craig.cahillane@LIGO.ORG - posted 17:48, Wednesday 08 June 2022 - last comment - 14:17, Friday 10 June 2022(63510)
June 2022 HAM1 incursion - Day 1 (PSL and REFL path power measurements)
Keita, Craig, Dhruva, Ryan S, Georgia Today we went into HAM1 and measured some powers on the REFL path using the integrating sphere from Rick and Francisco. We set up the integrating sphere such that it was connected to the PSL racks H1:LSC-EXTRA_AI_1_OUT16, which was calibrated into Volts in alog 63342 The requested input power was set to 1 watt. Sphere Calibration ------------------ Francisco emailed me the calibration of the sphere from V/W:-4.6786 V/W +- 0.11%at a temperature of299.87 K. This responsivity is for a 1047 nm beam, at roughly 246 mW in the integrating sphere. So for us, the responsivity at 1064 nm should be 1.6% higher, or-4.754 V/W +- 0.11%. Measurements ------------ Ryan and Georgia recorded the GPS times of our PSL and REFL path measurements. First we measured the high power input PSL beam that bypasses HAM1 on it's way to the IMC. This beam is 12.5 inches above the HAM1 table. We mounted the sphere on a crazy contraption to get this measurement. Next, we measured the high power beam promptly reflected from the PRM. Finally, we measured the lower power beams after the first 90:10 beam dump. Ryan made a diagram of the locations of these measurements based on DCC D1000313 (see Figure).Measurement Label GPSstart time Volts from Integrating Sphere -------------------------------------------------------------------------------------------------- PSL to IMC input A 1338750048 -4.42 Closed light pipe, same location A 1338750178 0.002 Out of REFL viewport, before M14 B 1338751500 -3.92 Closed light pipe, same location B 1338751605 0.002 Out of REFL viewport, before M14 again B 1338751753 -3.94 Reflection off M14 (90:10) C 1338751905 -3.62 Misaligned PRM, same location C 1338752032 0.002 Reflection off M17 (50:50) D 1338752190 -0.144 Reflection off M1 (50:50) E 1338752915 -0.0642 Misaligned PRM, same location E 1338753052 0.002 Reflection off M7 (HR) F 1338753209 -0.0727 Took lunch, during which there was some alignment problems with the PRM as we tried to lock PRMI without REFLAIR photodiode, and the HAM2 and HAM3 ISIs were brought from offline to online, so we are not sure if we can trust this measurement fully: Reflection off M5 (HR) G 1338761550 -0.064 Misaligned PRM, same location G 1338761607 0.002WFS alignment problems --------------------- Analysis is ongoing, but the above numbers seem to make sense mostly according to in-our-head calculations. We had a problem with changing alignment during lunch. The REFL WFS A DC light fell dramatically, by eye we could see that the beam was too low. REFL WFS B DC also fell slightly. We did not notice this before taking the measurement at location G. We only noticed when we could not close the DC centering loops around the RMs. At lunchtime, two things happened: we tried to lock PRMI using the POP diode (which is currently in air since HAM1 is vented), and while doing this changed the PRM alignment slightly. We also took the HAM2 and HAM3 ISIs from offline to online, which can shift the beam. We checked by eye the location of the REFLAIR beam on the West wall of the LVEA, it seemed very close both before and after lunch. Georgia and Keita did some forensics, and moved the PRM to a place where we could recover the power on the REFL WFS and close the DC centering loops around the RMs. At this point we had around 2500 counts on each PD, whereas before we had around 4000 counts. Then, Keita tried slowly moving the PRM back to where it had been. This brought the counts back up, indicating that our REFL beam was clipped somewhere during measurement G. At some point on this journey, we railed RM2. This is when we decided to leave for the day. To Do ----- To do for the power budget: 1) Figure out RM saturation issue (was this always a problem)? 2) Analyze this data, including normalizing for laser drift and calibrating into power. 3) Lock PRMI with carrier 4) Finish measuring REFL path with integrating sphere. 5) With PRMI locked, measure POP path (will have to switch to REFL diode for control)
Non-image files attached to this report
Comments related to this report
Looking at the return loss in reflection we get 3.93/4.42 ~ 89% efficiency—this includes the mode cleaner throughput, 2 passes through the Faraday and the PRM reflectivity. This is higher than some other estimates mentioned in alog 63004.
Secondly, the splitting ratio on HAM1 inlcudes a 90:10 and 2 50:50 splitters for an expetced 2.5% efficiency towards the in-vacuum photodectors. The measurement itself indicates a ratio of 0.064/3.93 ~ 1.6%, or about 35% lower than the nominal splitting ratio. This explains our unexplained losses in reflection. (The above alog mentions clipping during the last measurement, but G and E are compatible, and any clipping would have to be downstream.)
FYI -- The SEI Managers for HAM2 and HAM3 had been off since the team went out to lock HEPI HAM1 on Monday afternoon (LHO:63479 and LHO:63459), so any attempts to lock the IMC and/or PRMI prior to Wednesday Jun 08 2022 at 13:26 PDT (20:26 UTC) would have suffered from the misalignment of the HAM2 and HAM3 ISI tables, because those respective HEPI platforms were offline. (the SEI managers were in State number 35 which is the "READY" state, in which both the ISI and HPI control systems are OFF). See attached trends.
Images attached to this comment
I partially analyzed the data from above. I went back using ndscope and curated some times of measurement data on H1:LSC-EXTRA_AI_1_OUT16 which were valid. I then grabbed those data spans using some code in the power_budget gitlab: I used IMC-PWR_IN as a normalization monitor for slow laser drifts, accounted for an offset of 0.00205 mV for the integrating sphere because this was the offset around the time of the low-power measurements where the offset will be more impactful, and calibrated the integrating sphere volts into watts using our 1064 responsivity in the main alog above.Measurement Label GPSstart time Volts from Integrating Sphere Power in Integrating Sphere [W] -------------------------------------------------------------------------------------------------------------------------------------- PSL to IMC input A 1338750048 -4.42 0.9279 Closed light pipe, same location A 1338750178 0.002 Out of REFL viewport, before M14 B 1338751500 -3.92 0.8254 Closed light pipe, same location B 1338751605 0.002 Out of REFL viewport, before M14 again B 1338751753 -3.94 0.8257 Reflection off M14 (90:10) C 1338751905 -3.62 0.7633 Misaligned PRM, same location C 1338752032 0.002 Reflection off M17 (50:50) D 1338752190 -0.144 0.0308 Reflection off M1 (50:50) E 1338752915 -0.0642 0.0139 Misaligned PRM, same location E 1338753052 0.002 Reflection off M7 (HR) F 1338753209 -0.0727 0.0157 Reflection off M5 (HR) G 1338761550 -0.064 0.0139 Misaligned PRM, same location G 1338761607 0.002Analysis --------Below are some splitting ratio and other interesting calculations. None of Daniel's conclusions above are changed.REFL efficiency = REFL beam / PSL beam = 0.8895 M14 refl = 0.9244 M14 trans = 1 - M14 refl = 0.0756 M17 refl = M17 refl power / (M14 trans * REFL beam) = 0.4928 M17 trans = 1 - M17 refl = 0.5072 Measured power incident on M1 = M1 refl power + M1 trans power from refl off M7 = 0.0296 W Estimated power incident on M1 = REFL Beam * M14 trans * M17 trans = 0.0317 W Loss estimate on M14, M17, M1, and M7 = 0.0021 M1 refl = M1 refl power / Measured power incident on M1 = 0.4705 M1 trans = M1 trans power from refl off M7 / Measured power incident on M1 = 0.5295 Measured power incident on M1 = M1 refl power + M1 trans power from refl off M7 Loss between M1 and M5 = 1 - M5 refl / M1 refl = 0.0018 PSL to M5 efficiency = M5 refl / PSL beam = 0.0150 REFL to M5 efficiency = M5 refl / REFL beam = 0.0168Input Optics ------------First we recall that PRM trans = 3.1%. Second, the IM4 transmission is 2427 ppm, thank you to GariLynn and Camille for tracking down this number which was measured in lab for a bunch of AOIs (see Table 2 here). Third, the IFI we think is around 0.977 forward throughput from The Advanced LIGO input optics paper paper. We are not sure what the backward throughput of the IFI is, so we assume it is the same. If we group all of our unknown IO losses into the IMC transmission, we can estimate it:IMC trans = REFL efficiency / (PRM refl * IFI trans**2 * IM4 refl**2) IMC trans = 0.9664This is much higher than we originally thought, because our PSL beam is lower than we thought it was.IM4 Trans ---------The raw recorded value on IM4_TRANS_NSUM was around 0.810 W. Corrected using Keita's HAM6 measurement yieldsMeasured power on PRM via IM4 TRANS NSUM = 0.897 WDoes IM4 trans make sense with our estimated power on the PRM from IM4 trans?Estimated power on PRM = PSL beam * IMC trans * IFI trans * IM4 refl Estimated power on PRM = 0.8740 WThis is pretty good: within around 2.5%Conclusions -----------This is another point in favor of Keita's input power calibration, in addition to e.g. alog 62936 and alog 63154. What's more, this measurement shows that current IM4 calibration is impossible, because it reports 0.81 W but we measure 0.8254 W on the REFL beam. We need to update our H1:IMC-PWR_IN calibration to be consistent with our number above, which is 7.3% lower. At the same time we should update our IM4 TRANS NSUM calibration, as well as the REFL and POP PDs since we will be lowering the powers on these paths during this vent. The true problem remains: why do the PRG and arm power measurements not hang together with this input power?Repos -----The code for grabbing the data is here: https://git.ligo.org/aligo_commissioning/power_budget/-/blob/master/code/psl_and_refl_path_ham1_measurements.py
Attaching some pics from Monday's HAM1 activities, including integrating sphere contraption and install.
Images attached to this comment
Late entry - I picked up the 3" witness sample which was laying in the center of the chamber, first thing Wed moning before work started. Particle counts just out side the door on 2 monitors were in the 10s or less.