I tried to calibrate our ASD spectrum into units Hz/rtHz.
To do this, I retook our demodulated beatnote spectrum in Vrms/rtHz from ~12Hz to ~12kHz , retook an open loop gain measurement of the PLL, and tested the actuation of the Marconi in kHz/Vpeak.
I'm pretty sure I did it wrong since our ASD reports a noise level of 1e-3 Hz/rtHz at 100 Hz.
The way I calibrated was by the following:
Hz/rtHz ASD = Vrms/rtHz ASD * PLL CLG * Hz/Vrms Actuation
where Vrms/rtHz is the original spectrum (plot 3 below), PLL CLG is the closed loop gain of the PLL equal to 1/(1+ OLG), and Hz/Vrms is read straight off the Marconi in 400 kHz/Vpeak and multiplied by 1e3/rt(2) to get 2.8e5 Hz/Vrms.
The PLL OLG is plot 2 below. The Agilent is unable to measure lower than 10 Hz. I feel like the solution is simple here... just make an analytic single pole and delay TF and extend it down to DC. In reality this control system is a bit weird since we are actuating on frequency but detecting phase, so we get a 1/f fall off which extends I don't know how far down to DC. Don't trust anything below 10 Hz.
Next step is to diagnose our North and South cavity FSS. Both are susceptible to "ringing", or high amplitude modulations at 10 to 20 kHz. "Ringing" is mitigated by turning off and on the Boost and dumping the integrator on the FSS boxes, but the ringing always comes back after a while. We suspect ringing is a result of the PZT railing from poor control loop design. Turning down the Common and Fast gains reduces the occurances of ringing, but eventually we want to turn up the gains as much as possible to get a high bandwidth and a lot of laser frequency noise suppression at low frequency. One theory Andrew has is one of our power supplies may not actually be supplying any current (see elog 1850). |