I was wondering what the IMC VCO calibration was for a CARM calibration. I had been using what's in IMC-F FM3, named "VCO", which is zpk([40],[1.6],536604,"n"), meaning the VCO DC gain is 268302 Hz/V times 2 for the AOM double pass. This calibration comes from Kiwamu's alog 5556 all the way back in dinosaur times (2013). The IMC VCO has been replaced a couple times since. In 2019 Daniel remeasured the VCO calibration (alog 47980), and got a different calibration of Hz/V. This year, again we replaced the IMC VCO, now with less mixing to avoid RF whistles. Jenne has run a few sweeps of this guy as well, so I took the one from alog 64140 and reanalyzed like Daniel did in 47980.Results: -------- VCO Fit = 79.2 MHz + -287.88 kHz + (114.46 kHz/V) v + (-0.39kHz/V^2) v^2 + 0.04(kHz/V^3) v^3 Operating Tune Offset = -3.39 V Slope at the operating tune offset = 118.5 kHz/VI'm going to update the IMC-F FM3 "VCO" with twice this last number (237000 kHz/V).Code ---- To run a fresh IMC VCO sweep: /ligo/gitcommon/labutils/imc_vco_sweep/imc_vco_sweep.py To analyze the data: /ligo/gitcommon/labutils/imc_vco_sweep/imc_vco_sweep_calibration.pyRelated alogs: -------------- Kiwamu's apocryphal IMC VCO calibration: alog 5556 Daniel recalibrates IMC VCO (but did not update IMC-F): alog 47980 Jenne's IMC VCO sweep analyzed here: alog 64140 Giacomo tracking the IMC-F cal: alog 5945 Daniel IMC-F frequency noise: alog 50099 O3 (incorrectly calibrated) frequency to DARM TF: alog 45352
Nancy, Evan
We applied this recalibration to O3b data and found that the Hanford IMC VCO spectrum is essentially at the refcav thermal noise limit in several places between 30 Hz to 2 kHz, as shown in the first attachment (red), with intrusions from PSL acoustic motion throughout (faded dots show the coherence projection from the PSL periscope X accelerometer). Below 30 Hz there is significant limitation from ADC noise and above 2 kHz there is something else, perhaps residual laser frequency noise. The thermal noise estimate here comes from Chalermsongsak et al. (2014).
Livingston's IMC VCO spectrum appears much higher for reasons that are unclear. The calibration at dc is consistent with the frequency readback channel and I cannot identify a whitening/compensation discrepancy. Livingston also is limited by PSL acoustic motion in many places. The amount of periscope motion is similar between Hanford and Livingston (second attachment), meaning that if we take these plots at face value they imply Livingston's frequency feedback system is 10–20× more sensitive to PSL motion than Hanford's.
It is also worth mentioning that this noise level at Hanford is not consistently this low among the different observing runs, with O1 and O3a being poorer than O2 and O3b. I assume this has something to do with varying loop settings but I have not investigated systematically.
From alog 66293: "One thing to note is that the tuning offset on the VCO medm screen is a factor of 2 lower than the analog tuning input at the rack (or as measured by the CTRL output of the common mode board)." So the IMC VCO calibration above is in units of the medm screen volts, which is a factor of two lower than the actual volts we care about in loop. We should use the calibration in alog 66293 going forward.