Fig. 8

(a) Linear regression of the sample variance \(\sigma _{total}^{2}\) over exposure time of a series of dark frame subtracted background images, according to Eq. 78, without the gain reference applied. The graphs follow the same color scheme of Fig. 1. The read-out noise \(\sigma _{read}^{2}\) is given as the offset and the thermal noise \(\sigma _{therm}^{2}\) as the slope of all quadrants. The uncertainties of the results are derived from the standard error of regression \(\sigma _{SER}\) and a 95\(\%\) confidence interval. The detector was cooled to a temperature \(T_{Detector}\approx -20^{\circ }\) C. (b) By applying the gain reference, the values change according to Eq. 61. (c) The row variance can be found as the mean value of the series and (d) does not change with the application of the gain reference due to rounding effects. Below the graphs, the table shows the noise sample variance of the read-out noise \(\sigma _{read}^{2}\), thermal noise \(\sigma _{therm}^{2}\) and row noise \(\sigma _{row,j}^{2}\) of the respective quadrants. These values are derived from the regression analyzes of the above graphs. For each noise, the table shows the values of the dark frame subtracted images (DF subtr.) and the same images with the gain reference (Gain Ref.) applied. We further calculated the theoretical values of the noises for the images treated with the gain reference (Theory), based on the values found in DF subtr. and the values from Fig. 5, according to Eq. 78. The values are in very good agreement, but we noted a small deviation in in the read-out noise of Q1, which is slightly outside of the measurement uncertainty.