Flat Fields and Snap Caps

Tags: astronomy, ccd, imaging, tao, observatory, monkton-stargazers, software-engineering, developers

I've been busy today doing a ‘5 minute job’ that took 9 hours =8-o

I’ve reconfigured TAO - our automatic robotic observatory - to use the SnapCap for doing electroluminescent panel flats instead of using Sky Flats. Sky Flats are nice but our tilt-up limit is rather low and doesn’t allow us to reach the right point in the sky, so ACP just crashes out of the flat plan. Lucky we have the SnapCap, from Gemini Telescope Design.

Ought to be simple, you might think. Nope! SnapCap uses an ASCOM driver, which we wrote for Gemini Telescope Design. It's modelled as an ASCOm Switch device, where switch 0 opens and closes the cover and Switch 1 (an analogue switch) controls the light on.off/brightness in 255 steps. ACP requires something that can be run from the command line, so my first stab was to write a PowerShell script. I could not get ACP to use it! I had it all working in a command prompt, but ACP would not budge for reasons I could not figure out. A permissions problem? Who knows! So I wasted about 2 hours trying to get that to work. In the end I had to write a little command line application in C# that ACP is happy to work with, and can control the SnapCap by proxy. It works like this:

C:> SnapCap.exe –cover close –light on –brightness 255

(does what it says on the tin).

For a flat field, the goal is to fill the pixels roughly half way, on average. Pixel values in the final images are measured in ADU – Analogue to Digital Converter Units but Full Well Depth is given in electron charge (e is the unit of elementary charge, the charge of one electron). For the KAF-6303E chip, full well capacity is 100,000 e but we need that in Analogue-to-Digital Converter units (ADU) which gives us a direct pixel value. To get that we divide by the camera’s “gain”, which for the STXL-6303E appears to be 1.65 e/ADU (it’s not in the specs!). This gives 60,606 ADU full well capacity (the old ST-8E was about 45,000). There’s a bit of an offset (the “pedestal”) so the real value is a bit less and 60K ADU would be a good working value. We want half of that for our flat fields, so about 30K ADU mean pixel value.

By experimentation, I came up with these values. They are all a bit low because I couldn’t find a value for the gain and I used the ST8 value, which is about 2.3 instead of 1.65, giving only 43K ADU full well depth instead of 60K ADU. Luckily, ACP computes the optimum exposures automatically.

Filter      SnapCap         Exposure        Max. Pixel
Name        Brightness      (seconds)       Value (ADU)

Luminance    30%             0.50            23370  
Red         100%             0.95            23363  
Green       100%             2.00            23540  
Blue        100%             2.10            23683  
Clear        30%             0.50            23424
Hα          100%             15.0            24094  
OIII        100%             11.5            24052  
SII         100%             20.0            23780  

Those figures are for full resolution (unbinned) flats. So in ACP, that translates to per-filter SnapCap brightnesses (in the range 0 to 255) for each binning level, of:

; [TPL]                  L    R    G    B    C    Ha   O3   S2  
PerFilterBrightness      96, 255, 255, 255,  96, 255, 255, 255
PerFilterBrightnessBin2  24,  64,  64,  64,  24, 128, 128, 128
PerFilterBrightnessBin3   12, 32,  32,  32,  12,  64,  64,  64
PerFilterBrightnessBin4   6,  16,  16,  16,   6,  32,  32,  32

…and ACP works out the exposure. These are a first pass estimate and we can refine based on the actual images we get.

And now you know why it took me 9 hours! All being well, we should see “dawn flats” appearing in the morning.

No Comments

Add a Comment

Let Us Help You

Find us on Facebook