Next: Discussion Up: No Title Previous: The Full-Frame biases

# The transfer and linearity curves

All images described in this section start at pixel (700,1050).

The image sequence collected for the purpose of generating transfer curves and linearity curves did not cover the entire area of the CCD and therefore 5 smaller bias images were obtained, covering the same subraster as the flat-field images.

```     bias_1.fits                Null        1.000   360   320
bias_2.fits                Null        1.000   360   320
bias_3.fits                Null        1.000   360   320
bias_4.fits                Null        1.000   360   320
bias_5.fits                Null        1.000   360   320
```

These images also show the same interference structure as the full-frame biases and a more pronounced glitch at the start of the overscan.

Table 3 lists the mean counts and standard deviation of each of these bias images.

Since the interference pattern is not coherent between images, difference images are not considered.

18 pairs of flat-field images at various integrations times were taken in the sequence shown below.

```   tc_0.1_1.fits                Null        0.100   360   320
tc_0.1_2.fits                Null        0.100   360   320
tc_0.2_1.fits                Null        0.200   360   320
tc_0.2_2.fits                Null        0.200   360   320
tc_0.3_1.fits                Null        0.300   360   320
tc_0.3_2.fits                Null        0.300   360   320
tc_0.5_1.fits                Null        0.500   360   320
tc_0.5_2.fits                Null        0.500   360   320
tc_0.6_1.fits                Null        0.600   360   320
tc_0.6_2.fits                Null        0.600   360   320
tc_0.8_1.fits                Null        0.800   360   320
tc_0.8_2.fits                Null        0.800   360   320
tc_1.1_1.fits                Null        1.100   360   320
tc_1.1_2.fits                Null        1.100   360   320
tc_1.5_1.fits                Null        1.500   360   320
tc_1.5_2.fits                Null        1.500   360   320
tc_2.0_1.fits                Null        2.000   360   320
tc_2.0_2.fits                Null        2.000   360   320
tc_2.8_1.fits                Null        2.800   360   320
tc_2.8_2.fits                Null        2.800   360   320
tc_3.7_1.fits                Null        3.700   360   320
tc_3.7_2.fits                Null        3.700   360   320
tc_5.1_1.fits                Null        5.100   360   320
tc_5.1_2.fits                Null        5.100   360   320
tc_6.9_1.fits                Null        6.900   360   320
tc_6.9_2.fits                Null        6.900   360   320
tc_9.3_1.fits                Null        9.300   360   320
tc_9.3_2.fits                Null        9.300   360   320
tc_12.5_1.fits                Null       12.500   360   320
tc_12.5_2.fits                Null       12.500   360   320
tc_16.9_1.fits                Null       16.900   360   320
tc_16.9_2.fits                Null       16.900   360   320
tc_22.9_1.fits                Null       22.900   360   320
tc_22.9_2.fits                Null       22.900   360   320
tc_31.0_1.fits                Null       31.000   360   320
tc_31.0_2.fits                Null       31.000   360   320
```

For each of the pairs of frames, the integration time, mean counts and variance of the difference of the two frames (divided by two) were computed for the subregion (20:300,20:300) of the CCD image. The median stacked bias image was subtracted from each image before calculation. The results are listed in Table 4.

Figure 6 is the linearity curve and a straight line fit for the first of each pair of exposures. Extrapolating the fit to zero counts gives a negligible shutter delay, as expected, since the LED is controlled directly by the shutter controller and no mechanical shutter is involved. Saturation is not evident in this plot.

The transfer curve (using only the first of each pair for the X-axis, the second gives similar results) is shown in Figure 7 together with a straight line fit. Saturation is seen in the last point of the plot, which is therefore not used in the calculation of the conversion factor. The conversion factor is 2.31 electrons/ADU which yields a readout noise of 5.3 electrons/pixel.

Figure 8 shows the linearity curves expressed in terms of count rate versus exposure levels. Any nonlinearities present are at an amplitude of

Next: Discussion Up: No Title Previous: The Full-Frame biases

Tim Abbott, tmca@cfht.hawaii.edu
Wed Jun 28 16:33:36 HST 1995