To look for slit mechanism repeatability, we moved the slit wheel by -12000 steps, then 12000 steps to see if the apparent positions of spectral lines changed after the motion. We only did 10 frames for this test, discarding the first image, in which we brought the slit to 3.6341 V.
by hand turning the mechanism knob.
EXES was looking at the gas cell, targeting 1335 cm^-1 H20 lines.
filter: 7.466um
decker: central arc
V_echelle: 7.9809 V
flip: high medium
lens: OPEN
A sequence of four 3-second images were taken at each iteration. I combined these by doing a linear fit to the pixel values across the frames, and recording the slope into a single 1024^2 array (using [3,6,9,12] seconds as the x-axis for the fit).
10 x 100 pixel box around a H20 spectral line, used to track changes in the slit wheel position after -12000 step,+12000 step motions.
Then for this very preliminary test, I focused on just one line, at [128, 516], and summed a 10x100 pixel box of this line to have a 1-d cut. Even though the line was strong, a ~20 adu amplitude sawtooth pattern (probably from 60 Hz noise in the readout) made fitting a Gaussian problematic. I smoothed the 1-d array with a 4pixel boxcar, and then fit a 6 param Gaussian.
The fit centers of the Gaussian (along the y-axis)
52.975
53.021
52.796
52.839
52.767
52.777
52.841
52.876
52.948
52.808
For a stddev of 0.09 pix and a maximum difference from the mean of 0.16 pix.
We probably want to know a great deal more about out repeatability than this - longer time baseline, effects of changes in cabin temperature ...
And as far as measuring it, I want to try to correct the 60 hz noise (or whatever it really is) without using the smooth function. But at least the mechanism seems very repeatable over short time baselines.
No comments:
Post a Comment