Night 5 was half LEECH and half HOSTS commissioning. The LEECH part was clear with 0.7-0.8 arcsec seeing.
The nulling commissioning made significant progress in several areas. We
- Optimized our gain parameters for pathlength control
- Measured throughput of each of our beams
- Tested feed-forward of the accelerometer signals
- Tested "fringe hopping" on sky.
- Tested vibration filtering within the loop.
The loop performed under conditions from 0.6-0.9" with integral gains of 400-700 and proportional gains of 2 (these will need to be rescaled once our corrector values are calibrated). The RMS values recorded with phasecam were 50-70 degrees RMS (300-400 nm at 2.2 um wavelength). With an I gain of 10 we see 200 degrees RMS. With a gain of zero we see ~500 degrees RMS.
The closed, vs. open loop amplitude spectra indicate we are providing correction at frequencies up to 50 Hz. This is consistent with our model of the system, if we assume a lag of approximately 2 ms between sensing and correction.
We confirmed that there is no vignetting on the right or left side with our nominal nulling alignment. The measured intensity ratio between the left and right beams was right/left=80%. This is lower than the February measurement of ~88%. The former would leave a residual of 0.3% and the latter would leave a residual of 0.1% in the null beam. This effect can be calibrated out, so either level is acceptable. We suspect this measurement may have been contaminated by clouds.
Accelerometer Feed Forward
The accelerometers were used to measure a prominent vibration at 11.5 Hz on the secondary mirrors. This signal was fed into the FPC and we successfully demonstrated reduction of the 11.5 Hz peak by roughly a factor of three (needs confirmation with the telemetry data). However, there were other prominent vibration peaks in the 9-15 Hz range, which meant that the total RMS was not significantly improved in our initial test. These vibrations appeared to be coming from the left tertiary (but not the right one). Prominent peaks were seen in both phase and accelerometers at 14 Hz in particular.
The ability for the phases sensor to move be n-lambda steps is important for setting the 2 um to 10 um setpoint. Denis tested a script during the afternoon to carry out this procedure automatically. The script adjusts the 2 um setpoint to one side and then the other of the setpoint which corresponds to the best null at 10 um. It then measures the 10 um flux imbalance for each dither and refines the setpoint.
We could not test this script on sky due to variable clouds, which prevented reliable 10 um flux measurements. However, we did confirm that the ability to hop from one fringe to another and change the 2 um setpoint both worked on sky.
We also tested the capability to measure the 12 Hz vibration from Phasecam and separately add a correction to the phase loop. This test did not show any improvement over the correction already being provided by the PID controller. We expected this may be the case, if the PID were properly tuned.