Dan Gould (ANU), Georgia, Sheila, Daniel, Vicky
Following up observations of scattering transients during PSAMS actuations with intentional scatter tests, using longitudinal excitations on various ZMs in the SQZ path.
- 1341597980 - modulating backscatter with ZM5 L test excitation of 0.3Hz, 5000 counts gain, OSEM longitudional RMS ~ 3 um.
- did many tests. found quickly that OPO locked/unlocked doens't matter to see this scatter. blocking green pump only turns off mean SQZ injection.
- proceeding now to test L-excitation on several ZMs, to figure out where scatter originates, with OPO unlocked
- 1341599400 - same exc in ZM4 L, looks same as ZM5 L. OSEM RMS L ~ 3um. ZM4 show same scatter shelf as ZM5
- 1341599645 - same exc in ZM2 L, btwn OPO/FC1. No scatter shelf from ZM2.
- 1341600029 - back to ZM5 L excitation. same OSEM L RMS = 3um exc.
- Is scatter from SFI2 un-optimized? Can we optimize SFI2 to suppress this scatter?
- 1341600390 - 1341601300 scanned some SFI2 TEC temperatures, from 25.2 - 30.2 degC. No noticeable change, but also based on LLO:57943, we realized this is too small below the noise to see anyways.
- Tried SFI1 TEC across same range, as expected nothing obvious.
- Is scatter from SFI2 un-optimized? Can we optimize SFI2 to suppress this scatter?
- 1341601528 - shaking ZM6 L - basically same response as ZM4/5 excs.
- 1341602829 - stopped excitations
- 1341602895 - closed beam diverter
Sheila's screenshot attached below: When we open the beam diverter, we see 14 uW on the OFI monitor PD, which is 1% of what reaches SFI2. LLO:57943 AS_C reports that there is 140 mW from the IFO to HAM6, so we are getting 1% of the light headed towards HAM6 scattered to HAM7 at the current OFI temp. The current OFI set temp is 25 C.
From the frequency cut-off and amplitude of the scatter shelf, we can calculate the amount of scattered light from this 5000 cnt, 0.3 Hz ZM5-L excitation and go from there. Attached screenshot shows DARM with excitations on various ZMs. Seems like, shaking btwn OPO-FC1 doesn't scatter, but shaking btwn OPO-IFO path shows scatter.
(Louis, with help from Shiela & Keita)
I modeled the scatter and estimate the reflectivity at ~900 ppm (asd) [which translates to ~0.81 ppm in power]. The data from OMC DCPD_SUM_OUT has been calibrated to RIN and closed-loop DARM corrected using pydarm_modelparams_PostO3_H1_20220614.ini. The ZM4 was shaken with OSEM RMS L ~ 3um, but I estimate something closer to 4.9 um rms. I'm not sure where the discrepancy is coming from.
The red and green curves in the attached plots are the DARM-corrected OMC DCPD SUM_OUT channel during nominal operation and during the ZM4 injection, respectively, in RIN/rtHz. The blue curve is the modeled scatter at ZM4. The scattering model and nominal DCPD SUM_OUT channel are summed and shown in orange.
Code and data used for this is in /ligo/louis.dartez/Documents/archive/SQZ_scattering_20220824
Wouldn't a change of 3um in the ZM4 position make a ~6um length change in the light path? And 12um if it is doubled passed?
At this time ZMs were in positions ZM4(-218, 309), ZM5(-396, 426), ZM6(880, 816)
@Daniel
Yes, a 3um change at ZM4 corresponds to a total length change of ~6um in the light path (per reflection). This is accounted for in my script (now moved to gitlab).
While chasing down possible sources for the 4.9<--->3 um rms discrepancy, it was suggested that maybe we were actually looking at the second shelf (i.e. the double bounce / double reflection shelf).
The attached plot shows the scattering model adjusted so that the "double bounce" shelf matches DCPD_OUT during the ZM4 injection, placing the single-bounce shelf cutoff at ~25Hz.
For the ZM4 injection to have resulted in two shelves, we would expect to see /something/ near 25Hz in the green trace to convince us of the presence of a scattering shelf. This does not seem to be the case and I think we can safely assume that the 50Hz shelf present in the initial injection corresponds to the single-bounce scenario.