Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-38095

Investigate LATISS "idle camera artifacts"

    XMLWordPrintable

Details

    • Story
    • Status: To Do
    • Resolution: Unresolved
    • None
    • None
    • None
    • Data Release Production
    • No

    Description

      mfisherlevine reported during the Monday calibration products meeting that odd features were observed when the camera was allowed to sit idle.  The attached images show two science images, 2023021400321 and 2023021400849 that have bright artifacts in amplifiers C13, C01, and C03.  These are exactly the same kind of features I've seen previously during bias construction.  I've attached are bias residual images (bias frame with only overscan and combined bias applied), 2022092700005 and 2022092700006 which show these features blink out when the time between exposures is short.  A short summary table of dateobs values is below showing that the exposure prior to the one with the artifacts is generally more than an hour.

      In browsing the RubinTV archive, I was able to find an example in which a smaller artifact is appearing with only ~five minutes of idle time, with 2023021400421 showing features that disappear in [422|https://roundtable.lsst.codes/rubintv-dev/summit/auxtel/monitor/event/2023-02-14/422]

      For bias construction, I've started skipping the first exposure, and the current LATISS defect map masks these areas to attempt to mitigate their impact.  However, knowing that it's very likely related to the idle time makes it a good example for the SITCom/ISR group.

      youtsumi, any thoughts on what may be happening, and how they may be corrected?

       

      exposureId dateobs feature present?
      2023021400320 2023-02-14T21:52:47.920  
      ​2023021400321 ​2023-02-14T00:20:17.74 ​Yes
      2023021400848 2023-02-15T07:18:20.033  
      2023021400849 2023-02-14T08:06:26.54 Yes
      2023021400420 2023-02-14T02:32:27  
      2023021400421 2023-02-14T02:37:57 Yes
      2022092700004 2022-09-27T17:57:50.702  
      2022092700005 2022-09-27T19:11:06.763 Yes
      2022092700006 2022-09-27T19:11:34.439  

       

      Attachments

        1. 2023021400321.png
          2023021400321.png
          3.45 MB
        2. 2023021400849.png
          2023021400849.png
          3.38 MB
        3. bias-2022092700005.jpg
          bias-2022092700005.jpg
          538 kB
        4. bias-2022092700006.jpg
          bias-2022092700006.jpg
          596 kB

        Activity

          youtsumi Yousuke Utsumi added a comment - - edited

          I think defects are generating charges. These charges can be cleared if one does a readout.

          We can mitigate this by taking additional bias image or ignoring the first image as you mentioned.

          We have implemented IDLE_FLUSH for different purposes but we could use this feature for mitigating this issue too. But it will add another complexity to the operation, especially when we want to power cycle CCDs, because we don't want to power cycle while the IDLE_FLUSH is running. We could add something to stop the IDLE_FLUSH before power cycling to the power cycle script.

          youtsumi Yousuke Utsumi added a comment - - edited I think defects are generating charges. These charges can be cleared if one does a readout. We can mitigate this by taking additional bias image or ignoring the first image as you mentioned. We have implemented IDLE_FLUSH for different purposes but we could use this feature for mitigating this issue too. But it will add another complexity to the operation, especially when we want to power cycle CCDs, because we don't want to power cycle while the IDLE_FLUSH is running. We could add something to stop the IDLE_FLUSH before power cycling to the power cycle script.

          A few points/questions:

          • Do the cameras (and I mean all our cameras, not just LATISS here) know how long its been since the last image readout? It should, right?
          • And given that, should this not be the definition of what we're writing into DARKTIME (modulo the usual "do we include the exposure time in DARKTIME" discussions)? I wonder if, with really good darks, and the correct values being written for DARKTIME, whether these would subtract nicely?
          • How feasible would it be to add a full frame clear before any images where the DARKTIME is above a few mins?
          mfisherlevine Merlin Fisher-Levine added a comment - A few points/questions: Do the cameras (and I mean all our cameras, not just LATISS here) know how long its been since the last image readout? It should, right? And given that, should this not be the definition of what we're writing into DARKTIME (modulo the usual "do we include the exposure time in DARKTIME" discussions)? I wonder if, with really good darks, and the correct values being written for DARKTIME, whether these would subtract nicely? How feasible would it be to add a full frame clear before any images where the DARKTIME is above a few mins?

          This is entirely a scripting issue, and doesn't require any changes to the camera itself, correct?  I just want to check that fixing this issue doesn't require more calibration reconstruction.

          czw Christopher Waters added a comment - This is entirely a scripting issue, and doesn't require any changes to the camera itself, correct?  I just want to check that fixing this issue doesn't require more calibration reconstruction.

          To conclude if this is entirely a scripting issue, we need to test if this issue can be solved by enabling IDLE_FLUSH.

          youtsumi Yousuke Utsumi added a comment - To conclude if this is entirely a scripting issue, we need to test if this issue can be solved by enabling IDLE_FLUSH.

          People

            czw Christopher Waters
            czw Christopher Waters
            Christopher Waters, Merlin Fisher-Levine, Robert Lupton, Yousuke Utsumi
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

            Dates

              Created:
              Updated:

              Jenkins

                No builds found.