Moving MOSS to main dome- Sept 2024

Sept 11, 2024, Stubbs.
This document lays out a plan for moving the Multibeam Optical Seeing Sensor (MOSS) from the AuxTel dome into the main telescope dome, for dome seeing diagnosis with ComCam. It will eventually be replaced with a MOSS II version that spans the full annulus of the Rubin pupil, but in the near term we’ll settle for the four-beam version with roughly 1 meter between beams.

The objective is to have MOSS serve as one of a number of diagnostic dome seeing instruments. MOSS has the singular advantage of sending light right down the same optical path that is traversed by light from celestial sources. The downside is we would need to interrupt night-time observations to point the telescope at it and take dome-seeing diagnostic data.

The anticipated use duration is the time that ComCam is on the main telescope, a few months at most.

We have the following aspects to consider:

mechanical & electrical, packaging, and alignment
software
power on/off
data analysis pipeline
integration with ComCam operations, streamlined operations.

Mechanical & Electrical
The overriding requirement is that we not place any optics or instruments on the main telescope at risk due to falling pieces. This can be accomplished by mounting the system on the inside of the enclosure, well away from any aspect of the Rubin TMA. We need access to AC power and internet connectivity. The ideal would be a hardwired ethernet connection but we’ll settle for WiFi.

The device comprises two optical benches. Four parallel beams are launched in the plane of one of them, with individual tip-tilt control that allows us to move the relative position of the four spots on the focal plane. Since the enclosure and the TMA are azimuthally independent, we can always orient the TMA in azimuth to point at the right place. We need to orient the launch platform such that the four beams hit the M1 mirror. We can then adjust the elevation angle and azimuth of the TMA until its optical axis is parallel to the beams. So the tolerance of the tilt of the device is quite large, likely in excess of 10 degrees. (Assume the MOSS device is 10m from the primary. Changing the elevation of the TMA displaces the beam on the primary by only a small amount.) An action item is to identify an appropriate mounting location on the interior of the enclosure, and initiate fabrication of any mechanical mount elements we need.

The ideal location would be to mount the device on the inside of the Rubin enclosure, about half-way up so that telescope would point at 45 degree elevation to ‘see’ it. Placing it close to the edge of the slit means that the azimuthal misalignment compared to normal observations is minimized. The location of the platform that is intended to support the CBP would be a good place. Placing it on a manually-operated limited-range alt-az mount will prove useful.

putting MOSS onto an alt-az mount would give complete freedom of operation and would greatly ease alignment. We could also place the beams anywhere on M1, like with CBP.

this would work for remote-controlled mount, we bought one already.

Both to constrain any potentially falling parts and to provide dust protection we should make a sheet metal housing to cover the optical components.

Software

Building a CSC to control the operation of the system will integrate it with the Rubin architecture and should allow for block-scheduled operation. For this version I propose we use remote-desktop access to the windows machine to tweak the tip-tilt orientations of the four emerging beams. That operation will likely be done only a few times, to place the beams a desired separation apart.

The CSC under development is documented (HERE).

Operation

There are multiple ways we can run data collection for this device:
a) Operate ComCam in standard full-frame readout mode, and send pulses into the camera at roughly the same rate that images are being taken. Ideally we’d use shutter state to determine when it opens, pause a appropriate time, send one flash, and then close the shutter and read out the camera. If we don’t have access to shutter state then we’d just run MOSS open loop at a pulse rate slightly slower than the camera data collection rate. This ensures only one pulse is delivered per image.

b) run in guider mode after positioning the beams onto different ROIs, which could be on the same CCD or spread across them. This gathers data faster but is much more demanding regarding alignment and stability.

c) Run in strip chart mode with either continuously-on source or else pulsed. We take 2 seconds to read out 4000 rows so the row rate is 0.5 msec per row, or about 1 Mpix/sec per amp. As long as each strobe pulse duration is less than 1 msec we will get unstreaked PSFs. If the source is pulsed at, say, 100 Hz and the spots are placed near the serial register, we’d get roughly 200 realizations of differential image motion in a 2 second image, and temporal Fourier coverage of image motion out to 50 Hz. Alternatively, if we run in strip chart mode with the source continuously-on then we get only 1-d image motion information but temporal Fourier coverage increases to hundreds of Hz. The strobed-strip-chart mode seems the most advantageous since we don’t suffer the overhead of switching to guider mode and we get hundreds of differential motion realizations per collected image.

Separately, we wish to control power on-power off using a PDU, in a way that conforms with Rubin standard practice and expectations.

action items:

identify one or more places to mount MOSS optical benches inside main dome.
implementation review and safety review.
CSC implementation
decide on whether GUI control of stages is adequate
Figure out how to run a PDU, and run electrical power to the mounting location
Ask for AuxTel engineering time to develop data collection and analysis methodology. This could all be done in the daytime!
develop sheet metal enclosure
make parts list and start procurements, drop ship to Tucson for forwarding to Rubin or assemble in Boston?

questions-

what is going into the ComCam filter array? Is there a ‘blank’ ?
where can we get AC power on the inside of the MT dome?
what is the fastest reasonable cadence of ComCam images? 15 sec? 5 sec?
what does an observing sequence look like?
Make sure ComCam will operate in strip chart mode, ie readout with shutter open