Shannon Dulz

AST 311

September 30, 2015

Detector Lab Report

Abstract

            In this lab, we obtained flat, dark and bias images on the CCD cameras in order to characterize background levels of these CCDs for later processing of images. We calculated the dark current rate and average bias levels.

Introduction

            As part of every astronomical image obtained with a CCD, there are background counts of electrons resulting from the detector itself. To account for these background levels, three types of calibration images are taken: bias, darks and flats. Bias images account for noise introduced from the readout of the detector. Biases are taken as zero-second exposures with a closed shutter. Dark images account for the accumulation of noise due to the electronics of the detector and it tends to grow linearly over the integration time. Dark images are taken at the same exposure length as flat images (or scaled to the same length) and with the shutter closed. Flat images account for pixel-to pixel variations in the detection of light. Flat images are taken of an illuminated screen (or the sky at twilight for sky-flats) with the shutter open at an exposure length results in the image being evenly illuminated but not yet saturated. During processing, the bias frames are averaged into a master bias and subtracted from all the other images including darks and flats. Then, the darks are averaged into a master dark, which is subtracted from the science images and the flats. Lastly the flats are averaged into a master flat, which is divided from the science image pixel-by-pixel.

Procedures

            For this lab, we tested the CCD cameras without a telescope. The cameras were connected to our laptops via USB cables. Then the program Maxim DL was used to collect images. To connect the camera in the program, under the settings tab on the program, we clicked set-up camera and insured the correct camera model is shown. Also insure, swap chips is off, guide chips is internal, and ext trigger is off. Under the options tab, turn off rotation orientation and auto dark subframe extraction. Finally, click connect.

            To begin taking images, under the expose tab, adjust the exposure time to required time. Images must be saved after each exposure unless Auto Save is used. We began by taking 10 bias images of exposure time zero seconds with the cap on the detector to block light. One of the bias images is below.

Next, we obtained flat fields with the camera pointed at an illuminated piece of paper. We took several test images at various exposure times to insure counts would be higher than bias levels but not saturate the detector. We choose an integration time of 0.01 seconds and took 10 flat images, an example of which is shown below.

While this image, appears identical to the bias image, the counts are in fact much higher and a slight grating in counts from top to bottom is noticeable. Using this same integration time of 0.001 seconds we obtained 10 dark images with the cap on the camera, an example of which is shown below.  

Lastly, in order to calculate dark current levels we obtained 1 image each at 1, 5, 10 and 20 second exposure times with the cap on the camera.

Results and Discussion

For this lab, we obtained multiple flats, darks, and bias (shown in the table below).

Image Type	Integration time (seconds)	Number of Images
Bias	0	10
Flat	0.01	10
Darks	0.01	10
Darks	1, 5, 10, 20	1 each

           

            Using IDL, the biases were averaged together by summing the counts in each pixel of each image and dividing that number by the total number of images. This produced a master bias frame the median of which was 2006, giving us the average bias count. The darks of exposure time 0.01 were averaged in the same way leading to an average dark count for those images of 1789. Lastly the flat fields were averaged into a master flat, the median of which was 4299. The average counts of the biases should not normally be higher than that of the darks. This variable bias level could be due to temperature and we will recalculate the bias level from the dark current rate discussed below. While bad pixels were evident in individual images by vastly different count levels in single pixels, these few dozen pixels averaged out when the images were summed together. The flat fields we obtained were very evenly illuminated with only a slight grating of counts from top to bottom of the image. To obtain the dark current rate, we obtained darks at 1, 5, 10 and 20 seconds. We took the median of these images individually to obtain the table below.

Exposure time (seconds)	Average counts
1	1026
5	1031
10	1037
20	1036

We expect to see the dark counts increase linearly over time, the slope of which would give the dark current. While the linear fit is far from perfect due to the few number of images taken, we obtained a dark current of 0.5 counts/sec. The y-axis intersection of this linear fit is the bias level for these sets of images. For this we obtain 1028 counts as the bias level.  This calculation is a much more reasonable number than that obtained from the bias images themselves of 2006. This discrepancy could be due to temperature, especially since the bias images were the first ones taken as the CCD was still warming up. The graph of this linear fit is below.

Conclusions

            In this lab, we took flat, dark and bias images to begin to characterize the CCD detectors. We found an average bias level over 10 images of 2006 counts. When recalculated using the linear fit intersection of the dark current, we obtained a much more reasonable 1028 count bias level. We obtained a very low dark current rate of 0.5 counts/sec. Lastly, we took very even flat field images with average counts of 4299. The low background levels of these detections especially with regard to the low dark current will allow for high signal to noise ratios when they are used for later labs.