Big Johnny and I hacked a function generator output into the cross-connect of the 80 MHz VCO driver so that we could modulate the
amplitude of the light going into the RefCav. The goal of this is to measure the coefficient between cavity power fluctuations and the
apparent length fluctuations. This is to see if the thermo-optic noise in coatings behaves like we expect.
To do this we disconnected the wire #2 (white wire) at the cross-connect for the 9-pin D-sub which powers the VCO driver. This is
called VCOMODLEVEL (on the schematic and the screen). In the box, this modulates the gain in the homemade high power Amp which
sends the actual VCO signal to the AOM.
This signal is filtered inside the box by 2 poles at 34 Hz. I injected a sine wave of 3 Vpp into this input. The mean value was 4.6 V. The
RCTRANSPD = 0.83 Vdc. We measure a a peak there of 1.5 mVrms. To measure the frequency peak we look in
the FSS_FAST signal from the VME interface card. With a 10 mHz linewidth, there's no peak in the data above the background. This signal
is basically a direct measure of the signal going to the NPRO PZT, so the calibration is 1.1 MHz/V.
We expect a coefficient of ~20 Hz/uW (input power fluctuations). We have ~1 mW into the RC, so we might expect a ~20 Hz frequency shift.
That would be a peak-height of 20 uV. In fact, we get an upper limit of 10 uV.
Later, with more averaging, we get an upper limit of 1e-3 V/V which translates to 1e-3 * 1.1 MHz / 1 mW ~ 1 Hz/uW. This is substantially lower
than the numbers in most of the frequency stabilization papers. Perhaps, this cavity has a very low absorption?