Krishna

I've attached some plots showing the tilt measurements with Compact-BRS. The calibration is only approximate right now (+/-50%). More accurate calibration will happen soon.

Quick Tutorial:

Recall that the instrument consists of a beam-balance with two fiber interferometer readouts at left and right ends of the balance. The fiber-tips are located on a translation stage driven by a PZT stack each (100V, 11 micron range). Photodiodes (PDs) measure the output of the interferometers. The PZTs are used in a low-frequency feedback loop to keep the PDs at mid-fringe. The low-frequency tilt signal is therefore in the PZTs, while the high frequency seismic tilt is in the PDs. The raw displacement signal is converted to angle by dividing by the arm-length (roughly 15 cm from the center). The beam-balance is also damped with capacitive actuators using the PZT control signals.

The fiber interferometers are operated with a large gap (using a collimating lens) and thus are individually limited by the frequency noise of the common laser. However, taking the difference of the PD signals allows us to subtract frequency noise to the extent that the cavity lengths are matched. Crudely,

PD 1 = Angle + L1 * Frequency noise

PD 2 = -Angle + L2 * Frequency noise

Therefore,

SUM = (PD1+PD2)/2 = Frequency noise*(L1+L2)/2

and DIFF = (PD1-PD2)/2 = Angle + Frequency noise*(L1-L2)/2

The attached plots show data measured this afternoon. The first plot shows the PZT and PD signals for each cavity. After the flexure replacement on Monday, the instrument is slowly settling but is continuing to drift in one direction as seen from the plot.

The second plot shows the ASD of these four signals. As explained above, the tilt at higher frequencies is in the PDs and the low-frequency tilt is in the PZTs.

The third plot shows the ASD of the SUM and DIFF channels. Note the smooth 1/rt(f)-ish slope in the SUM channel (at low frequencies) indicating that it is limited by the frequency noise, as expected. Unfortunately, there appears to be excess noise above 10 Hz, which look like acoustic pickup of some sort. This was taken during a noisy LVEA, so it is possible that after the clean room fans are turned off and it gets quieter, the noise will go down. The DIFF, does dip below the SUM as expected - indicating that we are getting some frequency noise suppression. It is not clear what the noise floor of the DIFF channel is yet. IF there is a factor of ~3 frequency noise suppression (at a minimum), then the noise floor ought to be a factor of 3 below the SUM channel (~ 30 picorad/rt(Hz) at 10 Hz). But if we are limited by acoustic noise, then the noise floor could be worse.

The fourth plot shows coherences between some interesting channels.

Compact BRS is performing reasonably well at the moment. Stay tuned for more data/plots when things are quieter. I'll also try to compare c-BRS with other local sensors.