Stellar Observations Network Group (SONG): SONG in New Mexico

The calibration spot moves up and down as the calibration stage is moved. To move the calibration stage left/right one needs to adjust one of the screws on the mirror mount for the mirror in front of the Shelyak. I believe that turning the top screw CW (from behind the mirror) moves the spot to the right.

To position the calibration spot with the iodine cell, the cell can be tilted by adjusting the three screw on the front ring. Need to take care not to overtighten and potentially break the cell, so adjustments should be made by both loosening some screws and tightening others! Moving the cell with the bottom two screws (one in, one out) should move spot left/right, moving the cell with compensating motions of top screw and the bottom two together should move spot up/down.

Best focus for calibration spot determined by looking at image cross-sections at different focus values.

With iodine cell out:

With iodine cell in:

From on-sky tests, focus with iodine is 4625 microns farther back than without iodine. Calibration channel focus seems consistent with this.

Note that the spot brightness is fainter by ~15% with the iodine in, which doesn't seem great…

Given the fibers in the focal plane, it's not possible to illuminate all of the pixels in the detector with light coming through the fibers. One doesn't want to flat field with the calibration light because of all of the pixels with low/zero light levels.

We obtained some pixel flats by removing the fiber input assembly and putting a diffuse light source (an LED lamp with a piece of paper over it) to get illumination across the whole chip:

You still see the spectral signature of the lamp, which has multiple LEDs in it to achieve different color temperatures (it was set to the hottest color temperature). The variation in intensity requires a large number of images to achieve high S/N, but the exposure times were extremely short (0.02s). We stacked 25 of these, then divided by a 25×25 smoothed version to obtain the following pixel flat, which has S/N>200 in most places (less at the bluest wavelengths, especially on the sides of the blaze function):

The plots on the right show horizontal and vertical cross-sections through the middle of the chip. The detector is remarkably flat, with typical variations well under a percent, so even skipping flat fielding wouldn't introduce a large amount of fixed “noise”. Upon higher inspection, there is a low level pattern seen here:

So flat fielding with this flat should help if trying to obtain data at S/N>100.

Sequences of calibration frames:

- UT2510xx : flats and ThAr through fiber 2 (science) every half hour for 12+ hours

- UT251011 : flats and ThAr through both fiber 1 (direct) and fiber 2 (science) every half hour for 12+ hours