We reattached the ETMy upper structure to the ISI on the cartridge test stand this afternoon. We have a few more dog clamps to attach but can do those anytime. SEI can get to work there tomorrow if they see fit. We'll work elsewhere.
Kiwamu with remote help from Cheryl
As a part of the high power prep, I did three things this morning:
Stray light:
I looked around the inside of the IOT2L and ISCT1 enclosures with an IR viewer to look for a stray light which needs to be properly dumped. I did not find a stray beam and therefore I didn't have to newly install a dump.
Power budget:
See the attached. When I made the measurement on IOT2L, I intentionally misaligned MC2 in order to have high and stable laser power on the IOT2L. As for ISCT1, I misaligned PR2 to avoid a fringe while maintaining the PRM reflection. Note that the IMC incident has been 1 W (see alog 9306) these days and during these measurements. Because the POP beam was so dark, I didn't even attempt to measure the power. Instead, I made a back-of-envelope estimation of the maximum power in the POP path with an assumption that the power recycled Michelson is locked.
Power at POPAIR_A (or POPAIR_B) = (1 W incident on PRM ) x (60 power recycling gain) x (250 ppm PR2's transmissivity) x (90% BS in HAM1) x (50% BS on ISCT1) = 6.75 mW
Of course, in reality, it must be smaller than this due to losses and etc.
Search for PRM parking beam:
Because we didn't check the parking beam during the last vent and also readjusted the alignment of the input optics recently, we need to find a good alignment for PRM again which gives us a parking beam coming out through the top view port of HAM2. Since a CCD camera was already mounted on the viewport, I used it to see if I can find the beam. I tried the good PRM biases (P=-330 and Y=-6100) that was established back in April (see alog 6168). Also, I enabled AWG and shook PRM in both pitch and yaw in order to somewhat expand the parameter region that I am scanning. However, I didn't find a beam or any sign of beam on the monitor. Also I turned the illuminator on and off to see if the CCD camera is actually looking at something useful, but the CCD kept showing a dark background regardless of the illuminator.
I am concluding that we need to remove the camera canister to see how the camera is doing and directly check the parking beam with a laser card or IR viewer.
Input filters were loaded and installed on HEPI-ITMX yesterday at 12:45pm
Matrices were updated on HEPI-ITMX from the 'X' direction convention to the 'Y' direction convention, yesterday at 12:54pm.
Position loops are not installed on this unit yet. Previous transfer functions were measured in the local basis, before the input filters. Hence, those changes, which were needed, should not impact commissioning activities.
(Corey, Doug, Sheila)
While doing the PSL Checklist today, I noticed that we were getting intermittant "Check Diode Chiller" WARNINGS on the SYSSTAT.adl window. Sheila and I went to the Diode Room to check. There's a spout on there with Min/Max lines on it (it was closer to Min, so we topped it off). Peter King actually mentioned that this is spout is for the Crystal Chiller.
The lower LCD Display still was flashing with "level warnings", and Sheila said it was probably for the lower resevoir which has a big cap (this one is the Diode Chiller).
I emailed the PSL group and Doug ended up going out to fill the Diode Chiller resevoir at around 11:30am. He said it took 4 gallons to top it off!
And Operator Note: He used distilled water (NOT deionized water from lab---because it is corrosive).
(Alexa, Daniel)
With the cavity unlocked, i.e. only the straight-shot green beam, the power at the bottom of the periscope was ~12uW. The power at the green trans PD was ~2uW (after an 80-20 BS).
We also determined that the trans PD cables inside the table to the feedthrough were swapped between X and Y. That has now been fixed.
Attached is a coherence plot. Basically all X motions on the ISI on the two platforms have similar coherence with their Optical Lever Pitch. Interesting though, the ITM with a 0.56Hz notch has much broader coherence below this frequency whereas the ETM has no notches and much less coherence elements below 1Hz. Maybe there is area here to improve.
Starting at 12:01pm.
Work completed on HAM4 at 12:30pm. Model was recompiled and restarted. Hugh and Mitch are now plugging the sensors in to check out the model wiring.
HAM5-ISI model modifications are on-going. h1isiham5.mdl will be restarted once changes are confirmed on HAM4-ISI. Models will then be commited to the SVN and WP will be closed.
We connected the following cables in EX:
The RT link is now working and can be used to engage the boost filters for the arm cavity feedback to HEPI/SUS. The PD readbacks of th red path are now in the online system.
Well I got fooled here and at the moment I don't have the data to say anything. Referring to alog 2973, I said the loop closing on the HEPI position quieted the noise from the IPS that appeared to be getting to the T240s on the ISI stage1. However, Friday night when Jim, Fabrice and I were closing these loops, Jim reduced the gain on the T240s; I failed to confirm they were put back to their nominal high position. So the trend I posted indicating the great reduction in noise was only the reduction of the T240 gain. Yesterday afternoon I put the gain back to where it should be.
The attached 10 day plot shows the new story. The last part of the plot shows the period after the gain increase. It is smaller than the noisy period before last Fridays position loop turn on; but, remember Keita reduced the pitch request of HEPI and just by the scaling it may be completely due to this.
All is not doom & gloom though. All through the night, there were no T240 glitches and no watchdogs trips on either ISI or HEPI. Of course this could be completely attributed to the reduced tilting.
Heading to ISCt1
done at ISCT1 for now. X arm trans camera realinged, ALS light pipe opened and checked for stray beams from PSL path.
Stefan, Kiwamu
We did the following things tonight:
- We measured the BS, PRM and PR2 actuator transfer functions again and convinced ourself that they are not crazy. The BS actuator was found to have the opposite sign and 10 times bigger gain at 20 Hz.
- As a preparation for PRMI locking, we started using RELAIR_A_RF45. We found that it is 5 times smaller than RF9 in terms of the PRX optical gain. We could easily close the PRX/Y loop with RF45_I.
- Because we were not sure if we could close the MICH loop with BS, we went back to PRY to verify the behavior of the BS actuator in action. The idea is to optimize the LSC filters for MICH loop beforehand using PRY.
-We discovered that there was a length to pitch coupling in BS which caused 5-10 urad excursion in its angle. This was big enough to unlock the PRY loop.
- POPAIR_B_RF18 demod phase = 80 deg
Our message is that we need to put some more efforts on the BS actuator.
A comment on the attached plot:
They are the transfer functions from the suspension input to REFLAIR_RF9. The vertical axis is not calibrated to any meaningful units. The BS transfer function is artificially sign-flipped to see how much the phase gets delayed. Also we took a ratio of BS/PRM to figure out what kind of extra filters we need to close a loop with BS. You can see that the PRM and BS actuator responses are different and therefore we are working on BS.
We used the new prism protection shield on the PUM prisms during the welding of the ETMy monolithic in Dec and again this last week. We used the shield during the welding steps but not always during the lower power annealing steps, as was assumed by the users. We think the shield caused more stray beams from the welding laser than when we weld without it. We found more stray beam damage to curtains and garb than in any previous welding session, all of which we tried to mitigate. Possibly we could consider making the shield out of a different material than copper.
We also found that one of the prism glue joints suffered some heating, and subsequent charring, at the exterior of the prism when the shield was not in place during annealing. This occurred on only 1 prism. The other prism had the shield in place for all of the welding (as is standard) and for the annealing in Jan which is why we didn't see the charring. (We did not observe any stress fractures in the glass as was seem in the ETMx PUM!) So, the new shield has the following:
PRO
Prism joint is better isolated from heating/burning during welding.
CON
Shield causes more CO2 beam reflections which are difficult to manage.
Below are some pictures of the shield and the prism charring when the shield was not in place. Note, we had to cut a half circle view hole on each side of the prism for better visibility of the horn welding. Also, as shown in the pictures, it was difficult to get the shield positioned against the optic.
The good news is that our testing with the heat gun gave us great confidence in the stability of this epoxy even at high temperatures. My log book indicates that prisms started de-bonding at ~330 degrees C. And then only with shear force. The color of the heated glue was amber (see report at https://dcc.ligo.org/LIGO-T1300642 )
ISC length control offload path was added to HEPI-ETMX's model: h1hpietmx.mdl - See attached screenshot of the model.
The model was re-compiled and re-started - log attached.
DAC restart was required and performed by Dave.
Isolation and alignment offsets were turned back on, per the latest preferred configuration, and the ISC2CART matrix was updated. The outputs of the ISCMON block will be left off until ISC decides to try them.
Note: The signal offloaded to HEPI is the same as the one offloaded to SUS.
The new version of h1hpietmx.mdl was commited under the SVN -r6869
Work was performed under work permit # 4394 which is now closed.
Using a SUS control signal to drive HEPI will result in aliasing, since the SUS runs at a higher sampling rate than HEPI. This offload path should have its own IPC link, with an anti-aliasing filter installed on the sender.
also recall that drive in +x for HEPI ETMX will make the arm cavity longer, while the same signal applied to the mirror LONG input will probably make the cavity shorter, so watch the signs. ISC should pick a place to apply the signs so they get applied in a systematic way. I suggest that the ISC2CART matrix would make sense, and that a -1 in the LONG-in to X-out would make sense.
The large tilt requirement into HEPI for the TMS tilting really accentuated the IPS sensor noise when the HEPI loop was open. The HEPI had not been commissioned so the tilt was direct drive. On Friday Jim, Fabrice and I closed those loops and I did my best to restore the previous tilt to the system. We only got to within ~25urads cause the other loop closings caused more drive to the the V1 corner pushing the IPS to 29900counts with a watchdog trigger at 30000.
See the attached 8 day plot. The plot shows lots of things but mainly look at the lower right graph with T240 INMON. The first half of the plot is before the hepi position loops were closed. The quieter interval in the middle left, was during a noise study and we had no tilt (see the various hepi graphs). In the middle of the time frame we closed the ETMX HEPI position loops and you see the T240 channel go very quiet. There are no railing spikes that would have tripped the ISI. On Monday, Keita offloaded much of the HEPI tilt to the TMS because the HEPI IPS was close to its trip level and TMS needed a little bit more tilt. You can see the HEPI tilt change in the channel 9 & 12 graphs. However, there is no change in the noise at the T240 when Keita offloaded HEPI to TMS (I zoomed in and there is no difference in the time series view.)
Bottom line, The HEPI tilt will not cause ISI trips with the HEPI position loops closed. HEPI could take much of the TMS tilt back if needed.
See 9317--apparrantly not so uch, ah well.
