How to get raw data for non-linearity determination if the flat light source is not stable over time?
In https://jira.eso.org/browse/MIC-652 Udo Beckmann argues that
At J and H band wavelengths the temporal flux stability was good, as your results confirm, however things were different in K band, especially at wavelengths > 2200 nm. I observed a slowly increasing flux over time, starting after a few minutes. It turned out that the housing of the flatfield lamp slowly heated up, adding additional flux especially at long K band wavelengths. I could avoid this effect by actively heating the flatfield housing and regulating the temperature at a few degrees above ambient.
To which Robert Harris replies
We currently have no requirement on the stability of our flatfield light sources, just that they are able to produce a uniform illumination and that the flux is variable. This makes sense to me, as normally a flatfield is just checking the pixel to pixel response (if we were trying to check the linearity of the detector this would be different). By varying the flux from our bulbs we can then compensate for variations in flux without changing the MICADO detector settings.
We currently plan to use flats for non-linearity determination. Should we do anything special for this or can we assume that "varying the flux from our bulbs [can] compensate for variations in flux without changing the MICADO detector settings"? How do we know how to vary the flux?
See also Slack discussion: https://juheart.slack.com/archives/C01Q3QXDA92/p1622204445019700
For now I propose that we ask the detector team to measure the non-linearity with a stable source in the lab. We can then compare this with the non-linearity measurements in commissioning and determine how to proceed based on that. And that for now we will assume that this extra flux in K is either not an issue, or can be corrected / accounted for based on the tests during commissioning.
@verdoes what are your thoughts on this?