BCAM / Image Quality Correlations¶
Note: Similar plots (using a slightly different data set) for airmass < 1.1 cuts are available here: BCAM IQ plots with airmass cut
The image quality data from December through April has a distribution of r50 (the PSF flux radius) which looks like this:
Although the BCAMs measure mechanical---not optical---collimation, their low-noise readings should be correlated to image quality as long as there is some degree of coupling between the mechanical and optical systems. Here we show profile histograms of BCAM degrees-of-freedom versus PSF flux radius. The error bars are standard errors of the mean. There are clearly discernible correlations between BCAM data and IQ data.
Here are the same plots with the axes swapped:
In fact, since the hexapod is essentially driven by the Donuts, the measurements shown here necessarily include noise from the Donut system. If the BCAMs were also incorporated into the AOS, the correlation should be stronger and the noise lower. Since the BCAMs measure relative---not absolute---misalignment, all we need to do to adapt the BCAMs for use in the AOS is to subtract off the "best positions" as read off from these plots.
Here is a 2D plot of the BCAM (x, y) vs. PSF. The "best-fit" BCAM values (read off from the BCAM vs. PSF plots above) are marked by black lines. These lines are very near to the global minimum in the PSF terrain.
We can also compare the stability of BCAM and Donut data at a given pointing to see how constant the misalignments remain over time. Here we show BCAM and Donut data in the SPT East field near (HA, Dec) = (0, -45). The BCAM measurements here exhibit a lower spread than the Donut measurements.
Pointing and PSF¶
To ensure that the BCAM -- IQ relation isn't spuriously caused by exceptional seeing only at certain pointings, we created the following plot to show the range of (HA, DEC) positions in each PSF bin. In each PSF bin, the blue points are HA positions and the red points are DEC positions. The points have a low opacity to make denser region more obvious, but have no physical meaning.
Here is a scatter plot of the same data:
There are still unanswered questions about the discrepancies between BCAM measurements made from different BCAM pairs. The discrepancy is larger in the x (RA axis) direction. We have ruled out flexure in the BCAM mounts as a significant contribution. Presumably something else in the telescope or DECam system flexes or moves slightly. This possibility is supported by the observed larger errors/jumps seen in the Donut system along the x-axis as well.
Meanwhile, we continue to produce the BCAMs' global average, which is essentially what we want anyway, as ultimately any alignment system must handle asymmetries by producing the best, overall correction. This isn't an insensible thing to do, either, since the global average is a good measurement of mechanical collimation, as evidenced by: 1.) the strong correlation with image quality, 2.) measurements from the CTIO laser alignment system, and 3.) excellent, low-noise measurements taken during the SV hexapod tests.
Here are new vector plots of the January BCAM LUT. This is the same exact data as presented before, but with a more intuitive coordinate system for the vectors. Previously, the +x axis was +HA, which is West; however, a vector arrow pointing along +x indicated a BCAM misalignment measurement along the +x BCAM axis, which is East. Similarly, a rotation around the BCAM +x axis was previously plotted pointing along the +x axis of the plot; however, this rotation will "look" like DECam is tipped toward the North (+y). These new plots adjust the coordinate system of the displacements / rotations to match the compass directions of plot's axes.
The interpretation of this is simple: DECam sags away from, but the bottom tips toward, zenith. This is an illustration of the BCAMs measuring a simple and intuitive mechanical phenomenon.