craig.cahillane@LIGO.ORG - posted 11:35, Friday 16 September 2022 - last comment - 11:28, Thursday 22 September 2022(64974)
Contrast defect estimate during DARM offset test
On September 1 2022 I moved the DARM offset to check the optical gain change in OMC DCPDs and AS_C (alog 64818). During that test, Evan found that the interferometer DARM offset carrier light is changing alignment into the OMC, by looking at the demodulated 17.1 Hz PCAL line seen in ASC OMC A/B. This suggests that when we reduce the DARM offset, part of the drop in OMC DCPD power is due to misalignment. I plotted the DARM offset vs the OMC DCPDs and AS_{A,B,C} during this test (Plots 1 and 2). I also plotted the 410.3 PCAL line optical gain vs DCPD power, similar to alog 30573. Some things to note: 1) EDITTED: The DARM offset vs OMC DPCD SUM light is a sort of circular plot. When we "change the DARM offset" by moving H1:OMC-READOUT_X0_OFFSET, this is setting the light we want incident on the OMC DCPDs according to the settings on our IFO DC READOUT screen. So this plot is telling us that our DARM offset math is working. It does imply that about half of our power incident on the DCPDs is actually contrast defect light when at the lowest DARM offset setting here 2) The AS_C and ASC-OMC_{A,B} seem to disagree on how much power is entering HAM6, which should not be true since OMC-{A,B} are calibrated to AS_C according to Keita's integrating sphere measurement. Likely due to full IFO alignment change vs single bounce alignment during calibration. 3) The optical gain vs DCPD power (plot 3) seems to give a contrast defect light estimate that agrees with the April OMC scans (2.6 mW). 4) From first principles, the antisymmetric power P_as should follow:P_as = P_in * PRG * Signal Recycling Gain * Arm Reflection Derivative * k^2 * ΔL_dc^2 + P_junkFor us at the time of this measurement, we havePin = 48 W PRG = 52.8 SRG = 0.1 # not really known, but fairly robust to losses since it's antiresonant Arm Reflection Derivative = 262 k = 2 pi / λ λ = 1064e-9 mThese number imply that P_as should scale with ΔL_dc^2 like 0.6 mW/pm^2.Past contrast defect alogs: ---------------------------OMC scan from April 2022: alog 62694 Optical gain estimate O3: "alog 48365 Sheila estimates contrast defect via 45 MHz test: alog 56590 Valera and Anamaria's original idea LLO alog 42731 I measured contrast defect by sweeping DARM offset through zero alog 46142 Evan Hall measured contrast defect via DARM optical gain Oct 2016 alog 30573 Stefan makes the IFO DC readout math alog 18470EDIT: -----After a convo with Sheila, I found the alog 18470 where Stefan created this method of setting the DARM offset using the amount of light on the DCPDs. We don't really "set the DARM offset", we just set the amount of light on the DCPDs. This means our "zero DARM offset" is likely not really zero, and we likely have significant contrast defect light on the DCPDs with low levels of power.EDIT 2: -----Plot 4 shows the uncalibrated "DARM offset" counts vs DARM optical gain linear fit, which estimates that the true DARM offset zero happens at 1.53 cts.EDIT 3: -----Plot 5 is a short derivation of optical power vs optical gain. For optical power at the AS port, I removed a factor of 4 from Kiwamu's part 2 Section 3.1 equations due to our different DARM definitions. I also divided the optical gain by 2 because the electric field carries the modulation info, not the power.
Non-image files attached to this report
Comments related to this report
From the above measurements and numbers, we can report the contrast defect and set an upper limit on the homodyne angle.TEM00 contrast defect --------------------- tem_00_contrast_defect = P_src / P_bs = Power in SRC / Power incident on BS = (P_as / T_srm) / (Pin * PRG) = (2.6e-3 W / 0.3234) / (48 W * 52.8) tem_00_contrast_defect = 3.2 ppm Homodyne angle -------------- (upper limit assuming contrast defect light and DARM offset light are 90 degrees out of phase) contrast_defect = 2.6 mW total_dcpd_light = 23 mW darm_offset_power = total_dcpd_light - contrast_defect homodyne_angle = arctan2(sqrt(contrast_defect), sqrt(darm_offset_power)) homodyne_angle = 20 degEDIT: -----Anamaria advises that contrast defect refers to all junk carrier light in the AS port, including carrier HOMs. I revised the above to describe the "TEM00 contrast defect", similar to LLO alog 56336
I spent some more time trying to look at the ADF math and rather than fitting inferred squeezing metrics (a calculation which made many assumptions that might not have been completely valid), I tried to match the ADF signals themselves to a model ADF signal.
There is some degeneracy between the inferred homodyne angle and the arm power. I'm attaching two plots to this comment
The first plot (ADF_LO12) uses a model that assumes a 12 degree homodyne readout angle and 260kW of arm power.
The second plot (ADF_LO20) uses a model that assumes a 20 degree homodyne readout angle and 280kW of arm power.
I have reason to believe that the 20 homodyne angle is more accurate. Fitting the data to a model with more arm power generally requires a larger homodyne angle to make sense, along with the assumption that my measurements were further deviating from the true squeezing and anti-squeezing angles.
Non-image files attached to this comment
In the main alog, final PDF, I made a mistake calculating the optical gain. It should readAnalytic Optical gain --------------------- Pas/ΔL = 2 Gp Gs R_a' k^2 ΔL_DCand the relationship between AS power and optical gain should be the same as in LLO alog 56336:Pas = 1/4 a (Pas/ΔL)^2 + P_00To double check these calculations, I compared the simple pole DARM plant to a lossless, TEM00-only Finesse simulation. I also compared to my "lossy BnC model" used in alog 64928 to make sure the factors of 2 and √2 are correct. They all agree to around 1%, see the PDF. The comparison code is attached, and lives in .
Non-image files attached to this comment
