The ML50100

The FLI ML50100 is an amazing camera. With a pixel size of 6 microns and 8176 x 6132 pixels!!!

But, boy, is it tricky to calibrate the images. There are several challenges:

• Because of the large pixel size, the shutter is also very large - and hence slow. Flats need to be at least 8-10 seconds - otherwise the shutter will show up.
• Being front illuminated, the chip has RBI.
• The chip is so big, that it actually has two readouts - one for each half. And each readout has slightly different characteristics (bias, gain)
• Richard thinks that there could be bias drift depending on Ambient temperature

Here is a (stretched) 10 min dark master frame:

And a bias master frame:

The seams are clearly visible. 

First, calibrate light frames using the master dark:

Pre-dark calibration:

If you look closely, you can see the seam in the middle ...

Post-dark calibration:

Looks pretty clean. But if I now stack my (40) luminance frames (without alignment - I just want to check if the seam is gone):

The seam is clearly still there, i.e. the individual frames weren't completely cleared up. And this was even more complicated with the flat frames. Because our flat panel broke, we had to use sky flats, i.e. the exposure time is different for each. And because of the two different halves, we can't scale our darks, i.e. we have to take darks at the same duration as the flats ...

We have three theories where this could come from:

1. The RBI issue - though Richard said at 10 minute exposure and -15C we should see much if any
2. The bias is dependent on the ambient temperature
3. Our darks aren't clean (we had some darks with clear light in them - maybe they all suffer from minor light leaks ...)

I ran two experiments to check this:

1. Darks with roof closed vs. open

I took several 10 min dark frames with the roof close and the roof open (with some moonlight):

Overall		Left		Right
Roof Open	Roof Closed	Roof Open	Roof Closed	Roof Open	Roof Closed
1017	1016	1011	1010	1021	1019

The levels are just so slightly higher when the roof was open...

2. Bias at different ambient temperatures

I took a stack of 20 bias frames at various ambient temperatures:

Ambient Temp [C]	Level - Overall	Level - Left	Level Right
20.9	1,012	1,007	1,016
23.5	1,012	1,007	1,016
26.3	1,012	1,007	1,015
28	1,010	1,005	1,013

Not much difference - except the first step from 28 to 26.3 degrees. The difference here was that the 28 degree bias were taken with the roof close, the other when it was open...

This seems to indicate that we have some light leaks, but not so much that ambient temperature plays a role.

If this is indeed the case, then there isn't much that we can do about it now... But in a month, I'll be at the site again (for some leftover work). At that time I could take better darks by either covering the scope (in particular the front) or taking the camera off completely and take darks with the front fully covered.