90prime Bias Images in May/June 2004

C. W. Engelbracht; last updated 2004-7-2

All Bias Images

Each panel of the table below shows the 25 bias (zero) images obtained during the May 2004 90prime run or the 20 images obtained during the June 2004 run, one chip per panel as indicated along the top row. The top ten images in the May panels are from May 9, the middle 5 from May 10, and the bottom 10 from May 11. The top 10 images in the June panels are from June 20 while the bottom 10 are from June 21. The overscan subtraction method is summarized in the second column.

Date	Overscan Subtraction Method	display range	Chip 1	Chip 2	Chip 3	Chip 4
May 2004	none	900 - 1100
May 2004	median of whole region	-30 - 30
May 2004	cubic spline fit along column	-30 - 30
May 2004	median of each row	-30 - 30
June 2004	cubic spline fit along column	-50 - 50

Observed Behavior

• the overall bias level changes from night to night (contrast the first two rows of each panel in row 1 of the table with the last three rows of each panel)
• the bias level often has a gradient in the column direction that changes from night to night that will not be removed with a simple median overscan subtraction (most clearly seen in chips 1 and 2 in row 2 of the table - contrast the first two rows of those panels with the last 3 rows)
• the bias level sometimes displays an abrubt transition along columns (see the first image for chip 2 in the second row of the table - a column plot of the right half of that image is shown in Figure 1)
• the bias level sometimes displays broad (~800 pixels) features along rows at a low (~ 1 ADU) level (contrast the first two rows of the chip 1 panel in the third row of the table with the last three rows - a column plot of the image in the last column and middle row of that panel is shown in Figure 2)
• the overall bias level increased (e.g., by a factor of ~2.5 on chip 1) between May and June
• the difference between amplifiers increased (e.g., from 1 ADU to 7 ADU on chip 1) between May and June

Figure 1

Figure 2

Conclusions

• overscan subtraction is necessary to track changes in the bias level
• a low-order fit to the overscan region is necessary to remove structure in the bias level

The final combined bias (zero) images are shown in Figure 3, with chips 3 and 4 on the top and chips 1 and 2 on the bottom. These images were created using a cubic spline fit to the overscan region and are plotted with the same display range of -50 to 50.

Figure 3
May 2004	June 2004