# Fix ccd=33 unmasked defect upstream

XMLWordPrintable

#### Details

• Type: Story
• Status: To Do
• Resolution: Unresolved
• Fix Version/s: None
• Component/s: None
• Labels:
None
• Team:
Data Release Production

#### Description

In DM-19841 we found that the ccd 33 mask was insufficient, but the root cause was that the bad amp had a value of 65k. When we interpolate these bad amps they should really be replaced with something that could be more robustly interpolated over like nans. Look into best way to fix permanently.

#### Activity

Hide
Christopher Waters added a comment -

Looking into this issue prompted DM-20112, which in turn prompted further investigations to confirm that would resolve thinigs.  I think the issue here is:

1. The bad amplifiers have values = 65535.  The camera configuration sets the saturation level to 49000, and suspect to 50000.
2. IsrTask.maskAmplifier() identifies these amplifiers as bad, but does not set the mask to "BAD".
3. These pixels are not "BAD", and therefore are not set to the median value, which itself is skewed because these pixels are  not excluded.
4. Interpolation attempts to smooth the amplifier interface (which is flagged with a defect and marked "BAD"), and is using the unmasked pixels to base one edge of the interpolation, resulting in the bright edge showing up.

Fixing issue 2 and folding in the changes from DM-20112 appear to resolve the issue, blanking out the two bad amplifiers on a test exposure, and providing a reasonable amplifier interface.

Show
Christopher Waters added a comment - Looking into this issue prompted DM-20112 , which in turn prompted further investigations to confirm that would resolve thinigs.  I think the issue here is: The bad amplifiers have values = 65535.  The camera configuration sets the saturation level to 49000, and suspect to 50000. IsrTask.maskAmplifier() identifies these amplifiers as bad, but does not set the mask to "BAD". These pixels are not "BAD", and therefore are not set to the median value, which itself is skewed because these pixels are  not excluded. Interpolation attempts to smooth the amplifier interface (which is flagged with a defect and marked "BAD"), and is using the unmasked pixels to base one edge of the interpolation, resulting in the bright edge showing up. Fixing issue 2 and folding in the changes from DM-20112 appear to resolve the issue, blanking out the two bad amplifiers on a test exposure, and providing a reasonable amplifier interface.

#### People

Assignee:
Reporter:
Watchers: