J. Kissel, S. Dwyer, K. Venkateswara, J. Warner, T. Shaffer

I stumbled in on a conversation between the four listed above about how to deal with wind, so I immediately started taking notes. Here're some of the conclusions:

- There are *three* fault conditions of the beam rotation sensor (BRS) at EX. Under these conditions the BRS should *not* be used for sensor correction:
   (1) The BRS software is dead. In this case, the raw time series of the ADC input for the sensor has flatlined and does not show the usual ~8 [mHz] sine wave resonance of the beam's suspension.
   (2) The BRS has rung up. If the raw input shows the 8 [mHz] resonance, but the amplitude exceeds ~200 [ct] then the sensor output begins to go non-linear and should not be used. Note that the analog USB DAC that is pushing the analysis software's output into the CDS system is the limit on the signal. This is why, last week, we were seeing what looked like an 8 [mHz] square wave in the time series -- the rotation sensor was so rung up, that the software's requested output signal exceeded the 20 [Vpp] (+/-10 [Vp]) range of the USB DAC. This occurs at +/-16k [ct] in the CDS ADC.
   (3) The BRS's gravitational damper is on. If the gravitational damper is ON, then the BRS has rung up enough that (2) is happening.

- If none of the above error conditions are present, then the BRS can be used to subtract tilt from the GND T240 sitting next to it, and therefore can improve sensor correction. HOWEVER, it won't improve any-and-all sensor correction: the sensor correction filters need to have authority in the frequency region where we expect the GND instrument is dominated by tilt. Therefore, in the current nominal configuration of the BSC-ISIs at LHO -- where we're using Dr. DeRosa's notch-like sensor correction filter that only has authority in a narrow band around 0.5 [Hz] and otherwise using the ISI T240s in feed back down to 45 [mHz] -- improving the GND T240 sensor with the BRS  won't do you any good, because the GND T240 is typically only dominated by tilt only below 90 [mHz].

- Krishna's data from LHO aLOG 16465 demonstrates that even during 5-10 [mph] winds, the following configuration for the ETMX BSC-ISI is equivalent to the nominal configuration:
   (1) Push the blend up to using the 90 [mHz] blends on the X DOF only.
   (2) Use the tilt-corrected GND T240 + BRS super sensor
   (3) Switch to using the Mittleman, low-frequency, "tilt free" sensor correction filters which have authority down to 10 [mHz]
As such, we suggest that this should be the windy configuration for ETMX.

- Jim plans to accelerate his campaign to get the tuned sensor correction MATCH gains pushed into filter banks, such that the nominal value for the *actual* gain of the MATCH bank is 1.0

- T.J. Is going to write us the beginnings of an ISI Configuration manager in order to make it easier to remember all of this.
It will do the following
   To go to windy configuration (in order):
   (1) Switch ETMX ST1 X and ETMY ST1 Y blends from 45 [mHz] to 90 [mHz] (wait to finish)
   (2) Ramp OFF the gain (ramp to zero) of the ETMX Sensor Correction MATCH bank.
   (3) Switch ETMX GND STS X direction input matrix element from using STSB to STSC (i.e. turn ON the BRS correction)
   (4) Switch the ETMX ST1 X Sensor Correction filter from the "SC-rdr" (i.e. the 0.5 [Hz] notch-style DeRosa filter) to the "Mitt_SC" (i.e. the Mittleman, low-frequency filter)
   (5) Wait for the ETMX ST1 X output to settle (which should be the impulse response time of the Mitt_SC filter, ~100 [sec]) -- should be 100 [sec]
   (6) Ramp ON the gain (ramp to one) of the ETMX ST1 X sensor correction MATCH bank.

   To come back from windy configuration (in order):
   (1) Ramp OFF the gain (ramp to zero) of the ETMX ST1 X sensor correction MATCH bank
   (2) Switch the ETMX ST1 X Sensor Correction filter from "Mitt_SC" back to the "SC-rdr"
   (3) Switch ETMX GND STS X direction input matrix element from using STSC to STSB (i.e. turn OFF the BRS correction)
   (4) Wait for the ETMX ST1 X output to settle (the SC-rdr filter's impulse response is smaller and shorter, so maybe ~30 [sec] is fine)
   (5) Ramp ON the gain (ramp to one) of the ETMX ST1 X sensor correction MATCH bank.
   (6) Switch ETMX ST1 X and ETMY ST1 Y blends from 90 [mHz] to 45 [mHz] (wait to finish)
   
This above process assumes that the MATCH TRAMP time has been set to 15 [sec] and is monitored by the SDF system, that Jim has moved the Mitt_SC filter into the MATCH bank (and removed it from the IIR SC bank) and has moved the tuned match gains into filters.

----------
For a description of the Derosa 0.5 [Hz] notch-like sensor correction, see SEI aLOG 645.
For plots and design of the Mittleman low-frequency, tilt-free, (typically used in Z) sensor correction filter, see SEI aLOG 594