is expected, but the amplitude of the highest-frequency pulsar injections (above 1 kHz) are significantly lower than before. The small residual discrepancy does not seem to be explained by the difference between the old and new inverse actuation filter curves that Evan G. posted yesterday. Perhaps both the old and new inverse actuation filters simply amplify the 1000-2000 Hz band too much (by 20-30%)? The figures below show 24-hour second-trend plots of the excitation channel envelope and 4-minute spectrum snapshots taken at 6-hour intervals, along with samples of sudden shutting off of the injections. Figure 1 - 24-hour trend (min/mean/max) of the channel for old actuation, showing the envelope of injections, which is affected by the rotating antenna pattern of the interferometer w.r.t. 15 different points on the sky with various intrinsic source polarization and strengths. Figure 2 - 24-hour trend for new actuation - one can see a small drop in amplitude, driven by the highest frequency pulsars for which the inconsistency between old inverse actuation filter and new actuation function is largest Figure 3 - 4-minute spectrum at 22:34 UTC on May 24 (old actuation) Figure 4 - 4-minute spectrum at 22:30 UTC on May 25 (new actuation) - approximately one sidereal day later Figure 5 - 4-minute spectrum at 04:34 UTC on May 25 (old actuation) Figure 6 - 4-minute spectrum at 04:30 UTC on May 26 (new actuation) - approximately one sidereal day later Figure 7 - 4-minute spectrum at 10:34 UTC on May 25 (old actuation) Figure 8 - 4-minute spectrum at 10:30 UTC on May 26 (new actuation) - approximately one sidereal day later Figure 9 - 4-minute spectrum at 10:34 UTC on May 25 (old actuation) Figure 10 - 4-minute spectrum at 10:30 UTC on May 26 (new actuation) - approximately one sidereal day later Figure 11 - Glitch induced by sudden shutoff of CW injections with old inverse actuation filter Figure 12 - Vertical zoom of glitch Figure 13 - No glitch induced by shutoff of CW injections with new direct application of inverse actuation function Note that the new trend (Figure 2) is a little smoother than the old one, as expected, without the amplification of tiny glitches seen with the old inverse filter. Another manifestation is the much cleaner noise floors seen in the new spectra, away from the injected lines.
At Rick's request, I am attaching more information about the desired injection strengths. Attached are a time series plot and a csv file for 10 seconds of H1:CAL-PINJX_CW on May 24 when the old time-domain IAF was in use, along with a spectrum and csv file for a minute, starting at the same time (rectangular window, no overlap, amp spectrum - not density). Graphs and files were generated via ldvw. This sample of May 24 data starting at 22:34 UTC corresponds closely to what should have been injected on May 25 at 22:30 UTC, i.e., the first pair of spectral snapshots above.
Peter in PSL enclosure 14:55 UTC Jeff B. moving chiller with forklift 15:13 UTC Lock loss after sitting on LOCK_ALS. A short time after the ISI ITMX stage 1 and 2 watchdogs tripped. 15:23 UTC Filiberto to beer garden to take measurements for running cables 15:33 UTC Jeff B. done 15:43 UTC Filiberto done 15:56 UTC Peter to LVEA to toggle NPRO noise eater switch 16:00 UTC Peter back 16:12 UTC Peter to PSL enclosure 17:10 UTC Fire department truck through gate 17:22 UTC Betsy looking for parts in LVEA west bay 17:35 UTC Jim W. to end X to install anemometer in desert (WP 5902) 17:32 UTC Peter done 17:44 UTC I toggled the NPRO noise eater switch 17:46 UTC Dave loading ASC filter modules 17:55 UTC Started attempting to lock 18:09 UTC Lock loss on CARM_5PM 18:28 UTC Lock loss on ENGAGE_ASC_PART2 18:47 UTC Re-requested READY_FOR_HANDOFF for OMC_LOCK node in order to get to DC_READOUT The ISS first loop did not engage because the PSL AUTOLOCK was off. Kiwamu turned it on. 18:50 UTC Jim W. back 19:55 UTC Tour leaving control room 20:06 UTC Travis to end Y to reset camera 20:44 UTC Have been sitting at DC readout. Commissioners back. Evan H. changing 9Mhz modulation depth. 20:47 UTC Travis back 21:24 UTC Evan H. done Going to NOISE_TUNINGS for Sheila and Terra to run measurements Got stuck on increase power. Sheila changed guardian code to continue. 22:11 UTC Kiwamu and Nutsinee to CO2 Y table to remove beam block 22:15 UTC Jim W. to end X to retrieve anemometer from desert 22:44 UTC Kiwamu and Nutsinee back. They did not remove the beam dump. 22:47 UTC Jim W. back 22:53 UTC Kiwamu and Nutsinee back to CO2 Y table
The laser power guardian checks that the laser power is within 15% of the requested power, and doesn't return true unless it is. The rotation stage isn't always this accurate (right now the requested angle matches the measured angle, but the requested power is 12W instead of 10W). Now that we are powering up in multiple steps, this means that the guardian gets stuck with a warning to adjust the rotation stage by hand or search for home.
We edited the threshold that is used to check the power, so now it will return true if we are within 25% of the requested power.
Following a two month hiatus we have gotten the daily archival of H1 running DAQ and filter files running again.
This runs as user root on script0, as a cronjob running at 27 minutes past the hours of 9,4,8,12,16,20.
Many thanks to Evan for reminding us that this service was not operational.
After several loads of the ASC DHARD filter, the archived H1ASC.txt file increased in size from 4,000 lines to 117,000 lines overnight. Now that the DAQSVN archiver is running again, this becomes an issue when someone tries to display the differences in the loaded filters. These problems can be avoided by doing a full filter load (from the GDS_TP MEDM screen) if you see the difference list getting out of hand.
Attached are 7 day pitch, yaw, and sum trends for all active H1 optical levers. Big dips on Tuesday May 24 caused by the recentering of the OPLEVS.
All the DBB cables were checked that they were connected properly (they were). The incident power into the diagnostic breadboard was adjusted to be ~110 mW - it really should be 135 mW. However at this lower power level, the interlock still tripped. The interlock tripped even though the shutter did not open. I blocked the input hole to the diagnostic breadboard so that no light could enter. Under this circumstance the interlock still tripped. So obviously it wasn't a light level associated problem even though the interlock message indicated it was. I turned off and then on the two AA and AI chassis associated with the diagnostic breadboard. This seems to have reset the problem. That said, there's still some work to do to bring the diagnostic breadboard truly back.
The signals for the Diode Room chiller flat lined at 23:09 on the 22nd. Need to reboot the H1PSLCTRL0 computer. As this will take down the PSL will do the reboot early Friday morning.
Sheila, Keita, Haocun We took several more measurements on 90MHz signals at AS port yesterday. In summary, the noise at 90MHz is really large compared to signals. Even when the MC was unclocked, the spectrum from RF IN had a peak of -88dBm at 91MHz, and the signals were around -71dBm both at RF in and RF MON when the IFO was locked. There are large signals at 54MHz in RF IN (-45dBm), and -68dBm in RF MON. We put in an 91MHz Bandpass Filter (Lark Engineering MC91.5-H2.5-3BA), but this did not help with the noises at 91MHz. (RF IN: -78dBm; RF MON: -70dBm @91MHz)
To rephrase the above and add some:
Forget about RF MON for 90MHz for now as it's dominated by the 90 MHz LO leakage/pickup or something inside the demod chassis. Even when the RF input on the demod chassis is terminated we see -71 dBm in RF MON at 90 MHz.
RF level from the WFS at RF IN on the demod is about -71 dBm when IFO is locked (and -88 dBm when MC is unlocked, which is 17 dB lower than locked and is therefore negligible). This sounds small, but we have a 90MHz notch on the WFS head itself.
54MHz signal is much larger than 90MHz in WFS 90MHz output (-45dBm for 54MHz VS -71 dBm for 90MHz on RF IN of the demod). It wasn't clear if this was doing anything bad, but inserting 90MHz BP between the triplexer and the demod didn't change the demod output significantly. It's not clear to me if BP properly blocks 54MHz when triplexer assumes 50 Ohm resistive load and BP adds complex load, though.
Last night, we ran a quick TCS test where we attempted to minimize the intensity noise coupling to the DCPDs by changing the CO2 differential heating.
It seems that the following CO2 setting gives a much better intensity noise coupling when the PSL power is 25 W:
This did not improve the recycling gain so much. It seems to have increased by 2% only.
According to a past measurement with a lower power PSL of 2 W (alog 26264), a good differential CO2 power had been found to be P_{co2x} - P_{co2y} = 270 mW (or probably less than 270 mW because I did not explore the lower differential power).
This could be an indication that ITMY has a larger absorption for the 1064 nm light such that the differential self-heating linearly changes as a function of the PSL power. We should confirm this hypothesis using the HWS signals.
[The test]
No second or third loops engaged, DC readout, no SRC1 ASC loop.
Drastic reduction of the intensity noise coupling was observed mostly between 4:24 and approximately 5:00, indicating that reducing the CO2Y power helped improved the coupling. After 5:00 UTC, we did not see a significant reduction. This may mean that we might have been already close to an optimum point where the coupling is minimized. The attached shows DARM spectra from various time during the test.
A broad peak at around 400 Hz is my intentional excitation to the first loop with band-passed gaussian noise in order to check the coupling to the DCPDs. As shown in the spectra, the reduction from the beginning to the end of the test is about a factor of 5. As reported in 27370, broad noise above 100 Hz up to several kHz is indeed intensity noise and therefore we see the noise floor in this frequency band decreasing too.
Because of the error in picomotor assignment, there's a good chance that the CO2Y laser is severly clipped and not properly aligned to the test mass. The result of this would be strong higher spatial order lensing (non-quadratic) on ITMY. We're certainly seeing an excess in lensing as measured with the HWS but the exact nature is unclear.
If there is signifiant higher order mode lensing then the best differential effect will be from having the CO2Y laser set to zero. However, this will not be the optimum lensing for the power recycling cavity.
So, poor CO2Y lensing is at least consistent with and a plausible explanation for requiring 0W on CO2Y to minimize intensity noise coupling while observing reduced PRC gain.
I looked back the intensity coupling of this particular day. See the attached.
The coupling at 100 Hz, even though the coherence is not high, seems to be too high by a factor of two or so comparing agaist the measurement from this February (25476). When dCO2 was sort of adjusted (red curve), the coupling at around 400 Hz and above seems comparable to what it was in this February.
Kiwamu, Sheila, Hoacun, Evan
We've worked on all four hard loops today.
First we designed a new compensator for CHARD P, which roughly inverts the plant at 2W and should be stable all the way to 50 Watts according to the model. (Screenshots of filter change and before and after OLG measurements attached.)
We tried this for DHARD, but it was not stable durring the CARM offset reduction. Instead we made a filter that is the ratio of the new compensator to the old, which is engaged on resonance.
Then we had a look at the hard yaw loops, and adjusted the boosts in both loops to move the zeros to higher frequencies.
This is all in the guardian.
Things not in the guardian:
We have seen several times that engaging the ISS 3rd loop supresses the CSOFT P instability (engage second loop output, set 3rd loop gain to 0.04). I tried once to engage it at 20 Watts, and it broke the lock.
When we approach 40 Watts, there is an instability at nearly 2 Hz in yaw that breaks the lock. It is clear from the transfer function of both CHARD and DHARD yaw that there is a problem here. I tried a lag filter (FM4 in both DHARD and CHARD Y) to avoid the instabilites. These can come on by hand at 20 Watts, and do avoid the instability but just shift the problem to little bit of a lower frequency where we have a little gain.
After adding these filters I became unable to get test points using DTT (the test points were all full for ASC, so I cleared them, but I still couldn't get anything from dtt. I tried closing my dtt session and starting over, but that didn't work either).
I tried powering up one last time, it looked like an instability in DHARD P was the problem this time.
I turned off the XARM ring heaters to prepare for a PI test tomorow.
Related alog: 27349
I have coarsely adjusted the demodulation phases for the 331.9 Hz calibration line in CAL CS. Now the CAL CS model returns a live cavity pole value in unit of Hz. The first attachment shows the current filter settings. The second attachment shows where to look for getting the live cavity pole value. This screen is linked from CAL_CS_OVERVIEW.adl; clicking a pink box in the upper left takes you to the screen shown in the second attachment. I have adjusted the demod phase of the Pcal line to get the cavity pole between 340 and 360 Hz. Also, I set the demod phase of the DARM error demodulation to 180 deg so that the imaginary part of the calculation is negative. We still need to do a careful measurement of the cavity pole with the usual swept sine to make the live value more accurate, but for now this should be good enough for commissioning purpose. I believe that the current accuracy of the live cavity pole frequency is about 10%.
Additionally, I have bumped up the excitation amplitude of the 331.9 Hz Pcal line by approximately a factor of 10 for obtaining a high SNR. I will leave the excitation amplitude as it is.
At 05:04UTC
BSC5 annulus IP is flickering between red and green state. BSC8 aIP was in the red a few days ago, but seems to have recovered.
Someone told me BSC5 was not connected.
I just added BSC5 and BSC6 to the overview this morning (alog 27377).
Oh! So these channels are now available. We just need to run cables from the pump to rack.
Transitioning over to the new, improved inverse actuation filters (see LHO aLOG 27176), I removed the old filters from PINJ_HARDWARE and installed the new inverse actuation filters into the PINJ_TRANSIENT bank. Since the CW injections will apply the inverse actuation function to condition their signals before sending it to the PINJ path, we don't want the filters in PINJ_HARDWARE any longer. CW injections are running right now using this new actuation function. Keith R. and I made a test of this earlier to verify the output was roughly correct (Keith will post an aLOG about our work). He wants to see this running over the next few days to gain confidence that the injections are operating normally.
Reference to Keith's aLOG regarding today's CW injection test. Attached is a comparison of the old and new inverse actuation filters. Note the discrepancy at low frequencies because the old version assumed only a free-mass response instead of the real force-to-length suspension model. Also, different approximations were made at high frequencies leading to different magnitude and phase effects. Recall the new filter is valid within 5% in magnitude and 5 degrees in phase up to 2 kHz, when taking into account that waveforms must have an assumed 200 us advance.
Keith R. noticed that the CW injections appeared to be ~10% smaller in amplitude now that they are no longer passing through the old inverse actuation filter. To see if the new design is the cause, I plotted the inverse actuation transfer function in Foton and plotted the results of the old/new. Note that there is also an old AI2 filter module that didn't appear to be in-use lately. In any case, I plot all three curves in the attached plot. Importantly, the inverse actuation transfer function is larger in magitude for the Old version compared to the New version by about 10% at 2 kHz. Meanwhile the Old w AI2 is smaller than the New version. This difference is the likely explanation to Keith's observation.