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

Star off fiber, not saturated

For each science spectrum, the flux is extracted in each order using a boxcar extraction, and a wavelength solution from a ThAr frame is attached. We then determine the maximum flux between 5560 and 5570 Angstrom. We convert this to electrons per second for a star with m=0 using the exposure time, V magnitude of the star, and gain; we'll loosely call this throughput. This is plotted in the top panel of the following plot (other panels are not relevant here, but are the observed flux in the exposure and predicted S/N based on this):

Clearly, there is some variation in the throughput, which is expected because of variations in transparency and also variations in seeing, leading to different fiber losses at the focal plane. However, the exposures with iodine (+ symbols) are clearly significantly lower than those without iodine (squares) by a factor of roughly 2-3 (we've been using 2.5x the exposure time for iodine, and you can see in the middle panel that this roughly brings observed fluxes to be the same); i.e., iodine throughput is ~40% of the non-iodine throughput.

The horizontal lines represent throughput achieved at the Tenerife SONG node for several different stars (median of many observations), I believe through iodinec cell, from some paper….

To understand the source, we observed a star on the guide camera (off of the fiber hole) with and without iodine cell, and plot the cumulative counts in successive circular apertures. Top plot shows the two cumulative flux curves, and bottom plot the ratio.

The profile with the iodine cell is both less concentrated and also never achieves the same flux as the profile without the iodine cell. Of course, the comparison assumes that the transparency is the same for both exposures. The total flux suggests about a 20% loss of flux. Comparison of the total flux ignores the fact that some light is really absorbed by the iodine, but the response of the detector is broad and probably peaks in the red, so the expectation is that this may not be expected to be a lot. Some total light loss is expected from the uncoated interior surfaces in the iodine cell. Paul Butler estimated 4-6% at each surface, so 10-12% overall, while we may be seeing 20%.

The pixel scale of the guide camera is about 0.16 arcsec/pixel, so the hole to the fiber corresponds to about a 5.5 pixel radius. At this radius, the ratio is about 50%, perhaps a little bit higher. This is close, but not equal, to the 40% measuredabove from the spectra. Of the 50% loss, perhaps a bit less than half comes from the total throughput, while the rest comes from the degradation in image quality.

Paul also mentioned that the ends of the cell are likely to be bowed a bit, but we don't know how much image degradation this might lead to.

The following images shows a focus sweep of a star with the iodine cell in place (at two contrasts). Compared with similar images without the iodine cell, the core is blurrier, but there is also some very asymmetrical light outside of the core of the image.

The following images show coarse focus run with and without iodine cell in place. Aberration is clearly present with the iodine cell.

To look at the overall throughput of the iodine cell in a different way, we looked at flat fields taken with and without the iodine cell. The ratio of these is shown in the following plot.

This suggests an overall throughput of 75% or less, under the assumption that the same amount of calibration light is incident on the fiber; our calibration spot is 200 microns in size for a 50 micron hole, and we have adjusted the iodine cell so that the hole is in almost the same location when the cell is in place, but it is not exact, although we think it is close enough that roughly the same amount of flux should be going down the hole.

Interestingly, the iodine data here was taken after the iodine cell heaters had been turned off for 12 hours, with the cell around 13 C, but iodine absorption is still present!

Fiber spool 241586 (the better of the two) was pulled through the conduit.

Using fiber 2 for the science fiber, as it had consistently highest measured throughput. Absolute and relative throughputs with cal lamps were similar to those measured using the fibers in the lab.

In the focal plane, fiber 2 is the lowest, with fiber 3 above it and fiber 1 at the highest row. At the maximal separation between the different fibers, fibers 3 and 1 overlap in the redder orders. The ferrule was rotated (to about 45 degrees) to avoid this. Without a shutter in the direct calibration feed, the direct calibration fiber was left unconnected.

First light on 251002 obtained spectra! However, images with iodine cell appeared poor and resulting spectra had much lower throughput than without iodine, perhaps due to combination of image quality and guiding on poor images.

Modifications made to change guiding algorithm to ignore saturated pixels to allow for longer exposure times to average out the seeing.

251003 observed several stars. Throughput without iodine doesn't look too bad when compared with Tenerife measurements, a little lower, but not dramatically so. Throughput with iodine still looks poor. In plot below, three dots are from SONG paper with photon rates from these three stars, claimed to be median of many exposures. Repeatability of multiple observations of multiple stars is encouraging from the point of view of guiding.

251006 repeated observations of several stars.